FROM SPEECH TO SONG: An of interdisciplinary investigation rhythm in English and Spanish Rosalia Doctor of Philosophy The Rodriguez-Vazquez in Linguistics and English Language University of Edinburgh. 2007 Table of Contents Declaration I Abstract II Dedication Ill Acknowledgements IV Introduction 1. Rhythm as 1 the connecting element between language and music 1.1. What is 1.2. rhythm? 5 Rhythmic hierarchies 1.3. The 1.4. 5 8 study of rhythm: prosody 11 Linguistic stress and its relation to rhythm 12 1.4.1. Introduction: stress and accent 1.4.2. 13 Types of languages in relation to accent 13 1.4.3. Stress 15 1.4.3.1. Definitions 16 1.4.3.2. Phonetic realisation of stress 17 1.4.3.3. Stress 1.4.3.4. The 1.5. 2. placement rules: a preliminary account relationship between stress, weight, length and tenseness Summary 27 typologies: the dichotomy between stress-timed and syllable-timed languages. English and Spanish 2.2. The as typological models prosody of Spanish 2.2.1. From Latin to Spanish quantitative system 35 43 assignment in Present-day Spanish 46 2.2.2.1. Harris's analysis amended by Roca 2.2.2.2. Navarro Tomas's 2.3. The 35 and syllable structure 2.2.1.2. Vowel systems 2.2.2. Stress 27 35 2.2.1.1. Stress and the evolution of the 2.2.3. 20 25 Speech prosody 2.1. Prosodic 18 analysis amended by D'lntrono et al. 46 50 Syllable-compression phenomena in Spanish: synaeresis and synalepha55 prosody of English 66 English to Present-day English 66 assignment in Present-day English 73 2.3.1. From Old 2.3.2. Stress S3 2.3.3. Stress and related features 2.3.3.1. Vowel length and reduction 83 2.3.3.2. Vowel quality and tenseness 89 2.3.4. 2.4. Syllable structure and syllable weight 91 Summary 92 prosody 98 3. Verse 3.1. Introduction: poetic rhythm and metre: their relation to speech prosody. Analysis of verse prosody in English and Spanish 3.2. Verse Ill prosody in Spanish 3.2.1. Introduction: from Latin 3.2.2. The 3.2.3. 3.3. Verse principle of isosyllabism in Spanish syllabic 119 126 131 English alliterative verse to Present-day English accentual131 verse syllables in contemporary English Synalepha in English: a 139 verse comparison with Spanish 141 149 Summary relationship between speech, 4.1. The interaction between 4.2. The 4.2.1. verse verse and music 153 prosody and musical rhythm 154 text-setting problem in the existing literature Janda and Morgan (1988) "El acento dislocado castellano" and Morgan 157 - pues cantado - and Janda (1989) "Musically conditioned stress shift in 160 Spanish revisited" 4.2.2. 112 115 verse prosody in English 3.3.2. Stresses and 4. The verse... displacement of stress in Spanish poetry 3.3.1. From Old 3.4. quantitative verse to Spanish syllabic Syllable counting and its related devices 3.2.4. The 3.3.3. 98 Hayes and Kaun (1996) "The role of phonological phrasing in sung and chanted verse" 162 4.2.3. Hayes and MacEachern (1996) "Are there lines in folk poetry?" 4.2.4. Hayes and MacEachern (1998) "Quatrain form in English folk verse". 178 4.2.5. Hayes (in press) "Text-setting as ii constraint conflict" 172 186 4.2.6. Dell and Halle (in press) "Comparing musical text-setting in French and 189 English songs" (forthcoming) "Text, tune and metrical form" 197 4.2.8. Kiparsky (2006) "A modular metrics for folk verse" 202 4.2.9. Summary 207 4.2.7. Halle 4.3. Object of study: Folk 209 song .....210 4.3.1. Definition and characteristics of folk song 4.3.2. 4.3.3. The importance of lyrics in the study of folk 4.3.4. Text-tune relationships in folk 4.3.5. The choice of a specific folk 4.3.5.2. song corpus Burgos Kennedy's Folk songs of Britain and Ireland Analysis of the folk song corpus: 220 220 221 overview and general observations.. 221 226 4.3.7.1. Case study I: "The farmer's boy" 226 4.3.7.2. Case study II: "Alegria, caballeros" 235 4.3.7.3. Conclusions Stretching out the theory: Art 243 244 song 4.4.1. Definition and characteristics of art song 244 4.4.2. Art song in Britain 245 4.4.3. Art song in Spain 246 4.4.4. Case studies 247 4.4.4.1. Case study III: "Loveliest of trees" 248 4.4.4.2. Case study IV: "Olas gigantes" 259 4.4.4.3. Conclusions 5. 217 218 4.3.7. Case studies 4.4. song song 4.3.5.1. Manzano Alonso's Cancionero de 4.3.6. 215 Transcription and classification criteria 269 Summary, conclusions and further implications 271 References 279 Appendices 292 Chapters 1-4: 292 Appendix I: Glossary of terms and symbols 292 a) Terms: 292 iii b) Symbols: 297 Appendix II: Instances of synalepha and synaeresis in folk songs a corpus of 444 Spanish (Manzano Alonso 2003) 298. Chapter 4: The relationship between speech, verse and music 299 Appendix III: Index of OT text-setting terms and constraints with examples. 299 Appendix IV: Mismatches between stresses and beats in Manzano Alonso (2003) 305 Appendix V: Mismatches between stresses and beats in Kennedy (1984) 359 iv I declare that this thesis is my own work, and other degree or professional qualification. I that it has not been submitted for any Abstract general theoretical frame of this dissertation has to do with the study, from an interdisciplinary and interlinguistic point of view, of the typological dichotomy between stress-timed and syllable-timed languages, inasmuch as this distinction is valid at all. As a preliminary step, I carry out a comparative examination of the basic prosodic characteristics of English and Spanish, in order to then analyse the standard versification systems of these two languages. In the central part of my dissertation, I explore the most important text-setting Optimality Theory constraints as applied to a corpus of English and Spanish folk and art songs. My main objective in carrying out these three-level analyses is to check whether the actual setting of verse to music responds to some kind of underlying rhythmic constraints common to language prosody, verse prosody and music, and whether those constraints are ranked differently from language to language. The conclusions have to do with a correspondence between the timing typologies of language and the rhythmic typologies of music. I find clear inconsistencies or mismatches between speech prosody, on the one hand, and verse and music rhythm, on the other. These inconsistencies work differently in a syllabletimed language like Spanish than in a stress-timed language like English. While in the first type of languages I find a natural counterpoint or dialogue between speech prosody and musical rhythm, in the second type this counterpoint tends to be considered arhythmic. In other words, I establish a difference in kind in relation to the dialogue between prosody and music for each of the two types of languages. In English, the level of agreement between the two stress-patterns is really high, while in Spanish the counterpoint between the two patterns is used as an expressive device. The II Os meus pais, Filomena O meu e Virxilio. irman, Virxilio. Sempre. Acknowledgements foremost, this dissertation is indebted to the work, support and encouragement of my two supervisors, Professor Heinz Giegerich and Professor Raymond Monelle, whose contribution to the final result goes far beyond the First and They believed in this project from the very beginning, asserting the value interdisciplinary enterprise. They made enormous efforts to always facilitate my work by offering me both human and material resources whenever I needed them. They devoted long hours to the thinking and discussion of each and every page contained in the final manuscript. They came up with ideas and thoughts which enriched the writing-up process substantially. Last, but not least, they supported me in every possible way. academic. of such an grateful to the numerous people who have contributed to this work by generously sharing their invaluable knowledge, experience and ideas with me: Dr. Javier Perez-Guerra, Dr. Francis Dell, Dr. Carlos Piera Gil, Prof. Nigel Fabb, Dr. Ricardo Bermudez-Otero, Prof. Donna-Jo Napoli, Dr. Patrick Honeybone, Dr. Katie Overy, Prof. April McMahon, Dr. Chris McCully, Dr. John Halle, Dr. Marta Falces Sierra, Dr. Jean-Louis Aroui, Prof. Jose Dominguez Caparros, Prof. Peter Nelson, I am very Rafa Martin Castilla. accomplishment of this dissertation would not have been possible without the of the Fundacion Caixa Galicia, the Arts and Humanities Research Council, the Xunta de Galicia and the School of Philosophy, Psychology and Language Sciences (University of Edinburgh). The financial support Special thanks to all the staff and fellow postgraduate students at the department of Linguistics and English Language and the faculty of Music (University of Edinburgh), and the department of English, French and German (University of Vigo, Galicia, Spain). Also, to my English Language students at both institutions, who turned my working weeks into truly enjoyable learning adventures. big thanks to the Edinburgh University Music Society, the Edinburgh University Chamber Orchestra, the Noteworthy Trio, the Rosalba Duo, the Orquesta Clasica de Vigo and my group, Nao d'lre, for giving me the chance to cultivate one of my biggest passions. A very friends Marta Perez Quiroga and Helena Miguelez Carballeira, with whom I long hours of fruitful conversation, lots of laughs and a few tears for some years now. jE que sexan moitos mais! To Nerina Sunara, for always saying the right words at the right time. To my have shared Colin, in my life has shown me that beauty exists even beyond good poetry and symphonic music. I will never find enough words to thank you for this happiness. It's been a long time waiting... your presence IV Rodriguez Vazquez, my best life-long companion. Thanks dialectic capacity, optimism, generosity, eagerness to leam, sensitivity, captivating joy and, most of all, your unstoppable idealism. To my brother, Virxilio for your true friendship, Last and forever first, to my parents, Filomena Vazquez Paz and Virxilio Rodriguez Silva, for giving me the language of my dreams, for your always on-time words, your welcoming smiles, your openness, your silent and constant longing to have us close, your support, your immense love, your inspiration, your sense of freedom. Thanks for this boundless love for Vos sodes a Sodes todo Querovos. a small comer on the Iberian Finisterre. mina aldea, a mina identidade, o meu tesouro. o que me da a alegria. "Agora teno saudade do fiituro que hei de andar lonxe da mina cidade. Cupulas brancas E sinto na o ar: etemidade o corazon a sonar" (A. Aviles de Taramancos) V Introduction is, What art know you And if that's true, I soul, but can as well modestly and simply give more you rhythm, in paints, building blocks [...], sounds in music, or I do: it is nothing than rhythm. don't have to burden myself with imitation or with whatsoever: bus tickets, oil poetry, as or you just name any material or words in it (Schwitters 1993[ 1926] :229). In 1926, the avant-garde artist Kurt Schwitters used these words to refer to the essentially rhythmic nature of such. Strange as it might seem, artistic work that aimed at being described any as Schwitters's categorical assertion about the ontology object could be transferred to the linguistic field and thus become of the artistic a partial definition for language. In this as piece of work, I that of artistic both more that the status of natural language is the While painting a or a sculpture can analysed thoroughly, music, like speech, some on kind of multiple impose produced and perceived in real can only be experienced once over a the listener the necessity to devise, consciously - perhaps the most obvious regular structure on any and then period of speaker.2 This fact imposes both or one - on the unconsciously, systematic organisation of the sound stream. Rhythm is means a are be visually apprehended at time, the length of which is chosen by the speaker and same specifically, music, in two respects, namely that essentially semiotic systems and both are time.1 language and, assume one of the to which people resort in order to flow of sounds occurring in time. As Liberman (1979:313) observed, "[a]ll temporally ordered behaviour is metrically organised", an assertion which points to a deep relation between what Schwitters referred to as 'art', that is, poetry and music, on the one hand, and everyday speech, on the other. Such a relation would be mediated The fact that speech, metrical structure verse, by rhythm: and music all have hierarchically organised implies [...] a common ' cognitive origin. Not only are I am using the term 'language' in a broad sense, referring to any human manifestation that articulates thoughts by means of an arbitrary group of signs, whose main goal is to serve as a vehicle for communication (see Halliday 1973). For further discussion about the temporal nature of music, see Cureton's (1993:68-72) classification of arts into 'spatial arts' and 'temporal arts'. 2 For clarity's sake, I use 'speaker' to refer not only to the role of the individual that utters words or sentences, but also to the role of the performer of a musical composition. 1 the principles of organisation surprisingly similar for all three faculties, but they also allow for the same play-off between abstract construct or underlying structure and actual realisation (Couper-Kuhlen 1993:112). Several branches of linguistic theory have dealt with the role of rhythm in speech and language, highlighting the similarities between music and speech (Liberman 1979). In fact, Metrical Phonology draws rhythm in music. From a more on formal tools developed for literary perspective, there have been studies done poetic rhythm (Attridge 1982) and on on the technical analysis of text-tune relations (Cureton 1992). The initial theoretical when talk about we known assumption for the rhythm, three levels prosody, which is the formal system used verse tool for poetic scansion and (iii) as a rhythm and metre. The main focus of this work is the relation between these three levels. The core research question that this work addresses has to do with the rhythmic typologies of language and how they music. What timed into play: (i) speech rhythm, also come 'speech prosody', which refers to the rhythm of ordinary speech, (ii) as musical of this dissertation is that purposes are as Spanish, and English? This question entails English? What are the status of the adapted to a verse an an verse and vocal allegedly syllable- allegedly stress-timed language, such as series of others, which need to be answered before final conclusion: What a manifested in the differences in the nature of stress between language, such reaching are are the prosodic characteristics of Spanish and the rules of stress assignment in each of these languages? What is syllable in each of them? How are those prosodic characteristics in Spanish and English? What role do stress and the syllable play in the versification systems of the two languages? How is Spanish and English? What are verse set into music in the generalisations to be made about the matching of speech rhythm (i) with poetic metre, and (ii) with musical rhythm/structure? In order to find divided my appropriate on for all the above questions, I have dissertation into four main parts. In chapter review of the current theories focus answers the main stress, and I on one, I carry out a preliminary language rhythmic typologies. In chapter two, I rhythmic element in the so-called 'stress languages', namely analyse the evolution and present status of stress assignment rules in Spanish and English. In the third chapter, I explore 2 some of the methods developed for the scanning of verse in stress-timed and syllable-timed languages, and explore a theory that relates linguistic prosody and poetic prosody in Spanish and English. In the fourth chapter, I analyse the process of text-tune alignment in English and Spanish in order to establish the differences in the musical setting of stress-timed verse and syllable-timed setting stems from not verse. The methodology applied to the analysis of text- series of articles published in recent a years, mainly - although exclusively - within the Optimality Theory (henceforth OT) framework. I review each of those papers and Kaun - Janda and Morgan (1988), Morgan and Janda (1989), Hayes (1996), Hayes and MacEachern (1996), Hayes and MacEachern (1998), Kiparsky (2006), Hayes (in press), Dell and Halle (in press), Halle (forthcoming) order to be able to establish an - in analytical apparatus which extracts the most relevant elements from them. In this respect, I must point out that the method applied to my analysis does not strictly belong in the OT school. I carry out a series of analyses of specific case studies, which have been carefully chosen a vast among corpus of songs. This subgroups, corresponding to the subgenres of folk and art show a number of introducing the case specific characteristics which on the one of English songs reasons why this song. These subgenres carefully explained before studies. The first subgroup is constituted by 444 from Manzano Alonso's Burgos, are comprises two corpus (2003) compilation of Castilian folk hand, and 239 songs extracted Cancionero de extracted from Kennedy's (1984) collection in Folksongs of Britain and Ireland, corpus songs songs on the other.3 There are several of songs constitutes the analytical focus of this dissertation. First, folk songs have been passed from generation to generation by people often non-musicians in this respect, kind of 'native exceptions are - who inherited them as - most popular 'composers' had created them; the matching of lyrics and music must necessarily respond to some language of text-tune alignment', where both the rules and the clearly part of the 'song grammar' of that specific tradition. The collections with which I collectors themselves, am dealing have been only minimally edited by the something which ensures 3 the 'native' quality of the songs. On Chapter 4 contains an in-depth analysis of two folk songs (4.3.7) and two art songs (4.4.4.). I have analysed the whole corpus of 444 Spanish folk songs and 239 English folk songs in terms of syllable/stress/beat mismatches (see 4.3.6.). Details of such an extensive analysis are shown in appendices IV and V. also 3 the other sung, hand, the genre to which they belong which means that any instrumental parts - song are is obviously intended to be - composed bearing in mind vocal Spanish texts parts, and vice versa. We can therefore presume that the prosody of the rhythm of the music according to traditions of Spanish word-setting, will relate to the and that the corpus same will happen with the English texts. The second subgroup within the is constituted by a sample of two art namely Butterworth's (1911) Six songs extracted from two collections, from A Shropshire lad', and Mompou's (1971) Becquerianas. The analysis of art songs is intended rhythmic constraints applied to folk whether the 'native' which have been written device considered songs as a test to songs also work for by trained composers, who intentionally step musically 'incorrect' or find out away songs from any aesthetically 'unacceptable'. My main objective in carrying out these three-level analyses is to check whether the musical score - responds to language prosody, ranked setting of verse some verse - either written before or after the instrumental kind of underlying rhythmic constraints common to prosody and music, and whether those constraints differently from language to language. My conclusions show a correspondence between the timing typologies of language and the rhythmic typologies of music. The analysis of the folk and art brings to light clear inconsistencies hand, and one differently in like are a verse or mismatches between speech prosody, and music rhythm, on songs on the the other. These mismatches work syllable-timed language like Spanish than in English. While in the first type of languages I find a a stress-timed language natural counterpoint or dialogue between speech prosody and musical rhythm, in the second type this counterpoint will tend to be considered arhythmic. In other words, I establish difference in kind in relation to the the two types a dialogue between prosody and music for each of of languages. In English, the level of agreement between the two stress patterns is really high, while in Spanish the counterpoint between the two patterns is used as an in what expressive device. This confirms the hypothesis that there is we could call difference 'stressability' of syllables in English and Spanish. The originality of this dissertation is that it shows that such a difference is manifested not only at the speech level, but at all three domains - speech, this a study, music. 4 verse and, as discovered in Rhythm as the connecting element between language and 1. music language and music Given that both natural constituted by a small, are sound systems - that is, they are productive number of sound units arranged according to very certain theoretical parameters -, it seems plausible to assert that there are certain aspects of music theory which bear a strong resemblance to some of the work done in phonology. As Jackendoff (1989:15) points out, "musical analysis might be brought to bear current on some there is no doubt that problems in phonological theory". At some of the a more practical level, methodological tools used in music analysis parallel certain aspects of phonological structure. Actually, scholars such as Halliday (1967, 1994) and Liberman (1979) introduced their theories by presenting rhythm as underlying building block in phonology and invoked chants and poetry as the evidence for the of music alignment of rhythmic beats with speech. Moreover, Liberman's use clearly emerged from an notation represent metrical to interfertilisation of work on music and on structure language. Suprasegmental phonology, that is, the branch of phonology that deals with intonation and rhythm, is the field where the most fruitful parallelisms arise. The next sections concentrate on 1.1. I review some an of the works that deal with capacity, in order to establish expand explanation of the rhythmic phenomenon. In section a rhythm as a general cognitive comprehensive definition for it. In section 1.2. I important characteristic of rhythm, namely its hierarchical nature. In on an section 1.3.1 concentrate on the discipline that deals with rhythm in language, that is, prosody. Section 1.4. is devoted to the analysis of stress and its role in the rhythm of the so-called 'stress 1.1. What is of the chapter. beginning of this chapter, rhythmic perception and production is of our basic mental certainly preceded is a summary rhythm? As mentioned at the one languages'. Section 1.5. presents our capacities and in the evolution of the human species it almost ability to use language (see Blacking 1973; Downing 1995). It widely assumed that the comprehensive and stabilising factor of our temporal rhythm. Despite this acknowledged fact, there is still much consciousness is confusion with regard to the definition of rhythm itself. 5 One of the first contemporary of a sequence treatises on rhythm defined it of events in time which produces as "that property the mind of the observer the on impression of proportion between the duration of the several events or groups of events of which the sequence is composed" (Sonnenschein 1925:16, Underlying this definition, there is with the nature of a rhythm. First of all, rhythm is are they are subject to time and Time and can space are terminology (see Kant 1884) forms' in Kantian a mental construction.4 The only be perceived through the the objects of the observer's experience and, space. emphasis). series of principles commonly associated 'sequence of events' to which Sonnenschein refers observer's senses; my - as previous intuitions such, they - 'a priori in relation to perception; in other words, they are the conditions that make experience possible. In this sense, time is transcendental than space even more of outer say because, while the latter is the organising form experience, the former organises both inner and outer experience. We could that perception does not constitute experience unless it is real for one's consciousness. An event in the own physical realm must be ordered with respect to the 'here and there' and also to the 'before and after'. Secondly, given that rhythm is dependent on inner time, it is a serial phenomenon. The events that constitute the basis for rhythm to be constructed take place in sequence, one after another. Thirdly, the conception of rhythm is based either conscious sense or on quantitative measurements, unconscious, delimited by the action of stress, which produces proportion between the duration of different of segments.5 a As Barry (1990:262) observes, The organisation of time is always the search for points of reference and articulation, and durational spans are created by information shaped between those points. It is through points of articulation that time be structured into durational spans, design can can and through defined characteristics be understood and remembered. 4 Although rhythm can effectively be conceived as a mental construction (see Pike 1945; Abercrombie 1967), it is important to note that it has physiological correlates (see Luce 1971; Glass and Mackey 1988). 5 Stress is the basis of rhythm: "Quantity alone cannot have been decisive for rhythm. Therefore, with it we have not only in a regulating but in a shaping capacity - stress" (Weil 1960:675, in Allen 1973:100). - 6 Cooper and Meyer (1963:16) define rhythm unaccented beats more this, an event beats is grouped in relation to are a group composed by at least one as an "the way accented in which one or one".6 According main beat, and one or more to weak arranged around it. Malof (1970:2) complements these definitions by pointing to the fact that the observer's also with other sense of rhythm has to do not only with the beat, but linguistic features such as pitch, loudness and duration, which are markers of prominence. Rhythm is based periodicity. These two determine whether iteration, which is, in turn, manifested in repetition and on are the most important perceptual hints used by listeners to sound stream is a rhythmical or not. In Attridge's (1982:77) words, perceive To a a regular rhythm is to comprehend a sequence of events as pattern in time, with two mutually reinforcing features, repetition and periodicity. That is to stimulus say, a series of stimuli is understood occurring again and again, and these repetitions occurring at equal, or at are same felt to be least equivalent, temporal intervals. It is not objective, measurable phenomenon, but repetition is A certain amount of the as an a perceptual an one. essential condition for rhythm to be such by the listener. In music theory, it is said that repetition, both perceived as small- and large-scale, can be considered the prime form-building agent and the most fundamental construction principle of musical form (Barry 1990:71). Similarly, in literary theory, readers reminded that "in all levels of language, the are essence of the poetic artifice has to do with periodic repetitions" (Jakobson 1966:23). Repetition makes it possible for listeners to order objects into create a recurrences coherent a specific pattern, design. This has processing so consequences: that their it reduces complex information to intelligible shape and coherent structuring; the coherent the structure of the perceived object, the tend to be learned and remembered. repetitions 6 are perceived to occur at more more easily the information will Periodicity has to do with the fact that equal intervals in time. It contributes to the of accents, namely 'metrical accent' - analogous to beat in a bar, and 'phrasal accent' - analogous to sentence stress -, which refers to the most prominent beat of a musical phrase. These two concepts are directly related to the notions of 'metrical hierarchy' and 'grouping hierarchy' that I develop in the In music, as lexical stress in speech, there are two kinds which refers to the strongest next section. 7 (see Dowling establishment of the so-called 'invariants' inside the listener's mind and Harwood 1986): structural constants have been established, once change in local pattern features would be processed as such, never as a any surface change in the underlying structure. Periodicity complements repetition and reinforces the rhythmic impression. Once repetition and periodicity have been fixed in the mind of the listener, rhythm satisfies 1.2. If can be projected into the future, expectation at the an same time as that each event continually so that event itself generates a new one. Rhythmic hierarchies we accept the statement that all human actions, artistic or non-artistic, have rhythm, we are measured in implicitly asserting that the time in which one way or Conversational metre. another, that is to say, it has a any action takes place is perceptually regular pulse or speech is clearly less regular than poetry and than most forms; nevertheless, experiments have shown that, in conversation, musical "speakers even attune to each other's rhythm, speech [...] of both" (van Leeuwen Liberman's so that the same pulse underlies the 1999:43).7 This is the basic idea underlying original thesis work, based on the so-called 'Metrical Organisation Hypothesis'. This hypothesis, mentioned in the introduction, states that "all temporally ordered behaviour is metrically organised" (Liberman 1979:313). According to this broad conception of metre, of two separate relative strength of beats, and the grouping hierarchy, which has to do with the phrases.8 In order to decode regular pulse a or a metrical sequence, beat pattern, that is, salience structure on a listeners try to find beat rate in listener, who terms 'strong' those beats which are not salient. Studies in This conception is not a comfortable a more or range.9 less Metre beats, and that structure is instinctively inferred by the 7 that rhythm consists hierarchies, namely the metre hierarchy, which has to do with the organisation of beats into imposes we can say are salient and 'weak' those which rhythm perception point to the fact that, in order to be Saint Augustine had already observed that "the role that metre plays in readily in speech" (Epistola CI, in Dogil 1979:45). 8 One of the metatheoretical problems of dealing with prosody is that the technical terminology is used in many different ways. 'Metre' seems to be one of the most controversial terms. I am using it here in a broad sense, referring to any kind of regular pattern, and not just to the study of poetic patterns. 9 At the metrical level, the strong impulses in the rhythmic sequence are called 'beats', and the weak ones are called 'offbeats' (Attridge 1982:77). all movements is new. seen more 8 perceived as such, beat intervals cannot be much longer than 1.5 seconds much slower than 40 beats per beats Another occur. or tempos minute, where 'tempo' refers to the rate at which perceptual restriction is that beat intervals are always divided, preferably into patterns of equal subdivisions in 2:1 ratios. More complex subdivisions The are difficult to encode by listeners (Dowling and Harwood 1986:196). conception of a periodic alternation between strong and weak beats, that is, the metrical hierarchy, can be represented graphically. Traditionally, it has been represented either in the so-called 'prosodic notation' or in 'metrical notation' (Liberman 1979; Selkirk 1984). The latter derives from Liberman's hypotheses, which any suggested the existence of two abstract structures, form of complex behaviour and the other structures were called 'metrical relation between the two was one of them imposed one imposed on on the time line. Those pattern' and 'metrical grid', respectively, and the defined as their 'alignment'. (1) gold The ring is just a symbol (metrical grid) / x In / (prosodic notation) x/x/x prosodic notation, there is no hierarchy among syllables. In the grid notation, the hierarchy is clear: the beat. At each hierarchical three times that of the the different accented more dots, the stronger the level, the distance between beats must be either two immediately lower level. Given that the grid can or actually represent a hierarchy of prominences, both between strong and weak beats, and also among strong more and weak beats themselves, this latter type of metric representation is transparent than the former. As we can see, Metrical Phonology Liberman and Prince (1977) and Liberman (1979) - 9 initiated by is the most relevant model for rhythm and intonation in language, and the fact that it employed in music puts these two in connection. - uses some of the tools music, "the metrical organisation of In a composition is designated in the signature and measured by the bar lines" (Cooper and Meyer 1963:4-5). The time is, grouping. concept of 'bar-line' connects metre with the second hierarchy, that Departing from the comprehension of metre, listeners organise sounds into groups. The sounds of natural and speech are divided into syllables, words, phonological phrases are organised into feet, phonological words, so on. In music, these groups are called bars, (Fabb 1997:25). Poetic sounds so on phonological phrases, utterances and motives, themes, phrases, periods, theme-groups, sections and the piece (Lerdahl and Jackendoff The 1983:12). hierarchy of from the listener's rhythmic groups emerges points of structural culmination within delimited structural contains a grouping peak around which the other elements delimitation of groups similarity. In musical terms, the listener tends to terms of register, also symmetrical pauses; subject to the listener's the will" most organised. The are as that distance and are coherent in groups are preferred to non-symmetrical Apart from all the perceptual variables that determine grouping, the latter is is not dictated to group texture, dynamics, timbre and tempo; also, events that are delimited by clearly-defined ones. group events each spans; has to do with Gestalt principles such response to own imagination and creativity: "the grouping of sounds by their real relations, but is completed by us, in an act that is subject (Scruton 1997:29). This is related to the fact that grouping is important variables which the performer can one of the manipulate in projecting a particular conception of a piece (Lerdahl and Jackendoff 1983:63). Summarising, rhythmic organisation consists of two hierarchies, namely metre and grouping, which interact so that the listener perceives any sound stream as being structured. Metrical structure consists of beats arranged hierarchically while grouping consists of more or less extended units arranged hierarchically. Once metre has been established and groups roughly delimited, listeners expect the given parameters to remain constant until the end of the poem or composition. The kind of rhythmic response to anticipated points of structural resolution gives rise to hierarchy of prolongational regions defined by those points of arrival, points of departure (Cureton 1992:124). 10 as well as a by study of rhythm: prosody 1.3. The The term 'prosody' comes The Latin word accentus is from Greek prosoidia, which a use of prosoidia appears Well at any rate you can agree three elements, words, mode, and - - - pros=ad, in Plato's (1998:99) discussion of the social and emotional correlates of the different a musical modes and the different kinds of speech - 'parallel to song'. morph-to-morph translation of the former and oide=cantus. The earliest recorded Republic, in means they imitate: easily enough that song consists of rhythm. Yes, I agree to that. As for apply as as the words those we are concerned, then, the principles will same have just laid down for words not set to music, both for their content and form. -True. - - And surely the mode and rhythm should suit the words. Certainly. The previous text asserts the long-acknowledged relationship between words, one hand, and melody and rhythm - on the other. In its - on the the so-called 'suprasegmental aspects of speech' origins, prosody was a part of which dealt with the grammar study of accent. Accent, in turn, referred to the phonetic properties of syllables and words in relation to the measure of rhythm, especially in verse and, more generally, those of the different metres and forms of verse. Currently, the term 'prosody' is used to refer to two different, though closely related, fields. When it refers to the study of the elements and structures related to the rhythmic and intonational dimensions of speech and language in general, it is known as 'linguistic prosody' or 'suprasegmental phonology'. Suprasegmental phonology is broadly concerned with the analysis of the phonological features that time span greater than that of a single significant sound or occur within a phoneme. As opposed to segmental features, which have to do with the sounds of individual vowels and consonants, suprasegmental may occur on prosodic features - intensity, duration and pitch the level of the syllable, the word, the phrase sentence, and may of or or even thus discriminate between different meanings for the - the complete same stretch speech. One of the most important theories within suprasegmental phonology is 11 Metrical Phonology (Liberman and Prince 1977; Liberman 1979), which deals with study of the alignment of speech sounds into the conventions that allow a a temporal grid, as well linguistic community to discriminate between with the as prose and verse. The second field to which analysis of rhythm in the verse, prosody refers is literary prosody, which deals with that is, with the laws of versification. This second meaning of prosody has always been the most commonly used among scholars, probably because it is the most directly connected with the original meaning of the word. Thus, it is not surprising that this term was originally applied by Halle and Keyser (1971) exclusively to the distribution of stress in poetry. According to them (1971:13 9-140), "an adequate theory of prosody [... ] should be expected to [... ] help us a to understand the nature of metrical verse and illuminate the relationship between speaker's everyday linguistic competence and his ability to judge or unmetrical and complex as Nevertheless, years - as or verses as metrical simple". explained in the previous section, in the last twenty five mostly since the publication of Liberman and Prince (1977), and Selkirk (1984) - linguists' understanding of speech rhythm has deepened significantly, to the point that the statement made by Metrical Phonology that the distribution of stress in the words and phrases of speech is metrically conditioned has accepted. As result, the phenomenon of patterning in normal speech has been taken to form part 1.4. As a come to of the general theory of prosody. Linguistic stress and its relation to rhythm already pointed out, all languages articulate the flow of sound in way, be widely where weaker segments languages differ from certain units over one are grouped around a salient can the kind of accent used rhythmical What makes another is the mechanism used to assign prominence to the others. Section 1.4.1. introduces the prominence devices one. a topic and argues that all be included in the category 'accent'. Section 1.4.2. analyses by different groups of languages, which gives rise to the typological distinction between tone languages, pitch-accent languages and stress languages. Section 1.4.3. concentrates English belong in this on stress group. 12 languages, since both Spanish and 1.4.1. Introduction: stress and accent The issues addressed in this section have to do with the nature of the which articulates to an otherwise unstructured flow of sound in linguistic stress. There several questions which constitute the referring core of my phonological field: What exactly is stress? Do all the languages in the world have stress? If not, which other How many types of stress are As several scholars existing treatises articulating devices there? Are there How is stress manifested in actual of the am comprehensive investigation of this often marginal subject dissertation and prompt a within the are language. I phenomenon any may languages have? rules for the placement of stress? speech? (Allen 1973:86; Hyman 1977:37) have pointed out, most on stress use this term as a synonym for 'accent', especially point out that accent is when referring to emphasis. Nevertheless, it is more general concept which refers to the mechanisms that all languages display in necessary to signal the prominence of certain syllables order to "[sjtress-accent [...] is a property found in many or a words within utterances: if not most languages of the world. it is often observed to play a key role in the phonological and/or What is more, can in turn be subdivided into three main classes, one of which is stress, the other two being tone morphological structure of languages" (Hyman 1977:37). Accent and pitch. Depending languages languages can - which of the previous accentual mechanisms they show, be classified into stress languages, tone languages and pitch-accent for a full discussion of this topic, dissertation focuses seems on on stress see Beckman (1986). Even though this languages and their characteristics, in what follows it important to explain the basic differences between stress languages, tone languages and pitch-accent languages. This I undertake in section 1.4.2. 1.4.2. As Types of languages in relation to accent already pointed out, not all the languages of the world articulate the flow of sound by means of stress. There tone and are another two devices present in pitch-accent. Languages are many languages, namely classified according to the kind of accent which they use. Most languages in Europe, like, for instance, English and Spanish, languages. This means occurring typically in that they make every use of stress - a phonetic peak word (Allen 1973:86; Hyman 1977:39); for 13 an are stress or climax overview of the current definitions of 'stress', go to section 1.4.3.1. - in order to mark syllable prominence and, therefore, highlight boundaries between words. Stress is, therefore, a string of sound which would otherwise constitute tool used to parse a phonemic sequence. only one syllable per As word culminative function show the basic principle in stress languages is that receive primary stress (Hyman 1977:38), (Allen 1973:87). This does not degrees of stress. Actually, the well-known languages such unit can - as main one; of - has one and only are of secondary stresses in one main primary stress. or occurrence The of the majority of languages belong in the category of tone languages (Laver in order to commonly known obvious that they permitted to have patterns occurrence be conceived of as 'echoes of the latter' (Allen 1973:90). can 1994:465). Tone languages have been defined movements fact which that languages cannot mean secondary stresses is in fact directly related to the the former The a English does not affect the general principle that each accentual normally the word occurrence shapeless by stating that the most general property of stress is its be summarised can a consequence, a as those languages which signal prominence. The different levels than that tones languages show a wider of pitch culminative. In stress languages, the fact that it is not are not a primary stress per word reduces the possible stress in disyllabic words to unstressed-stressed contrast, given or movements pitch 'tones'. Given that tones operate in each syllable, it is as more use can range or stressed-unstressed. By be combined freely and placed on any syllable, tone of patterns for disyllabic words, from high-high, high- low, to low-high, low-low. Despite this richness of combinations, the placement of tones shows some restrictions, which Attending to the kind of unit be classified into word-based tone word-based tone a languages variable number of can on be either phonological or grammatical. which the tones operate, tone languages can languages and syllable-based tone languages. In a contour-pattern is associated with the entire word, syllables (Laver 1994:162). Within this group, over there is a subgroup of languages, like Slovenian, the Lewis dialect of Scots Gaelic, Swedish and Norwegian, which exploit two contrasting patterns more on many words of two or syllables, of which the first syllable is stressed, while monosyllabic words 14 show no languages tone (Laver tonal contrast 1994:463).10 In Pike's (1948:3) words, syllable-based those "having significant, contrastive, but relative pitch are each syllable". This pattern of a series of such tones on identify the word. Most of these languages found in Central America, Africa and Southeast Asia - Chinese is, language. Depending based tone relative can height of pitches - serves to probably, the best-known example of this type of what feature of the tone behaviour is on languages are the syllables of the word on more salient, syllable- be subdivided into register tone languages, where the level, rise fall, or or combinations of them - used by the speaker is the relevant feature for word identification, and contour languages, where the relevant feature is the Chinese and Thai. There shape of the pitch also are curve, like in Vietnamese, Mandarin languages that show some a combination of both register and contour tone systems (Laver 1994:466). There is a third group of languages that could be placed in between stress languages and tone languages: pitch-accent languages. In pitch-accent languages, tone is accent assigned to the most prominent syllable of the word. According to this, pitchlanguages are tonal in the be combined and render sense a that only one a sense that the assigned feature is contour; on the other hand, syllable per word can they a tone are stress which languages in the receive the tonal prominence, that is, in pitch-accent languages tone is culminative. Examples of pitch-accent languages be found in Africa and Asia - Japanese is can one can of these languages. 1.4.3. Stress Since this dissertation deals with two stress languages, it seems in order to explore in depth the notion of 'stress' and its characteristics. Section 1.4.3.1. introduces of definitions for stress. Section 1.4.3.2. explores the phonetic correlates of stress. Section 1.4.3.3. constitutes rules. the 10 an - a series that is, physical introduction to stress-placement Finally, section 1.4.3.4. analyses the potential interactions between stress, one hand, and weight, length and tenseness, on - on the other. languages seems to be controversial. While Laver (1994) classifies them languages, Roca and Johnson (1999) group them, together with Lithuanian and SerboCroatian, as pitch-accent languages, while Garding (1977:7) clearly states that "the Scandinavian languages are not tone languages". as The status of Scandinavian tone 15 1.4.3.1. Definitions When dealing with stress languages, one needs to take into consideration the various conceptions to which the term 'stress' has given rise in the linguistic field. Its depending definitions vary From • "the a on the point of view from which stress is examined. physiological perspective, stress has been defined as comparative force with which the separate syllables of a sound-group are pronounced" (Sweet 1878:203). • "force of breath-impulse" (Abercrombie 1923:19). • "greater amplitude of sound-waves, [...] produced by energetic movement, such as pumping more • more breath, bringing the vocal chords [...] closer together for voicing, and using the muscles oral articulation" of means more vigorously for (Bloomfield 1933:110-111). "kinaesthetic sensations of muscle and pressure changes" (Heffner 1949:224- 225). From • "an the a psychological point of view, stress has been said to consist of impulse (primarily of place by first a psychological nature) which increase of an pressure expresses itself in in the speech-canal and approximately coincides with the point of greatest pressure" (Classe 1939:37). • "relative • "the a strength or loudness" (Trager & Smith 1951:36). degree of force with which a sound or syllable is uttered. It is essentially subjective action" (Jones 1956:243). • perceived prominence imposed within utterances (Bolinger 1958). • a mental phenomenon picked up by the hearer (Lehiste 1970; Weinreich 1954). All the above definitions involves an extra degree of highlight the fact that the realisation of stress energy, an psychologically apprehensible. The first reference to the phonetic cues duration. In fact, although stress necessary to refer to the phonetic recognise stress as impulse, both physically perceptible and group of those definitions makes constant that indicate stress, namely pitch, loudness and can cues be analysed as a phonological concept, it is which both speaker and listener such. 16 use in order to 1.4.3.2. Phonetic realisation of stress regard to phonetic realisation, there With stress, are three main strategies used to signal namely pitch change, greater duration and greater intensity (Bolinger 1958, in Hyman 1977:40). In what follows, I shall examine these three phonetic cues in an attempt to determine their real importance both in the realisation and in the perception of linguistic stress. First, 'pitch' is normally defined as "that attribute of auditory sensation in of which sounds may terms primarily sound scale is "that aspect of auditory sensation in terms of which sounds of auditory sensation in terms of which sounds 'long'" (Crystal 1969:121). In clear correlation between these three concepts, frequently characterised by intensity, in that order, as a well In as as be many stress in such a way be ordered by an on a languages there is that a a stressed syllable a greater "increased articulatory precision" (Allen of richer vowel systems in the occurrence longer duration and louder intensity, while the vowels of unstressed syllables reduced in both loudness and obvious in length (Allen 1975:80). This correspondence is not Spanish, where higher pitch and louder intensity do not render syllables; in Spanish, they are on average are on an average (1998:49-60) point out, this is syllables the concepts 'open' and 'closed' syllables - where in Spanish stressed syllables are more 1.5 times longer than unstressed 1.3 times longer. As Yavas and Goldstein even more noticeable in non-final are open explained in section 1.4.3.4. only 1.1 longer than unstressed syllables compared to 1.6 in English and 1.5 in German. French case, may English, for instance, higher pitch in the accented syllable corresponds length. Stressed syllables in English - the syllable. accented are may pitch change, by greater duration and by 1973:93) which has to do with the with upon running from 'soft' to 'loud'" (Crystal 1969:113). Last, 'duration' scale from 'short' to is scale from low to high. It depends and waveform of the stimulus" (Crystal 1969:108). Second, 'loudness' is "that aspect on a on a the frequency of the sound stimulus, but it also depends upon pressure ordered be ordered seems to be an in-between where successive syllables within a sense group are very similar in duration, pitch and loudness, with the salient exception of the last, which is, 17 on average, higher in pitch, slightly softer and much longer in duration (Delattre 1966, in Allen 1975:78). In sum, not all stress languages manifest prominence using the same devices In English, syllable is in the same extent that its constituent segments duration tract way. one more prominent than another to the display higher pitch, greater loudness, longer greater articulatory excursion from the neutral disposition of the vocal or (Laver 1994:450), something which is not the case in Spanish, where higher pitch and loudness do not necessarily entail longer duration. 1.4.3.3. Stress placement rules: a preliminary account previous section, stress languages As mentioned in the are diverse with regard to the phonetic realisation of stress. They also differ in the types of rules that assignment. While others show With main types govern stress languages do not have specific stress placement rules, some complex system of rules. a regard to the of stress: on presence or the one absence of rules of placement, there are two hand, lexical stress, which does not respond to any rules, and whose position is thus unpredictable phonologically; on the other hand, grammatical stress, whose position is morphologically conditioned (Hyman 1977:39). Grammatical stress is in turn subdivided into morphological stress and demarcative stress. In the but case of morphological stress, the position of stress is free largely determined by the morphological structure of the word. Morphological stress serves a status of a distinctive function, as its presence word - for instance, Spanish termino ('I finish') finished'), English import (noun) few not rule disambiguates the morphological vs. import (verb). and in termind ('he Generally speaking, there pairs of words distinguished exclusively by their stress occur vs. - are most of which could comparable contexts (Allen 1973:88-89). Demarcative stress is fixed by always commences or occurs on ends with the the same syllable, indicating either that syllable in question, or a ends with the next syllable word - this type of stress is found in Czech, Icelandic and Hungarian, falling on the initial syllable; in Armenian, on the final syllable; in Polish, on the penult (Allen 1973:87). Summarising, the primary difference between morphological and demarcative accentuation is that the former can, in certain morphological structure of a word. 18 cases, help to indicate the . Independently of the an the presence or idiosyncratic tendency to place stress absence of rules, each stress language shows on a specific syllable, fact that is related to a phonological and morphological characteristics of that language. The placement of stress in each language conditions the perception of native speakers. As Jakobson, Fant and Halle (1951:10-11, in Allen 1975:78) observed, Knocks produced at perceived claimed as groups by a intervals, with even of three separated by The pause hears the the word stress in the are interval after every is usually Frenchman to a knock after the languages involved: in Czech the stress is syllable, in French, When the knocks pause one are perceptions correspond exactly to the position of louder. The different first a pause. Czech to fall before the louder knock, by fall after the louder; while a Pole initial third louder, every on the final and in Polish, on on the the penult. produced with equal loudness but with a longer third, the Czech attributes greater loudness to the knock, the Pole, to the second, and the Frenchman, to the third. Despite this perceptual idiosyncrasy, it would be possible to state that there certain universals with initial stress and all regard to linguistic stress placement: adapted from Hyman (1977) in Table 1 placements the penultimate and - on Medial Finnish a) On the Total number of b) On the penult: 77 languages: 114 Spanish and Portuguese 2nd syllable: 12 Table 1: Stress There have been way the most common Final No dominant French English Total: 97 Total: 113 placement placement in different languages. hypotheses that point to a close relationship between assignment and intonation. As Hyman (1977:44) states, these hypotheses conceive stress a some - the initial syllable: Initial stress language has general languages avoid both second syllable and also final syllable stress. As shown are on no are as a grammaticalisation of intonation at the level of the word, in such that pitch, duration and intensity mind of the are speaker-hearer: 19 encoded as components of a word in the Stress comes from intonation word-stress when the as [...] Intonation becomes grammaticalised suprasegmental features of pitch, duration, and intensity that would have characterised with the In come to word in isolation are encoded function in words not in isolation. Hyman's (1977:43) words, "languages will tend to assign stress to those positions where is word, and thus a a pitch-change will be most perceptible". In clausal intonation, always seems more common be the to than a rising one. This also happens in word stress. This why penultimate stress is reason so common: it corresponds to falling pitch. It is perfect, both phonetically and conceptually - there is of unstressed-stressed-unstressed stress tends not to be realisation of an syllables. According to this theory, the a alternation reason why placed word-finally is that final stress is less natural for the falling pitch a falling pitch a - this fall has to be identified within an only syllable: "[p]lacing stress non-finally will enhance the two functions of stress: i) phonetically, the fall from H to L is enhanced and is enhanced ii) conceptually, the culmination of prominence by virtue of the fact that a syllable lacking stress follows" (Hyman 1977:46). 1.4.3.4. The There are relationship between stress, weight, length and tenseness three variables which may languages, depending variables are on condition the placement of stress in different whether they act as phonological parameters length, weight and tenseness. I shall focus on or not. Those English and Spanish in order to show how those three variables interact in order to delimit the possible placements of stress. It seems in order here these variables have "basic unit in the and a an to specify the characteristics of the domain which all effect, that is, the syllable. A syllable can be defined as the prosody of all languages [which can be] subdivided into an onset rhyme. The rhyme is further subdivided into peak/nucleus and margin/coda" (Hyman 1985:1). The peak of a syllable is usually a vocalic element. What makes syllables complex in terms of weight is the presence of one coda. on Syllables elements in their vary depending on or more consonants in the the combinations of vocalic and consonantal rhyme. 20 Table 2 shows the most frequent syllable structures in English and Spanish (adapted from Laver 1994:531 ['C' stands for consonant and 'V' for vowel]), while Table 3 indicates the percentages of open and closed syllables in both languages: Syllable type CV V (light, open) (light, open) CVC VC (heavy, closed) (heavy, closed) CVCC (heavy, closed) Table 2: Most English % Spanish % 34 58 8 6 30 22 15 6 6 - frequent syllable structures in English and Spanish. Open 44% 70% Closed 56% 30% Table 3: As evinced Open and closed syllables in English and Spanish. by Tables 2 and 3 above, English enjoys permissible syllable types. Syllables may diphthong be closed by as their nucleus and may Spanish, the syllable nucleus contains be closed by at most one consonant absence of a vowel contrast, in closed that is, a contain a single vowel short vowel, as many as or long vowel or a four consonants. In may finally and two consonants medially. In the Spanish we simply have a contrast a between open and simple CV structure, syllables clearly predominate in Spanish (Dauer 1983:55). Unlike the group a variety in diphthong, and syllables syllables. Over half of the syllables in Spanish have open a greater Spanish, English is a weight-sensitive language, that is, it belongs to of languages which display a heavy versus light syllable distinction. In a weight-sensitive system, the segments within the rhyme that bear the weight of the syllable are two morae, between called 'morae'. According to this, and a a heavy syllable is said to consist of light syllable of one (Hyman 1985:9-10). The phonetic difference heavy and light syllables is that the former have "full articulations" and the latter, "reduced timing" (Allen 1975:83). 21 general, weight-sensitive languages follow In one of the following two patterns. A first group of languages treats a syllable whose rhyme consists solely of a short vowel consonant as light, whereas or - more order for or rhymes. This is the case 1985:5). The second group vowel light and as as a a as of languages treats a American and Asian placed on and at least that as a long vowel heavy as a consonant or not, light. This is the case short a - this but, group of some South- languages (Hyman 1985:5-6). heavy syllables, that is, syllables formed by one consonant a as English does, (Roca and Johnson 1999:357). The general languages is that stress, and may therefore or by a vowel pass it are case said to be quantityin quantity-sensitive open syllable position shows reluctance to accept on to a neighbouring syllable when this is feasible short vowel in a long vowel (Dauer 1983:55). As previously mentioned, stress systems require accenting of heavy syllables, sensitive (e.g. A-me-ri-ca). The result is that syllables of the shape V be English (Hyman English and other stress languages with the dichotomy heavy-light, stress In is short vowel or syllable whose rhyme has syllable whose rhyme has a final heavy, the language must also have -V: in Latin, most dialects of Arabic, syllable with closed a heavy. Syllable weight is thus tied to the independently of whether the syllable is closed by treats long vowel and/or a vowel tenseness (Hyman 1977:6); in other words, in rhyme to be counted -VC a syllable whose rhyme has is treated - length existence of vowel a or CV, which are said to light, sometimes function differently from syllables of the shape CVC, CV:, VC, C:, which are said to be heavy (Hyman 1977:47). According to this, we could say that, in languages like English, the weight of a syllable determines whether it can carry the stress of the word or not. There is sensitive a second variable which relates to weight and stress in quantity- languages, namely length. The relationship between stress, weight and length is quite complex. The most widely accepted hypothesis is that syllable weight owes its existence directly to the nature of stress itself and the consequent lengthening of vowels: "the tendency towards lengthening is the most obvious feature observed as accordance with this (CV) syllable in a a physiological correlate to stress" (Hyman 1977:48). In hypothesis, the language with a reason why stress may shy away from a light vowel length contrast, like English, is that the 22 vowel of that syllable would tend to lengthen and therefore might corresponding long vowel. In language without a stressed short vowel would be free to owes duration on lengthen, and therefore syllable. As syllable which has articulatory very no syllable weight (Hyman 1977:48). In its existence to the assignment of stress, because this entails the stressed a consequence, perceptual or with the vowel length contrast, the a distinction between CV and CVC would be observed length merge sum, a greater stress will be rebuffed by a reasons to stay short (Allen 1964). A complex example of the relationship between syllable weight and stress placement is Latin: if a light syllable is accented, the correspondent word ends with the next syllable but one - for instance facilis ('easy'); if a heavy syllable is accented, the word ends with the next syllable if that is also heavy compose') compdne, - or with either the next or the following one - compono ('I if the next syllable is light - ('she composes', 'to compose'). componere What is clear is that not only the structure but the segmental composition of syllables reinforces the difference between stressed and unstressed syllables in English. Spanish syllables to their are much equal than English syllables with respect more length and their retaining vowel quality (Pointon 1980:300-301). In Spanish, quality tends to obscure rather than accentuate the difference between stressed and unstressed syllables. This might be the reason between stressed and unstressed vowels is why the perceived durational contrast likely to be less striking in Spanish than in English (Dauer 1983:57). Assuming that segments have minimum and maximum durations, then based on structure alone syllable length is bound to vary more in English than in Spanish. The third variable which relates to stress is tenseness. and Halle a (1964:96), vowel compared to its lax counterpart is "produced with greater deviation from the neutral position of the vocal tract [...]; consequently a tense vowel Tenseness is 11 a tense According to Jakobson displays a greater deviation from the neutral formant pattern".11 directly related to length and weight. The heightened subglottal air dichotomy tense-lax has a specific physical basis. While Bell (1867), who first drew attention distinction, ascribed the decisive role to differences in the behaviour of the pharynx, Sweet (1878) put the chief emphasis on the 'shape of the tongue.' Later investigations, however, as summed up in Heffner's General Phonetics (1949), shifted the reference "from tongue elevations and tongue muscle tensions to laryngeal positions and air pressures" (Jakobson and Halle 1964:97). The to the tense-lax 23 pressure in the production of tense vowels is indissolubly paired with longer repeatedly stated by different observers, tense vowels duration. As has been are necessarily lengthened in comparison with the corresponding lax phonemes. Tense vowels have the duration needed for the vowels and, in production of the most clear-cut optimal comparison with them, lax vowels appear as quantitatively and qualitatively reduced, obscured and deflected from their tense counterpart toward the neutral formant pattern In stress languages such long vowels and as a seems reasons to single syllable case - a that in rather than the indeed be reduced and - but this are - achieved at the expense - the of the vowels, and this does not affect subjective syllable length. are reduced, but in ordinary speech addition, these syllables often endings that they do not syllabic /r/, /l/, or /n/ as high their in function words and morphological occur seem to carry as or a very much semantic information and thus subjectively shorter than full vowels in stressed syllables (Dauer 1983:57-58). All this has important consequences investigation. Although English and Spanish former the structure of the stress to a for the rhythm of the languages under considered stress languages, in the are syllable, the word, and the utterance, is determined by much greater extent characteristics whether stressed As can Spanish greater speed and 'ease of articulation' nucleus. In called that vowel reduction means phenomenon called 'synaeresis' proportion of unstressed syllables have schwa (/a/), seem lax by phonological environment rather than by stress. Actually, English not all unstressed vowels so vs. if it does, it operates differently than in given for vowel reduction in English consonants In be the or, of adjacent vowels in Spanish reduction is conditioned it Spanish, the dichotomies between tense in unstressed position sequence pronounced as short vowels do not exist, which vs. does not operate English. A (Jakobson and Halle 1964:97). than in the latter, where those units retain their or not. That may be why languages like English are 'stress-timed', while languages like Spanish are referred to as 'syllable-timed'. already mentioned, stress-timed languages have lexical lexical free stress loudness and unstressed -, or - usually realised by a complex set of changes in length, pitch contour, quality, which clearly make stressed syllables syllables, that is, there is have been called word stress a more prominent than clearly discernible 'beat'. Languages which 'syllable-timed' either have 24 no lexical stress - the case of French - or stress, if present, is likely to be realised by variations in the pitch contour more and in the loudness scale -, as happens in Spanish. In it could be said that the sum, greater the effects of stress on the linguistic system in both its phonetic and in the number of stressed-related rules that language, the more likely 1.5. a are in necessary realisation, of the a grammar language is to be stress-timed. Summary Chapter 1 focuses on the study of rhythm, a phenomenon which is articulated at multiple levels. Three of those levels, namely speech, verse and music, share some general structural characteristics. The hypothesis argued for in this dissertation is that the prosodic characteristics of a specific language both determine and by the other two levels, namely are determined and music. The questions raised have to do verse with the kind of mechanisms that rule this three-fold interaction. Section 1.1. defines rhythm as the impression of proportion between the duration of the various segments or groups of segments in quantitative measurements involved in the process a sequence of apprehending rhythm necessarily conscious and work at the perceptual level. In the analysed in this dissertation - English and Spanish delimits those segments or groups of events. The -, case are not of the languages the articulating device that of segments is stress. The markers of stress turn, pitch, loudness and duration, three features which work at the are, in level of the phoneme and the syllable. In section 1.2., the hierarchical nature of rhythm is explored. Rhythm consists of two hierarchies. The first relative one is the metre strength of beats, and the second do with the one hierarchy, which has to do with the is the grouping hierarchy, which has to organisation of beats into phrases. In phonological terms, the metre hierarchy has to do with syllables and stress levels, while the grouping hierarchy relates to the foot, the Section 1.3. phonological word, and concentrates on so. prosody or suprasegmental phonology, the discipline that deals with rhythm in language. Suprasegmental phonology analyses the phonological features that significant sound or occur within a time phoneme. The suprasegmental which stress is manifested - span greater or prosodic features by loudness, duration and pitch 25 than that of a single - may indeed means occur at of the level of the syllable, the word, the phrase thus discriminate between different the complete sentence, and meanings for the Section 1.4. is devoted to the the so-called 'stress or even same may stretch of speech. analysis of stress and its role in the rhythm of languages'. Regarding the type of accentual device that they use be in order to articulate an classified into stress languages, tone languages, and pitch-accent languages. Tone languages otherwise arhythmic sequence of events, languages can pitch movements in order to signal prominence. Pitch-accent use languages assign tone only to the most prominent syllable of the word; in this respect, they a mixed type of languages. Both Spanish and English are stress languages, but stress does not play the role in each of these languages. The same placement of stress is conditioned by the presence or absence of three phonological parameters, namely length, weight and tenseness. The syllable structure of English, where weight, length and tenseness play of stress, and vice versa. This is not the an important role, determines the placement with Spanish, where syllable structure case in principle, independent. All this has clear and stress are, rhythm of the two languages. While in English stress device, in Spanish it is only a language the - English as a an ultimate stress-timed language and Spanish the effects of stress on a of the language indeed the as a case of a - the rhythmic more language such as likely a the seems to - work in in both its are necessary in a language is to be stress-timed, which is English. 26 on syllable-timed linguistic system phonetic realisation and in the number of stress-related rules that grammar as a consequence analysed in chapter 2. This controversial classification that the greater sense function for the prominence device, that is, it does not alter the rhythmicality of the utterance. All this has classification of can consequences - Speech prosody 2. In the next sections I shall show that and, as a consequence, Section 2.1. constitutes a English and Spanish differ in terms of rhythm in the constraints summary carried out in relation to the the stressability of their syllables.- on of the different studies, analyses and experiments study of prosodic typologies, paying especial attention to those theories devoted to the statuses of Spanish and English as syllable- and stress- languages, respectively. This classification will be questioned throughout this timed section. In sections 2.2. and 2.3. I shall analyse the speech prosody of Spanish and English, exploring their historical evolution with regard to prosody in order to then concentrate specific theory of stress assignment for each of the two languages. analysis, I will exclusively In my pay attention to the phonological behaviour of languages, rather than to phonetic evidence. Section 2.4. presents the reader these with on a a summary. 2.1. Prosodic timed typologies: the dichotomy between stress-timed and syllable- languages. English and Spanish as typological models 1938, Trubetzkoy described languages from a rhythmic point of view, although he In got to develop a complex theory. Some years later, in 1945, the anthropologist never Pike proposed Pike's a series of criteria for the typological classification of languages. hypotheses were completely based according to which similar events they are different recur at on a temporal view of speech rhythm, regular intervals in time, that is to isochronous. The nuclear idea in Pike's theory is that languages will be depending on which elements constitute those isochronic events. of speech rhythm considered by Pike are, on syllable-timed rhythm, characterised by the fact that syllables come at The two types in time and, intervals. on the other hand, stress-timed rhythm, where stresses According to the above differentiation, there languages, namely syllable-timed languages as being rhythmically equal - - are two the group hand, equal intervals occur at equal major types of - where the stresses fall at the syllable is obviously considered major unit of rhythm, while in the latter this role is played by the foot 27 one where syllables tend to be perceived and stress-timed languages perceptually equal intervals. In the former the say, - a stretch of utterance which begins with a stressed syllable and includes all unstressed syllables following it, up to (but not including) the next stressed syllable. I shall now discuss the scientific basis of isochrony, the pillar on which Pike's theory is supported. Measurements have shown that strict isochrony does not really exist and that deviations from foot foot - isochrony depend on just like deviations from syllable isochrony the number of syllables in are related to the number of segments in a syllable. There is only limited evidence for stress-timing, but none any a of value for the physiology of stress-timing. Nevertheless, it has also been shown that, under certain conditions, such deviations tend to be psychologically disregarded in speech perception and, than it this as a result, speech tends to be heard really is (Barnwell 1971; Huggins 1972 issue, some b; Coleman speaker to produce isochrony, as well perceive it, which emphasises the fact that it has In relation to perceived would like to underline isochrony does not as on never the limits of been clear how isochrony might be permissible, and perhaps required, for much deviation from strict sentences to be 1974).12 rhythmical scholars (Lehiste 1977:255) have pointed to the necessity to determine the limits of the the listener to a, as more as regular (Cummins and Port 1998:146). In that, from presuppose a any case, I phonological point of view, talking about the existence of acoustic exactness (Giegerich 1985:185). Going back to Pike's rhythmic typologies, their further says are development. Abercrombie (1967) makes linguists have contributed to an important claim when he that Pike's two rhythm categories of stress-timed and syllable-timed languages mutually exclusive rather than a stress-timed - nowadays, Pike's classification is seen as a continuum dichotomy. Dauer (1983:51) emphasises the characteristics of purely languages, highlighting the fact that interstress intervals often contain different number of syllables, which forces those syllables to be either compressed spread out by the speaker in order to maintain 12 many In order to determine whether a or perceivable similar spacing of perceptual isochrony actually existed, Woodrow (1951) and Fraisse (1964) carried out several psychological experiments. First of all, they showed that, in order for sounds to be perceived in terms of a rhythmic pattern, their rate of succession must fall within a limited temporal range: if the rate is much greater than five repetitions per second, the sounds tend to be heard as a continuous stream; if, on the contrary, the rate falls below one repetition every two seconds, the sounds tend to be perceived individually (Hyman 1977:2). Secondly, they proved that, due to psychological factors, humans always tend to perceive temporal patterned events as rhythmically structured whether they are or not (Cummins and Port 1998:146). 28 . in stress-timed stresses syllables are never two stresses compressed will vary Focusing languages. On the other hand, in syllable-timed languages, on or spread out and, consequently, the distance between according to the number of intervening syllables. specific languages, Pike (1945) classifies English as prototype of a stress-timed language, while Spanish is considered a very the good example of a syllable-timed language. Up to the present day, nearly all the works that study of speech rhythm in these and other languages take Pike's theory deal with the as a or to theoretical point of departure, either to reassert the validity of his classifications contradict them. Abercrombie (1967), Ladefoged (1967, 1975) and Allen (1975), among many others, have supported Pike's views, contributing to the spreading and acceptance of the latter. These linguists follow Pike and enlarge the inventory of both stress- and syllable-timed languages, including Arabic, all Germanic languages, Russian and the already mentioned English among the languages of accentual isochrony (see Pike 1945; Bolinger 1965; Abercrombie 1967; Halliday 1967; Ladefoged 1967, among others), and French (see Delattre 1951, 1965; Abercrombie 1967; Ladefoged 1975; Catford 1975) been 1977), Spanish (see Pike 1945; Gili Gaya 1950) and Japanese (Ladefoged those of syllabic isochrony. Nevertheless, in the last among scholars some or criteria for the languages among which we can find Spanish - come up as on the strictly with new rhythmic typological classification of the languages of the world. English, there Most theorists agree on we can - syllable-timed, thus pointing to the necessity to In relation to what there have (Pointon 1980; Fant 1984) who have cast doubt classification of certain stress-timed years call an seems to its classification as a be little doubt concerning its status. stress-timed language. Its rhythm is "accentually measured rhythm" (Sosa 1999:53), which means that the primary stresses in without taking into account the unstressed syllables between them. As mentioned a sentence are situated at approximately equal intervals, before, speakers tend to adapt their rate of speech to the number of interstress syllables: if there is none, speaker talks faster, maintained so the speaker talks slower, and if there are several ones, the that the beat established by the stressed syllables is throughout the utterance. 29 application of the notion of isochrony to the prosody of English is not The at new Jones all. In fact, before Abercrombie wrote about the rhythm of English, or (1950:237-238) had already defended the isochronic nature of the interstress spaces as one of the most salient characteristics of the rhythm of English, stating that "[tjhere is a strong each other as the Pike tendency in connected speech to make stressed syllables follow nearly as possible at equal distances". Jones illustrated his theory with help of musical transcriptions, already pointing to the connection between speech rhythm and musical rhythm, an idea that Liberman (1979) would apply development of the grid theory. The scientific basis of isochrony later for the championed by Allen (1973:77), who argued that also range our was feeling that English is preponderance of intervals whose durations lie in the stress-timed arises from the narrow some years from 0.4 to 0.7 seconds, which is in agreement with the findings made by Woodrow (1951) and Fraisse (1964). Later research works replicated this idea. Ladefoged (1975:103) insisted on the various factors that, to his knowledge, are responsible for the isochronic phenomenon. According to him, the combination of those factors operates as regular rhythm, later. though there a statement (1981:44) own conspiracy to make English maintain which Roach (1982) and Dauer (1983) would Bolinger (1981) added he mentioned that were a a nuance to isochrony is a a use years the analysis of isochrony in English when phenomenon forced in production. In Bolinger's words, "[sjpeakers do adjust accents, where it is practical to do so (that is, where the stress and vowel structure permits), in order to get roughly equal intervals". Just like the former, is a Halliday (1985:272) emphasised the idea that "[tjhere strong tendency in English for the salient syllables to occur at regular intervals; speakers of English like their feet to be all roughly the isochrony as same length". The notion of the main regulating rhythmic device gives rise to the idea of the foot the basis of speech as rhythm in English. Catford (1977:182) states that "English speech tends to be delivered in a series of feet, and that feet tend to be isochronous and isodynamic". In contrast to all the existence of isochrony and, other scholars - Classe previous authors, who, on on the other, classify English (1939), Dauer (1983) - as one hand, assert the stress-timed, there whose opinion is regard to these two issues. For Classe (1939:85), isochrony 30 the may more are sceptical with exist, but only under favourable circumstances: "the groups numbers of differ too a widely; the grammatical connection between the word utterance or contiguous stresses, different so, author is Dauer groups must be similar".. topic has been proving that the number of syllables always directly proportional to the length between is strictly speaking, isochrony is just regard to the classification of English a very syllables; the phonetic structure of the component syllables must not Classe's main contribution to the in concerned must not contain as a (1983:52), who states that a perceptual illusion. With stress-timed language, the most sceptical no one has been able to prove that there is tendency towards isochrony in English. point, I have tried to summarise the view of several significant authors To this in reference English to two as a issues, namely the existence of isochrony and the classification of stress-timed language. The authors who defend isochrony do not have doubts about the status of existent any phenomenon English, whereas those who are not see isochrony as a non¬ convinced about the straightforward classification of language according to the isochronic parameter. Since the notion of isochrony seems a very useful tool for the analysis of the rhythms of different languages, for the purposes of this dissertation, I shall assume that English is an isochronic language. When one turns out to be analyses the status of Romance languages, the isochronic view even more be French, often quoted controversial. The most uncontroversial as "[sjuccessive syllables within language seems to being the perfect prototype of syllabic isochrony: a sense group are very loudness, with the exception of the last, which is, similar in duration, pitch, and on the average, higher in pitch, slightly softer, and much longer in duration" (Delattre 1966:78). French has been claimed to belong to the accentual 1982). Italian, Portuguese - on group (Grammont 1933; Wenk and Wioland the other hand, is claimed to be clearly stress-timed, while especially Brazilian Portuguese - shows a tendency towards stress- timing (Major 1981:350). Spanish has been described in either group. Pike's view of Spanish prototype of syllabic rhythm is shared by Gili Gaya (1950), Delattre (1965, Abercrombie as the 1966), (1967) and Olsen (1972), and rejected by Navarro Tomas (1918) Allen (1975) and Pointon (1980). Contrary to Pike's view is the Spanish linguist Navarro 31 (1918), who, Tomas affirmed that Spanish is remain stable. Allen that the there the basis of the results of a series of acoustic experiments, on are uneven in the duration of syllables, while accentuated feet (1975) reasserts Navarro Tomas's idea, emphasising the idea occasions when there accentual alternations in to be seem Spanish (Allen 1975:80-81). Other authors underline the irregularity of accentual groups compared to the relative uniformity of syllabic duration in Spanish. Delattre (1965, 1966) rejects the validity of Navarro Tomas's experiments and claims that rhythm in Spanish is syllable-timed in contrast to English, (1972). On the other hand, going Pointon nor a a step view with which Olsen a agrees further from Gili Gaya and Olsen's analyses, (1980:302) boldly suggests that Spanish is neither a stress-timed language syllable-timed language: The most has no probable solution seems to regular rhythm in the similar events, be they syllables be that, of sense an the evidence, Spanish on isochronous but there is or stresses, sequence some of form of segment-timing, in which the number and type of segments in each syllable, together with the duration of a presence or absence of stress, determine the syllable. However, Pointon (1980:302) admits that "to the ears of native English speakers, Spanish sounds syllable-timed, and it is not surprising that English speaking investigators should want to find syllable-timing, view shared tend to they by Bond and Fokes (1985), who assume are say even when it does not exist", a that non-native speakers of English psychologically the differences between their own languages - when syllable-timed - and stress-timed languages like English. Having in mind the controversies that Pike's typologies have generated, there have been categories some - attempts to expand the dichotomy by adding to it new rhythm for instance, the category of mora-timing, to which Japanese is said to belong -, to replace the traditional classification by vs. 'trailer-timed' -, or to inter-syllable intervals many - some relax the requirement for isochrony of inter-stress and stress- vs. syllable-based languages. Most phonologists accept the hypothesis that there features of both stress-timing and syllable-timing. As of empirical support other opposition - 'leader-' a are significantly, languages which show result, and in view of the lack for the stress-timing/syllable-timing classification, scholars such 32 as Dauer scale (1983) prefer to view speech rhythm not dichotomy but rather as a as a continuum, with two hypothetical languages, one perfectly stress-timed and or the other one syllable-timed, marking the ends of the scale (Dauer 1983; Bertinetta 1989): ► French Japanese In this Spanish vein, after carrying out rejects the idea of a a Greek English Portuguese series of perceptual experiments, Miller (1984) dichotomy. He defends the scalar hypothesis by saying that languages share features of both syllable and stress isochrony. For him, while languages belong clearly to one or the other group, some others fluctuate between the two (Miller 1984:75-76). Theorists who support the scalar hypothesis claim that the rhythmic languages result from the interaction of a variety of components. differences between this interaction, each language be given Depending on which will correspond to the position of that language can on According to Dauer (1983) and Bertinetto (1989), these that determine the score of a an overall rhythm the continuum are some seen score above. of the components language: a) Vowel quality: In stress-timed languages, stressed vowels have full articulation, while unstressed ones are vowels retain their distinct reduced centralised. In or syllable-timed languages, quality in both stressed and unstressed syllables. b) Consonant quality and quantity: In stress-timed languages, segmental quality and quantity distinctions are manifest in stressed syllables. In syllable-timed languages, quality, quantity and stress are independent. c) Syllable duration: In stress-timed languages, stressed syllables than unstressed ones and unstressed are usually longer syllables usually undergo compression. In syllable-timed languages, stressed syllables are only slightly longer. d) Compensatory shortening: In stress-timed languages, the stressed vowel is or less it. In compressed depending on more the number of unstressed syllables surrounding syllable-timed languages, there is no compensatory shortening. e) Syllable structure: In stress-timed languages, there is a wide variety of syllable types, and relatively uncertain syllable boundaries. In syllable-timed languages, CV and CVC predominate, with well-defined syllable boundaries. 33 f) Relationship between syllable structure and stress: In stress-timed languages, heavy syllables tend to be stressed, while light syllables tend to be unstressed. In syllable-timed languages, syllable weight and stress are independent. g) Position of stress: In stress-timed languages, the intervals between stresses relatively short; when they timed are long, new stresses can are be introduced. In syllable- languages, intervals have variable durations; there be stress clashes, can shift.13 not stress h) Intonation: In stress-timed languages, pitch change corresponds with stress. In syllable-timed languages, they are independent variables. i) Tone: In stress-timed languages, it falls only timed languages, it After actually based they could rhythm, even fall on hypothesis on seems to Pike's dichotomy seem. stressed syllables. In syllable- both stressed and unstressed syllables. examining the different theories that the scalar as can on on rhythm typologies, be the most plausible means on a seem to prove conclude The fact that it is that the two alternatives What these controversies if this rhythm is not based one. we can are not as at odds is that languages have perfect isochrony. The perception of a relatively constant periodicity between prominent elements, be they syllables stresses, is what makes them 13 or rhythmical. might object that the so-called 'emphatic constructions', which can take place both in languages and also in syllable-timed languages, are cases of stress shift. In this respect, I must say that, according to linguistic theory, there are two main types of stress shift (see Dogil 1979). The first one is phonologically driven: for eurhythmic purposes - in metrical phonology, the term 'eurhythmy' refers to the modification of lexical stress patterns to achieve strong/weak alternation at the surface level (Selkirk, 1984; Hayes, 1984) -, the main stress of a word is shifted to another syllable, so that the metrical pattern of the corresponding utterance is better formed rhythmically speaking - in English, the phenomenon known as 'iambic reversal' is an instance of this (Giegerich 1985:206-218; Hogg and McCully 1987:132-148). The second type of stress shift is semantically driven; it is subdivided into two further types: in the first subtype, known as 'contrastive stress', when speakers want to highlight the contrast between two prefixes or suffixes which are present in different words of the same utterance, they can put stress on those otherwise unstressed syllables - this corresponds to the so-called 'parallel constructions' by Chomsky (1971). In the second subtype, called 'emphatic stress', speakers can put an extra degree of stress upon the focus word, in which case we cannot speak about stress shift, or they can put this extra degree of stress upon another syllable of the focus word, in which case there is stress shift. For the purpose of this dissertation, where I am dealing exclusively with phonological issues, I will only consider the first type of stress shift as a proper instance of this phenomenon. Once this point has been made, we can say that syllable-timed languages, like French or Spanish, do not experience phonologically-driven stress shift. Someone stress-timed 34 2.2. The prosody of Spanish This section analyses the nature of Spanish prosody in order to define the most prosodic characteristics of this language. Section 2.2.1. explores the. salient and evolution of the Spanish prosodic system departing from Classical emergence assignment in Spanish, focusing Latin. In section 2.2.2. I deal with stress on two specific theories, namely Harris (1983) and Navarro Tomas (1965), both of them amended in subsequent works. 2.2.1. From Latin to Spanish prosodic characteristics of Present-day Spanish, it is In order to determine the necessary to overview the historical evolution of the Romance prosodic system, from Classical Latin to the present day. Section 2.2.1.1. explores the Latin quantitative system and its gradual disappearance, as well as the consequences of this process on the stress system of Present-day Spanish. Section 2.2.1.2. analyses the interaction between the vowel system, Classical on the structure of syllables and the assignment of stress in Latin, and how the evolution of the first has had noticeable consequences the other two. 2.2.1.1. Stress and the evolution of the As quantitative system already pointed out in section 1.4.2., the languages of the world can be classified according to the kind of prosodic features they show, that is, according to the which they articulate the flow of sound. Languages such said to be stress languages. This languages, means as as English or way Spanish in are opposed to tone languages and to pitch-accent that English and Spanish make use of stress in order to mark syllable prominence. As explained by Hyman (1975:38), the most important characteristic of stress is its culminative function, which entails that in a stress language only one syllable per word can receive primary stress. Although in all stress languages stress is culminative in nature, they further classified can be regarding the degree of freedom in the assignment of stress (Hyman 1975:204, 1977:39; Quilis 1993:388). In some languages, such as English or Spanish, stress stress' can virtually occupy any position within the word; these languages, where stress is unpredictable, and therefore lexical partially - and phonemic. Other languages, like French by the fixed position of their stress; these are 35 or Polish, are - are 'free at least characterised called 'fixed stress' languages, where stress is predictable, and therefore grammatical and non-phonemic. In languages with fixed or grammatical stress, stress signal used to a stress marks the It is word boundary, or a have a demarcative function, that is, it be morphological category of a specific word - see section 1.4.3.3. that is, free stress and fixed stress languages most common group - one of the exclusively. In fact, the is that of intermediate languages, which show lexical stress for of their lexicon but also some instances of words with fixed stress; moreover, the occurrence of one type occurrence can morphological function, whereby the position of important to note that stress languages do not always belong to two groups - most can of another of stress may or grammatical be overridden by the (Hyman 1975:230, 1977:39-40). This is indeed the case of Spanish and English, although the amount of instances which show grammatical stress as in both free stress Let languages is really reduced - that is why these languages are classified languages. have us now stress in stress look at Table 4 below, where a languages predictable stress - are - lexical all the possible types of unpredictable stress, and grammatical or or applied to examples in Spanish and English: Grammatical Lexical Morphological Demarcative Spanish Spanish Spanish English termino termin-o termin-o convert vs. convert (noun) (stem marker) (tense and Table 4: Di Terent types (noun) (verb) of stress in stress languages. instance of lexical stress the Spanish shows, termino ('term; boundary'), which could only be explained by means of a reference to the one hand, marker) Table 4 on the person an etymology of that word since there is no on noun prosodic justification for the assignment of stress to the antepenult. On the other hand, grammatical stress falls two verbal items, namely termino ('I finish'), where the penult is stressed in order to demarcate the end of the stem, corresponding to the third In on English, grammatical - and termino ('she finished'), where the morpheme -6, person singular in the simple past, is marked prosodically. morphological - 36 stress is applied to those instances such as (noun) and convert (verb), in order to determine the morphological status convert question. of the words in According to the classification of Spanish language, the stress of a hypothetical Spanish 4), and the Table vast majority of a which can going Before that, for the Nevertheless, Spanish nonverbal stress important residual effects of the Latin Stress Rule, something light on, means prosodic information must be provided in the lexicon with cases, number of shed is unpredictable (see termino in applies to adjectives and adverbs. This same reference to the stress of individual items. retains noun predominantly free stress, as a on Spanish.14 the stress assignment system of Present-day I need to point out that, contrary to the view held by Latin grammarians that Latin inherited the tonal system from Ancient Greek, it is widely accepted that Latin stress was of accent in which for reasons came Allen was an says spoken Latin underwent in which accent Penny that "early Latin had a type the predominant element [pitch-accent] [...] However, that remain obscure, have to pitch 1993:150).15 always dynamic (Lloyd (1991:42) explains the evolution of Latin stress and now a change in accent-type and energy-deployment dominated [stress-accent]". As (1964:4) explains, the rules for the location of stress in Classical Latin were simple: or on the antepenultimate if the penultimate is light. Obvious exceptions are It falls on provided in the forms penultimate syllable if this is heavy; the case of monosyllables, of syncopated and apocopated (e.g. illic < illice), and of disyllables with light penultimate. Classical Latin showed a words of three syllables, while light stressable not syllables contain any Allen's which or more - clear preference for stress to fall ones on heavy syllables in qualified preferentially although the latter could be stressed if a polysyllabic word did heavy syllable, or if the first syllable of a disyllabic word quotation does not specify which syllables could be classified ones as as non- as was light. heavy and light in Latin, something which turns out to be essential if we take into 14 Although Spanish nonverbal stress still shows some residual effects of the Latin Stress Rule, it is impossible to conclude that Spanish inherited the Latin Stress Rule directly (see Lipski 1997:549). 15 We must take into account that Latin grammarians translated the prosodic terminology directly from Greek, so their testimony is not reliable (Lloyd 1993:150). Nevertheless, some scholars still hold the view that Latin was a pitch-accent language before it became a stress-accent language (see Beare 1957:210). 37 account that, in polysyllables, the position of stress was determined by the length - and, therefore, by the weight - of the penultimate syllable. Latin polysyllables can be (i) the first classified into three groups: long penultimate vowel a second group includes a one and therefore, - penultimate with heavy syllable; (iii) the third and last by a single short vowel this case, we are a heavy penultimate syllable; (ii) the short vowel followed by two a clusters formed by consonant + /r, 1/ consonants, except formed is constituted by those words which have talking about a or by group a or more this would be, again, - comprises words with short vowel followed by a a penultimate one consonant - in light syllable (Penny 1991:41): E.g.: moli:nu, de:be:re ('mill', 'to have (i) Penult = long vowel to') (ii) Penult = short vowel + > two consonants (except cons. E.g.: sagitta, quadraginta ('arrow', + /l, r/) 'forty') (iii) Penult = short vowel (consonant) E.g.: filius, vetulus, corrigia ('son', 'old', 'strap') Table 5: Classification of Latin polysyllab es according to the weight of their penult. Thus, length in Latin had the function of determining which syllable in a word could be stressed, though it was always subject to two rules, namely the 'three-syllable' rule, which did not allow stress to recede further than the antepenult cases where there was a heavy syllable beyond that position shortening' rule, which would apply to was lightened and the stress could fall appeared to depend directly on 'quantity' is used so syllabic weight, and therefore, as a synonym not even in the and the 'iambic that its final syllable the penult. According to this, Latin stress (1964:5) words, in Latin "the rules of quantity Allen's term on disyllabic word, a -, - are for 'length', and on vowel length. In rules of stressability". The more specifically for 'vowel length'. Hyman (1975:206) connection between agrees with Allen in that Latin is the best example of the syllable weight and stress: "Stress is normally assigned to the penultimate syllable in Latin, except when that syllable is light. In this stress is assigned to the antepenultimate position, 38 as case, the in reficit ('she fixes/remakes'), penultimate syllable fi is light". Hyman's view coincides with that of where the Penny in that they both the penult was a that Latin stress light syllable; in that whether it matter argue was light always penultimate except when was case, stress fell on the antepenultimate, no heavy. Nevertheless, this clear-cut view of the or relationship between weight and stress in Latin is contradicted by Hyman (1977) himself, who states the exceptionality of Latin with regard to other languages with syllable weight phenomena. What Hyman (1977:53) observes is that, in Latin, the search for stressable a - heavy syllable "cannot - go back further than antepenultimate position [note reficit (antepenultimate stress because the penult is light) conspicio ('I contemplate') (antepenultimate stress vs. word with a an underlyingly penultimate stress. (1977:53) revised theory is that Latin had stress word to an syllables in a a Hyman's dominant antepenultimate stress, heavy penult would always attract stress. According to such a changed from - light syllable in heavy pre-antepenult)]", something which makes him question the assumption that Latin words have although on a an underlying initial position - that is, underlying antepenultimate position, a row so on a theory, the first syllable of the that the number of unstressed could be reduced. (2) facilius > facilius ('easier') Hyman's theory proves the controversial state of the discussion assignment of stress in Latin. For the purpose of our discussion, we on the will adopt Penny's widely-accepted view (see Table 5). As some regards poetic prosody, what was special about Classical Latin is that syllables, being phonemically heavy but not accented due to positional rules light accented syllables qualified - potentially note that there could also be stressable. This allowed, first, for the fact that syllables which were not stressed when a specific word was -, taken in isolation could be stressed in a as specific context and, second, for the phenomenon that certain primarily stressed syllables could see their stress 'devaluated' due to lapses a or clashes. We can positional factors develop a - that is, in order to avoid stress clearer idea of these two concepts by looking at poetic fragment in Latin (Horace, in Gasparov 1996:86), where the stressed 39 syllables are signalled with syllables are in bold: (3) a graphic accent, while heavy - and, in Latin, stressable - vi-tae / scelerisque Integer Honest life-of crime-of 'The person purus pure who is honest and a good citizen/ Non eget Mau-ris / iaculis neaue ar-cu Not need-3rdp. Moors-of darts-of nor does not need darts or bows/ Nec vene-na-tis / Nor venomous-of full or a bow-of gravida sagit-tis arrows-of quiver full of venomous arrows/ Fusee, pharet-ra... Fuscio quiver oh, Fuscio'. As the observe in the above one can in bold - one does not hand, we syllables - which ictus is an syllable, they so falls Stressable observe that there are are heavy and stressable because they contain assigned. On the other hand, there - and the ictus - are has to fall long vowel a words without on a any - to stressable light syllable. The most syllable on which Non-stressable (heavy) syllables with a metrical = word stress X Ictus = syllable (') = syllables with poetic word stress syllable E.g.: Eget, gravida venenatis Table 6: Stress and ictus in Latin. 16 word. On are rare: E.g.: Integer, vitae, Mauris, iaculis, arcu, every words with non-stressed, yet stressable is the first one, while the instances of light poetic ictus (') stressable fragment, the poetic ictus in each foot - represented necessarily coincide with the linguistic stress of the stress common case stress 16 Unless otherwise stated, all translations Latin-English and Spanish-English 40 are mine. X = light Latin stress assignment rules relied, on the one hand, on syllabic In sum, weight, which was a synonym for quantity, and, the other, on syllables and stressable syllables. Stressable syllables necessarily (as the existence of of syllables with regards to stress, namely stressed syllables, unstressed three types around on carry stress. This translated into in (3)) without giving rise to seen heavy, but they did not were idiosyncratic freedom to shift stress an any conflict between stress and ictus in poetic texts (Allen 1964:11). Latin, as most languages which have a vowel length contrast, had also vowel quality differences in the corresponding series of long and short vowels (Hyman 1977:48). There were two series of vowels, namely tense and lax, which roughly corresponded to the categories of stressable and non-stressable vowels. If a lax vowel became stressable in the vowels in a similarly, and, eventually, reduced, its stress system. fixed so a tense vowel which stressable was not accented would become lax on the by claiming that the Latin vowel system had a direct impact as vowel antepenult, depended essentially the phonological distinction on were mapped onto the distinction Latin are not Indeed, phonemic, reasons as for the it is in contemporary Spanish (Lloyd 1993:191). gradual disappearance of the quantitative system in exclusive to that language; rather, they languages. Scholars have long pointed out that a seem to be shared by most well-known characteristic of Indo- European languages in general is the tendency to eliminate phonological quantity a feature of the vocalic system with the fact that on quantity distinctions in language the presence or as (Lloyd 1993:180-181). This phenomenon has to do are degrees: long and short. Qualitative distinctions based and as length stopped being phonemic and became exclusively phonetic, stress itself became The on As already noted, in Classical Latin the position of stress, which between tense and lax vowels and, therefore, on the heaviness of syllables. soon - it would no longer be stressable. long and short vowels, which in turn between - relationship between stress, syllable weight, vowel length and quality could be summarised was were no tense word, it would not only be its length that changed, but also its quality - from lax to tense; The special circumstance that there absence of tension are not so universally reduced to two on limited. Quantity is, thus, less important than the other factors in the vocalic system of 41 the height of the tongue a language, that being the why the former reason favourable. In Latin, and became stress a can quantity non-distinctive easily be left behind when the conditions are not was not an independent distinctive feature phonetic feature which depended within the words, as well as on some on any longer the position of other factors (Lloyd 1993:179). In Beare's (1957:215) words, "by the fourth century A.D. Latin vowels their old quantitative difference, though they retained their qualitative difference [...] These qualitative differences In sum, to a losing retained by the Romance languages". the vocalic system in Latin evolved from a quantity-sensitive system quantity-neutral system, affected by several circumstances, amongst which qualitative the we (i) the relative inefficacy of quantitative distinctions in comparison with observe on are were ones; and (ii) the action of stress, which made quantity depend partially position of the former, and syllables exclusively. The quantity an as so concurrence it reduced quantity distinctions to stressed of these factors brought about a collapse of independent phonological feature in the vocalic system (Lloyd 1993:184). Spanish did not inherit the connection between stress, syllable weight and vowel length. As Quilis (1993:377) observes, after the complete collapse of the quantitative system, "quantity is not pertinent in Spanish, although it is true that in certain positions, and more often in a high register, observed".17 Thus, length in Spanish is just longer an short in a can be phonetic correlate of stress, and no a quantitative difference independent feature of vowels. That is, vowels a phonological lengthened from a sense; however, when they are not are inherently long stressed, they are or also phonetic point of view. In D'lntrono et aVs (1995:129) words, "quantity is, together with tone, the most important index of stress in Spanish. Tonic syllables 17 are regularly longer than short ones, so the total energy tends to be bigger". Hispanists (Garcia-Bellido 1993) do not agree with the view that phonological quantity is longer present in contemporary Spanish. Indeed, scholars such as Harris (1983) relate the assignment of stress in Spanish nominals to the quantity of the penult, thus building up a system parallel to that of Latin. On the other hand, Aske (1990), Eddington (2000) and Barkanyi (2002), among others, argue that "the apparent quantity sensitivity of Spanish nominals is a historical heritage, not an active constraint", and that "stress is rather the lexical property of words" (Barkanyi 2002:375); in other words, stress is lexical. Certain no 42 2.2.1.2. Vowel systems and syllable structure in Classical Latin consisted of ten vocalic phonemes, which could The vowel system be classified according to three features, namely degree of articulation and in openness, place of length. In spelling, grammarians usually represented the differences length by adding a macron or a micron on top of the corresponding vowel length is represented with the diacritic <:>. The phonemic status of those Table 7, ten vowels is shown in Table 8, which presents some Latin minimal pairs: Back Centre Front High /i://i/ /u://u/ Mid /e://e/ /o:/o/ Low /a://a/ Table 7: Latin vocalic system (Penny 1991:45). Short vowel Long vowel HP.C 'here' HIC 'this' VE.-NIT 'he came' VENIT 'he comes' MA.-LUM 'apple' MALUM 'misfortune' PO.-PULUS 'white POPULUS 'people' Table 8: Vlinimal As in most were in - mostly Romance poplar' pairs in Latin (adapted from Penny 1991:45). languages, Latin vowels constituted the syllabic nucleus. Syllables open in Classical Latin, a tendency which has been inherited by languages. As Lloyd (1993:86) highlights, "Romance has preferred from the time of Old Latin". The fact that the prototypical syllables to closed ones syllable in Latin had was open a consequence on syllabic nucleus, namely that they 1993:148). This fact important effect on open was not were the consonants that followed the "especially subject to erosion" (Lloyd exclusive to Latin; rather, it has always had an syllable-final consonants in Romance languages. In Spanish, for instance, codas have historically experienced some kind of erosion due to "the predominance of open syllables together with the tendency to make syllables as open as possible" (Lloyd 1993:552). Quilis (1993:368-371) 43 argues that Spanish evinces a clear tendency towards the most open syllables. The syllables that end with a vowel (V, CV, those, the syllable CV represents than half CCV) are of the Spanish syllabic structures; in contrast, the percentage of syllables with two common; among postnuclear consonants is very more low. Present-day Spanish has five vocalic phonemes (D'Introno et al. 1995:104): Front Back High i u Mid e o Low a Table 9: corresponding The cannot open and closed realisations happen randomly, and they specific phonetic contexts. Quilis chooses not to include be associated with them in the Spanish vocalic phonemes. description of the allophones corresponding to the vowel phonemes in Spanish (D'lntrono et al. 1995:93-94). In articulatory terms, there is not variation between stressed and unstressed vowels. In of the two are more positions is not very fact, the phonological inventory different, although the oppositions in tonic position fruitful than those in unstressed position, as exemplified in Table 10: piso 'apartment' peso 'weight' 'sediment' puso 'he put' poso a great pkso 'step' Table In any case, unstressed Spanish vocalic phonemes in tonic position. the smaller amount of energy which is generally applied to syllables makes articulatory tension lower, and affected. In their structure 0: so the quality of vowels is phonetic realisation, all the vowels tend to get closer to the acoustic of the schwa vowel When vowels are vowel tends to lose its (D'Introno et al. 1995:97-98). in context, their quality may change. The contexts where quality will be those where several vowels vocalised consonants such as liquids and voiced approximants 44 go one or a vowels plus after the other: - (4) Nunca he oido hablar de ellos ('I have never heard about them'). In (4), there is a succession of six vowels, only interrupted by the approximant- realisation of Idl. D'Introno et al. are sounds which do not schwa. In there are (1995:99-100) explain that, in such contexts, there general terms, the realisations which two contiguous vowels, particularly the falling towards the - for instance, are found when belong in the Spanish allophone inventory one are closest to schwa of which disappears. In the case of diphthongs, the quality of the second vowel normally gets altered ones, configuration of schwa - indeed, it is often pronounced as such - or even disappears (D'Introno et al. 1995:101-102). above, in Spanish the most common syllable structure is CV. As mentioned Nunez-Cerdeno word corpus, of the last Last (1999:211) has carried out in which he observes syllable in a syllable ends in a a study based on an electronic 91000- clear relation between the segmental structure word, and the placement of stress in that word: a vowel Last syllable ends in a consonant Penult 57,911 88.00% Ultima 26,642 97.80% Antepenult 7,327 11.10% Penult 512 2.03% Ultima 573 0.87% Antepenult 35 0.05% TOTAL 65,811 100% TOTAL 25,189 100% Table 11: Syllable structure and stress placement in Spanish. Table 11 illustrates three main facts about the and stress relationship between syllable structure placement in Spanish: (i) that the vast majority of Spanish words end in a vowel, (ii) that the unmarked stress position for words that end in a vowel is the penult and (iii) that the unmarked stress position for words that end in a consonant is the ultima. All this could lead us to think that Spanish has preserved the Latin stress position; whether it has also preserved the Latin quantity-sensitive system itself or not is, as we have noted, less clear. 45 2.2.2. Stress assignment in Present-day Spanish After that succinct seems analysis of the evolution of the Romance prosodic system, it in order to concentrate on the characteristics of stress assignment in Present- day Spanish. In sections 2.2.2.1 and 2.2.2.2. two classical theories are presented and amended. 2.2.2.1. Harris's analysis amended by Roca (1983) book, Harris offers In his Spanish stress system, and relying nouns, more Phonology. Framing his study Harris subject to the same some theory, Roca (1997) offers a revised significant new ideas. (1983:83) bases his analytical apparatus on the observation that the Spanish stress system follow from the peculiar morphological placement. In his view, Spanish nonverbal stress is unmarked stress a on words, where etymology plays an essential role, plus the contrast structures of vs. specifically, of the rules of stress assignment for analysis, adding characteristics of the marked of the most widely accepted accounts of the the theoretical tools provided by the school of Generative on version of Harris's one number of restrictions from the right edge of the lexical word (Harris 1983:85; Roca 1988:398). As a general observation, Spanish stress must fall syllables of the word, stress - never casa, of the last three removed further to the left. The least-marked type of especially if the word ends in (5) on one a vowel - falls on the penult (see Table 11). 1 o armario, encuesta, latifundio ('house', 'wardrobe', 'poll', 'estate') However, penultimate stress is unacceptable if the last syllable contains a falling diphthong - [ai, oi, ei, au, ou, eu], (6) *convoy ('convoy') As for ends with 18 a antepenultimate stress, the least marked instances when a word vowel: Penultimate stress is four times is the least occur frequent than oxytone stress and antepenultimate stress, which See Quilis (1993:400-403) for a thorough account of the frequencies of the patterns in Spanish. frequent different accentual more one. 46 (7) telefono, autentica, transito ('telephone', 'authentic', 'transit') Unless the plural will be so singular form of a - is antepenultimate, which implies that the noun the plural basically adds leaving the stress pattern unchanged -, -5 to the last vowel of the singular, antepenultimate stress in consonant-final severely restricted: words is (8) telefono, telefonos Antepenultimate stress words of Greek occurs in only two sets of forms: (i) a closed group of origin: (9) hiperbaton, metatesis ('hyperbaton', 'metathesis') and (ii) a small group of 'random idiosyncratic words': (10) regimen, Jupiter, Alvarez ('regime', 'Jupiter', 'Alvarez') On the other means stress hand, if the penultimate syllable is heavy, which in Spanish that it is closed is highly marked by or, a consonant or containing for Harris (1983:88), even a semivowel, antepenultimate impossible: (11) *masajista, *catapulta, * enigma ('masseur', 'catapult', 'enigma') Antepenultimate stress is also impossible if the last syllable contains diphthong: (12) *experiencia, *negocio, *calvicie ('experience', 'business', 'baldness') Last, oxytone stress is unmarked if the last syllable is heavy, that is, closed: (13) albanil, mirador, cajon ('bricklayer', 'viewpoint', 'drawer') 47 a It is final possible to have consonant-final words where the stress falls on a non- syllable; such instances (14) examen, ductil, are seen as azucar marked yet acceptable: ('exam', 'ductile', 'sugar') Nevertheless, if the final 'consonant' of a word is unacceptable, (15) as we have seen convoy, caray consonant. If they end in are a end in seem to be marked in that they feel less natural can end either in vowel, stress is prototypically placed on a vowel or in a the penult; if they syllable. Although exceptions to these rules consonant, stress falls on the last possible, and indeed well formed: widely attested in Spanish. In contrast with vowel- Spanish nonverbal items a are ('convoy', 'good gracious!') penultimates, these oxytones than the former. In sum, glide, non-final stress is above, while glide-final oxytones Vowel-final oxytones are final a numerous, stress will invariably fall on one of the last three syllables of the word (Roca 1997:631). Table 12 summarises the acceptable and unacceptable occurrences of stress in Spanish: Type of stress Consonant-final words Vowel-final Oxytone albahil, mirador, cajon convoy, caray Penultimate examen, ductil, azucar casa, (glide-final) words armario, encuesta, latifundio, *convoy Antepenultimate hiperbaton, metatesis telefono, autentica, transito regimen, Jupiter, *masajista, *catapulta, * enigma Alvarez *experiencia, *negocio, *calvicie Table 12: Stress in , Spanish. By observing the previous descriptive account of Spanish nonverbal stress, the question which arises is why antepenultimate stress with penultimate or final heavy syllables is restricted in Spanish. In order to account for that problem, Harris 48 (1983:88) states that Spanish has inherited from Latin the principle that "antepenultimate stress is impossible if the penultimate syllable has rhyme", which is the case in (11) and (12) above. According to this observation^ Spanish would have to be classified as However clear Harris's quantity-sensitivity rule are branching a a quantity-sensitive language. theory might quite numerous the exceptions to seem, a potential and varied in nature (Roca 1988:417). For once, referring to penultimates, although the majority of superheavy syllables, with three or more in a noun positions in their rhyme, must - and indeed do like tora[ks] and a few other Latinisms "a (Roca 1988:404). Second, weak" groups bear final stress, superheavy ultima is manifestly have already analysed two exceptional word above, Greek words and Spanish idiosyncratic words see - we - antepenultimate stress, something which contradicts the theory. But the most definitely overrides Harris's theory is that of syllable in a - with case that stressless penultimate heavy non-idiosyncratic, perfectly Spanish antepenultimate place-name like a Fromista, mirrored by some foreign words which seem to be freely exempted from the above constraint, like Robinson, According to Roca (1988:418), the turns the Washington occurrence or Manchester (Roca 1988:416). of such foreign words effectively regimen group into "an open class, thus increasing the pressure for a principled solution". As noted before, although most members of the class do indeed have a foreign Spanish. The stress or erudite flavour reason - see hiperbaton at least regimen is genuinely for the existence of these exceptions is indeed historical: the locus inherited from Latin has been almost Given that all the - completely preserved in Spanish. exceptions previously explored climb to amount, Roca (1988:417) concludes that an extraordinarily high Spanish stress is not quantity-sensitive. His arguments are reinforced from different perspectives, both by Hayes's theory that trochaic systems such as Spanish are prototypically quantity-insensitive (in Lipsky 1997:563) and by Trubetzkoy's idea that stress sensitivity to syllable quantity presupposes a systematic vowel length contrast (in Roca 1997:621), tense/lax contrast The sensitivity stress (Hyman 1977:48), both of which are as well as a absent in Spanish. contradictory evidence given by different authors to defend the quantity or insensitivity of Spanish leaves a theoretical gap with regard to Spanish assignment, which Roca (1997:627) tries to fill in by introducing the term 49 'accent', used to encode deviations from the normal pattern. To start with, Roca adapts the following rule from Halle (1991:145): "Accent [...] the rightmost (metrical) syllable". Subsequently, Roca formulates the principle that in Spanish especially if the corresponding rhyme is heavy stem-final stress is unmarked, followed by palatal consonant. Finally, Roca (1997:644) concludes that a we either on some Spanish words leftward from the can show weaker stresses, too. These the initial syllable of the word gramaticalidad ('grammaticality') primary stress - - or on occur are adjacent to each other or to avoided. analysis amended by D'lntrono et al. 2.2.2.2. Navarro Tomas's (1965) book, Navarro Tomas established long been regarded which have generative) ('generativist'), - generativo, gramaticalidad. The only condition primary (Harris 1983:86) - that is, stress clashes In his may even-numbered syllables counting non-primary stresses is that they cannot attached to the assignment of primary stress in Spanish, it is have dealt with the important to note that appear any principle must be lexically marked. deviation from this Once or as an a series of rules of stress assignment authority for any study of Spanish phonology. Nevertheless, Navarro Tomas's principles have recently been considered insufficient, and have thus been amended in Taking stress a only two main functions, namely in just by the same sequence (1995:161-173) make culminative first a tertiary stress. As assignment in a of phonemes. Navarro Tomas's on existence of three different In a tonic syllable, while the former points to the phonemic value of one Always relying stress distinctive and Spanish, that is, the position of stress distinguishes between meanings of words formed and a (D'lntrono et al. 1995:124-225). The latter refers to the fact that Spanish words have stress comprehensive volume by D'lntrono et al. (1995). different path from Quilis (1993), these authors start by stating that Spanish shows function a very observations, D'lntrono et al. important observation when they assert the degrees of stress in Spanish, namely primary, secondary second step, they proceed to explain the rules of primary nouns and adjectives. Spanish, the domain of stress assignment is what D'lntrono et al. (1995:162) call "the minimal word", which is constituted by what they call "a lexeme" and a gender suffix - pro to typically -a, -o, 50 although there are others. Their morpheme alone includes the lexical notion of 'lexeme' derivational suffix, the lexical morpheme together with this if there is or, one. any Stress is assigned according to two basic rules, which will be quoted hereafter as 'Rule A' and 'Rule B' (D'Introno et al. 1995:164-165): Rule A Syllable Stress Rule: The first closed syllable Closed - - counting from right to left - of the minimal word must be stressed. Rules B There are - Default Rule: Stress the second syllable. several observations to be made reference to the above rules, which I will illustrate with antepenultimates with the two words such as ('brother'), hermano Spanish words with some ending in consonant, penultimates with works for oxytones closed the behaviour of on a three following syllables or persona ('person') examples. Rule A closed penult and open. or cartera Nevertheless, ('purse') pose a problem to this rule, given that their structure would require them to be antepenultimates (see D'lntrono et al. 1995:165) - the solution to this apparent problem will be explained later. Similarly, Rule B does not antepenultimates like termino/sabana ('sheet') this is to establish case which end with consonant syllables a open - stress et inflectional There above all, - paroxytones which end with a must be assumed to be lexically marked if there is no for them to be stressed. The rest of the words get their or, this one lacking, to Rule B. In are other groups an any case, as interesting fact: stress analysis of words which do not D'lntrono never falls under examination leaves ('hypnosis'), hipnosis ('hepatitis') ('syphilis') without or s iflis an endings -sis, -tis, and -is do not count does not count at all - note that this solution, words like seem to respond to any of the further explanation (D'lntrono et al. 1995:165ff). First of a ('metamorphosis'), the -, morphemes. rules, and require the silaba ('syllable'). The solution in and proparoxytones with the last three (1995:165) highlight, this theory points to al. on - reason according to Rule A sofa ('sofa') ('corpse') like silaba phono-morphological like - like cadaver - apply to opposition marked/unmarked stress, where oxytones an vowel or seem to words hipotesis such as metamorfosis ('hypothesis'), explanation. D'lntrono et al. as hepatitis assume that derivational suffixes, and that final -s they do not change to form the plural. According to hepatitis or artritis would be obtained applying Rule B, 51 while the others would be labelled words -5 ending with final is not a -es, suffix, contrast with words like that so baul ('trunk') as jaula ('cage') or or Raiza, with derived from high vowels, are includes Rule B applies again. A word like Jupiter is said to be a maiz ('com'), with hiatus, diphthong where glides [u] in the first case and [i] in the second, followed by glides case like viernes ('Friday') and Hercules. In these words, the final lexically marked. Last, words such the lexically marked. The second as application of Rule B there is a process with another vowel becomes glide a - we can by means a we find a vowel; if we admit sustain the view that before the of which a high vowel in contact sliding. At this point, Rule B can no longer be applied to the glide, but rather to the vowel. explanation of the apparent exceptions The to establish sequence a apply Rule A - except when there is a derivational suffix A several derivational are more above forces D'Introno et al. in the application of the rules, which except when there are several derivational suffixes when there seen (ii) apply sliding and (iii) apply Rule B come into play when we means that Rule A a look at are a derivational group They end with morpheme ending in a a group vowel, which applies: (16) zapat-ero, ca-sita, There have they must be assigned stress according to Rule A. The second includes minimal words with - except - morphemes (D'Introno et al. 1995:167ff). The first includes words with the derivational suffixes -on, -cion, -es, -or. consonant, so follows: (i) morphemes. complicated apparatus must words with derivational - -, goes as words diminutive/augmentative - cas-ona with ('shoe maker', Tittle house', 'big house') several ending in derivational vowel. In this a morphemes case, applied to the last derivational morpheme, placing the stress on - class and the above rules are that morpheme: (17) cancioncita, poesla, zapaterla (Tittle song', 'poetry', 'shoe shop') In the last two words in condition that (17), the process of sliding gets blocked because there is requires the application of stress rules. In the 52 case of words with a an only suffix such observation, -acia, -ancia, -encia, -ario, -oria, sliding applies. As a last as words with some lexically marked, in the sense an added derivational suffix - -ito, -itico, -cito of case are that the suffix gets the stress. The exception to this is the suffix -tico, which does not admit stress, so it moves back to the suffix. In the - -ologo, stress is lexically marked the syllable before on the suffix, too (D'Introno et al. 1995:168-169). Once the rules for words with derivational suffixes have been D'Introno et al. which apparently override Rule A. He takes hermano, cartera/-o derivational closed (1995:168) find a good solution for words like hermano, suffix, which stress is applied, over and fourth adjectives as follows: above, calvicie to have though it is applied on a the primary stress assignment for Spanish general rule, stress as a can never be assigned to a syllable from the end of the word; put differently, the three-syllable rule inherited from Latin is still at work. if that word contains any those seen syllable. We could summarise the rules of nouns even or explored, Primary stress is assigned to the minimal word; derivational morphemes, stress is assigned to the last morphemes. Last, sliding is inapplicable if the minimal word has suffix which contains As for only one a one of derivational vowel. secondary and tertiary stress, D'Introno et al. (1995:172) establish a series of rules, summarised as follows: a) Primary stress is assigned, as in em-pe-ra-do-res ('emperors') b) Secondary stress is assigned to the first unstressed syllable in the word in case there is one -, as - in em-pe-ra-do-res. c) Stress is assigned to other syllables in accordance with the rules of secondary accent: tertiary stress next to primary stress, secondary stress next to tertiary stress, tertiary stress next to secondary stress, etc, as in em- pe-ra-do-res. d) The sequence stress + secondary stress + secondary stress changes into secondary tertiary stress, as in em-pe-ra-do-res. 53 In compounds and adverbs ending in -mente, stress gets assigned to the first lexical element, then to the second one; the stress of the first element is reduced, 1995:173).19 becoming secondary (D'Introno et al. (1995:412-414) implementation of Navarro Tomas's model D'Introno et a/.'s has the added value of rendering a brief, yet comprehensive, account of stress assignment rules in verbs, organised with reference to the different tenses in Spanish. First, in the present tense of regular verbs, both in the indicative and in the subjunctive, stress falls on the penult (except for the second person in the plural) independently of the type of syllable and of the preceding syllable. As for second person of the plural, stress falls on the last syllable, which means that Rule B is applied, then sliding in the first and second conjugations, and elision/synaeresis in the third one, where the thematic vowel is in the same syllable as the morpheme of person. Second, in the imperfect (indicative and subjunctive), the thematic vowel is stressed (note that it gets elided in the second and third conjugations). Third, the past tense carries stress remains on the on the thematic vowel, which is then elided, while the accent original syllable. The exception is the third lexically marked on -re marked with the stress on the and -ri, respectively. In non-finite forms Rule B is applied on the are participle and gerund, while the infinitive is lexically marked morphemes 19 of the singular, the morpheme -6, especially in the second and third conjugations. Last, the future and conditional morphemes person -ar, -er, on the last syllable - -ir. (1999:225) postulates that secondary stress is rhythmic in Spanish. Secondary the assignment of primary stress, as it does in other languages. It is, on the contrary, a manifestation of the metrical rhythm in the phrase. So, while primary stress is assigned according to lexical rules applied at the level of the phonological word, secondary stress responds exclusively to postlexical rules, applied at the level of the phonological phrase. Nunez-Cerdeno stress does not stem from 54 Non-finite forms Finite forms Subjunctive Indicative and Imperfect Present Past Stress Infinitive Gerund Part. Future/ Marked Rule B Rule B Conditional Suffix Penult Stress (except thematic thematic stress 2nd vowel vowel + morphemes elision -re, on plural) Marked on Table 13: Rules of stress Syllable-compression 2.2.3. on -ri assignment for Spanish verbal forms. phenomena in Spanish: and synaeresis synalepha As mentioned in sections 1.4.3.4. and English show 2.1., the so-called 'stress-timed languages' like tripartite connection between stress, vowel length and quality, and a syllable weight (Dauer 1983). In general terms, heavy syllables, that is, syllables rhyme is formed by whose to attract stress segments) are than means reduced or ones, a a consonant, are more likely "prevocalic segments in the syllable (i.e. onset are, in turn, and unstressed vowels usually undergo compression. an unstressed neutralised quality; this is to ones are position tends to be made shorter, say, it tends to become a schwa or, in III in stress accent languages (Giegerich 1992:287). Vowel reduction in direct vowel plus prosodically inert" (Hayes 1995:51). Stressed vowels centralised. A vowel in cases, - or a that, while stressed vowels have full articulation, unstressed laxer, and with some long vowel light syllables longer than unstressed This a English is dependent on word stress. This has, in turn, implications for the general rhythm of the language: in English speech, only the stressed syllables in that unstressed an utterance syllables are tend to be evenly spaced in time. This forced to hurry into short time-limits, a means fact which is responsible for the reduction and/or loss of quality of English unstressed vowels (MacPherson 1975:35). Syllable-timed languages like Spanish do not show the connection stated for stress-timed vowels - languages. The vowel system in Spanish, with just five length-neutral that is, neither phonologically short 55 nor long (see section 2.2.1.2.) - is responsible for the fact that weight depends different variable, which is, in this the open or checked status of syllables: in Spanish, a syllable is heavy case, exclusively when it is checked by state on a that vowels a consonant. According to this, it is impossible to heavy syllables attract stress in Spanish. On the other hand, given that are neither non-stressable long nor short, unstressed syllables do not necessarily count as syllables. syllable-timed rhythm of Spanish does not explain why this language The requires that stresses fall at specific points in the utterance; in other words, stresses at Spanish the level of the utterance in completely free, but rather dependent are not rhythm of utterances, which require the overall on certain number of rhythmical a 'leaning-points' at certain intervals. This entails that, at points where unstressed syllables which are very numerous, there will be devices. The former are As observed in the basis of a synaeresis and synalepha; the latter into play. There are consonants in Spanish. Vowels case a hiatus wi, eu, a iu, vowel can wo, be realised an -i or word, as as glides - ja, je, jo, ju] in spelling, realised a -u -, which leads to a case, aj, ej, oj, ui] -, an well as in different [u], [i] (also called a a -u as an realised a may onglide increase in sonority the outcome will be which entails combination; under these circumstances combinations a diphthong may arise. vowel is followed by an -i or a ou, note that there are no syllabic - be adjacent in or a length of the combination; in such Second, a Spanish) and [j], [w] (also called 'semiconsonants'). A vowel preceded by we, either may phonemes /i/ and IwJ 'semivowels' in [wa, dieresis and hiatus. five-way minimal contrast (see Tables 9 and 10). Vowels constitute words, in which The are 2.2.1.2., five vowel phonemes can be identified in Spanish on syllabic nuclei, and all the nuclei contain [au, come compression devices in Spanish, which have two corresponding 'decompression' two be need to increase the speed of discourse, of 'compression' must that certain processes means a over - the rising diphthong. as an offglide - decrease in sonority throughout the falling diphthong will arise. The [wu, iu, ji] are very rare in Spanish, while [uu] is completely unattested. Any other combination will normally make the two segments involved belong to different syllables, in an instance of hiatus. 56 the glides In sum, in a can be seen as the forms taken by /i/ diphthong with another vowel. Each glide immediately after the vowel with which it forms a diphthong, as seen ['bojna] boina 'beret' /Ml ['fwego] fuego 'fire' [eu'ropa] Europa 'Europe' are combined to form peak of prominence, which is the most most normal situation involves other vowel - high, medial level, the /u/ and any addition of a or low sequences that sonorous can occupy - can syllable. Every diphthong has one of the vowels one form a a vowel. As observed above, the being a high vowel, so that the the syllabic nucleus. Therefore, at diphthong those which involve are other vowel. Under certain morphological circumstances, like the plural morpheme, the two parts of the diphthong syllables different two ones, are not single syllable, but rather two independent clear the lexical Spanish glides. important to note that Spanish diphthongs, unlike English a or in Table 14. ['tjera] tierra 'land' phonemes which between immediately before III single phonemes which correspond to or can occur After vowel It is I'll /u/ when they occur Before vowel Table 14: The behaviour of vocalic or - ley [lei] vs. can be redistributed leyes ['lejes]; Taw', Taws' (MacPherson 1975:52-54). As I have would just mentioned, when two vowels which naturally form hiatus. This is Nevertheless, diphthong normally the even where hiatus is a when a preferred morphological explanation blo are case - in succession and pronounced in separate syllables, they when none are are some exceptional cases diphthong, something which does not have for example, in ('devil'), Ru-an-da (Nunez-Cerdeno in of the vowels involved is high. high vowel is involved, there over appear cases any like cli-en-te ('customer'), di-a- 1999:183). According to Canellada (1987:51-53), hiatus has seventy-five theoretical combinative possibilities in Spanish, twenty-five for each combination - two unstressed syllables, unstressed plus stressed syllable, and stressed plus unstressed syllable -, as observed in actual speech. Table 15 summarises the possible combinations of Spanish vowels. 57 /-v.v-/ < (from a more closed to > (from a more open a more open segment): /i/ and /u/ + /e/, /o/, /a/ to a more closed segment): /e/, /o/, /a/ + <:/e/, /o/ + /a/ > : = /a/ + /e/, /o/ (segments of equal openness): /i/, /u/ + /i/, /u/ = : = : Id, /o/ + /e/, /o/ /a/ + /a/ /-V.V'-/ <:/i/,/u/ + /'e/, /'o/,/'a/ >:/'e/, /'o/, fal + n/, /'u/ < : /e/, /o/ +/'a/ > : /a/ + = = : /'e/, /'o/ /i/, /u/, /'i/, /'u/ :/e/, /o/ + /'e/, /'o/ =: /a/ + /'a/ /-V'.V-/ < : /'i/, /'u/ + /e/, /o/, /a/ >:/'e/,/lo/, /'a/ + /i/,/u/ < : /'e/, /'o/ + /a/ /'a/ + /e/, /o/ = : /'i/, /'u/ + /i/, /u/ = : /'e/, /'o/ = :/'a/ + /e/, /o/ + /a/ Table 15: Hiatus in 58 Spanish. /i/, /u/ It must be observed that, in popular speech and in rapid educated speech, always preserved, hiatus is not so separate syllables of the same word or Id and lol assertion that to - - typically, two medial vowels, like /a/ and /o/, compressed to form are Spanish diphthongs belong in the that vowels which would otherwise belong in same a are not single syllable, a fact that reinforces the independent entities, but two vowels made syllable. This entails that, at the postlexical level, we have to accept the existence of both high and medial glides. If two adjacent vowels within the same word are made to correspond to a single syllable, the resulting phenomenon is called 'synaeresis'. The most frequent instances of synaeresis occur when two unstressed alike vowels come on other, something which gives rise to the fusion of those two vowels into same after the one of the quality: (18) neerlandes -> ner-lan-des ('Dutch') In contrast, it is not very frequent when one of the adjacent vowels bears a stress: (19) leer -> ler ('read') Synaeresis between unlike vowels is particularly rising sonority cases - /ia, ie, io, where hiatus is ua, ue, There are uo/. Nevertheless, there in sequences are some of exceptional preferred: (20) riamos -> ri-a-mos ('let du-e-to common us laugh'); diente -> di-en-te ('tooth'); due to -> ('duet') several possible causes for this unexpected choice of the marked option (Hualde 2002:217-218). One of them is that the hiatus is preserved because the root word also has hiatus: (21) riada -> ri-a-da ('flood') < ri-o ('river') f rio 59 Another explanation is that there can be perceived word boundary between the two a vowels, and the hiatus is preserved to signal it: (22) bienio -> bi-e-nio (meaning 'two years', from Latin). Last, the phonological context diphthongisation. This is the initial sequence can when case cause word starts with a for hiatus the preference a trill, or over when the word- is #(C)ia, #(C)io. The least common into contact to form a of the possibilities arises when two unlike vowels single syllable, particularly if one come of the two is lexically stressed: (23) teoria -> teo-ri-a ('theory') This fact forces us to observe a lexical contrast between sequences in hiatus and diphthongs, where the position of stress determines the realisation of one or the other (Hualde 2002:217): (24) pie -> pi-e ('I cheeped') The realisation of as the vs. synaeresis depends, pie -> pje ('foot') on the position of the word in the rhythm completely unemphatic, synaeresis is realisation of groups of vowels hand, group - upon linguistic factors, such when the word in question is likely to be found. On the other hand, the very as one hiatus or synaeresis has a sociolinguistic component, a fact that makes it difficult to reduce these features to a set of rules: uneducated conversation speakers make are also speech, synaeresis is a likely to a larger use use of synaeresis; educated speakers in fast synaeresis (MacPherson 1975:37). Beyond normal regular and completely accepted feature of Spanish verse, as will be observed in section 3.2.3. Quilis (1993:370) explains synaeresis in terms of the position of the tongue in the articulation of the vowels involved vowels (Table 16). The contact between two non-high gives rise to two different syllables; the contact between 60 a low or medial vowel, and a high vowel, contact is between different a or vice versa, non-closed, unstressed vowel and a a single syllable; when the closed, stressed vowel, two syllables arise. Resulting syllables: examples Articulation of the vowels Two gives rise to non-high vowels -> Two syllables a-e-re-o aereo ('aerial'), pe-le-ar pelear ('to fight'), le-a lea ('read!') Low/medial vowels + high vowel ai-re aire ('air'), eu-ro-pa (or vice versa) -> One syllable a-sia Asia, bue-no bueno Non-closed, unstressed vowel + a-bi-a habia closed, pais ('country'), stress vowel Europa, ('good') ('there was'), pa-is re-u-no reuno ('I gather'), ba-ul haul ('trunk') Table 16: D'Introno et al. "[w]hen there are two Synaeresis in Spanish. (1995:209) equate synaeresis with elision and state that, equal vowels next to each other, there is the unstressed one". There are four main elision rules 61 a tendency to elide according to this model: Formula Rule Vowel elision: elide degree of aperture and [a aperture, b point] V-> 0 / vowel if it is by another vowel with the followed same a [a aperture, b point] V point same of articulation. Elision S -» 0 / C, D ofsyllabic boundary, elide $ vowel, between a consonant and between a displaced vowel and a vowel, and vice versa. Sliding: turn displaced followed vowel into a [Medial] a Lax] Y #; # X come into contact in a rhythm single syllable: V#V -> (V) (V). This is known when the final vowel vowels of the or vowels of one word is syllable will fall that as or more vowels which belong group are combined to form 'synalepha', and often a occurs joined together with the initial vowel following word. A curious phenomenon is observed when synalepha brings unlike vowels together to form one [V, R] 'V Y # (D'Introno et al. 1995:223-224). of process takes place when two different words that in that _ [Vowel, word. A similar type or [High] / #X 'V stressed vowel within Table 17: Rules of elision to [highl or [+high, +lax] vowel if it is in contact with same [+medial, +lax] Thigh] V; V by another vowel. Vowel rising: turn a the V -> D / if it is preceded V D or a vowel [+high] into a one a V; upon the a single syllable. In that more or most case, the main stress perceptible of the vowels, not on the belongs to the most lexically significant word. (25) non-tonic pronoun se + verb unen This entails that the stressed vowel in synalepha may be a a -> seunen group ('they unite') of vowels brought together in vowel which is not normally stressed when the word to which it belongs is pronounced in isolation (MacPherson 1975:57). 62 The maximum of vowels joined (26) Subio onto the The most a together under synalepha is five: Eulalia al arbol -> su-bidaeu-la-liaal-dr-bol ('She lifted Eulalia tree') synalepha common sometimes it extends (27) Vencid over a vowels. It normally covers two words, but three: un enemy'); rompio groups two enemigo -> ven-cioaun-e-ne-mi-eo ('he defeated his en rom-pioen-a-plau-sos ('he burst out aplausos clapping') There are as for hiatus many combinatorial possibilities for synalepha as we have seen there are (see Table 15). Generally speaking, there are three conditions for synalepha to take place, namely that the tempo of speech must be fast, the register must be colloquial and, when more than two unlike vowels come together, they single syllable provided that the syllable contains fall in prominence, or a are 1975:56). In other words, when there are a closed one be combined to form continuous rise or a spread three between two open ones; (28) estoy aqui -> es-toi/a-kl ('I am over two or more a continuous peak of prominence. If a combination of vowels contains trough of prominence, the vowels not be a may a syllables (MacPherson vowels involved, there must if this happens, hiatus will take place: here') According to D'Introno et al. (1995:224-225), the processes of vowel elision, syllable boundary elision and sliding (see Table 17) also apply to synalepha. Elision is more frequent when unstressable one of the vowels belongs in a clitic, that is, an unstressed and monosyllable, like determiners, prepositions, non-tonic pronouns, relatives, conjunctions, etc. It is also possible to elide the vowel in a clitic followed by a different vowel, something that does not 63 occur in synaeresis. Any [+high] vowel in contact with of a different vowel a can [+medial] vowel does not happen Summarising what of four types we have explored slide and be across so syllabified with the latter, Sliding words. far, the union of two or more vowels may be in Spanish (Canellada 1987:54): /V.V-/-> Hiatus [-V.V-] Synalepha20 Reduction [-VV-] Diphthongisation [-SV-]; [-VS-] Elision [-V-] Table 18: As can Compression and decompression devices. be observed in Table 18, synalepha can be divided into three further subtypes, namely reduction, diphthongisation and elision. Diphthongisation is the prototypical type of synalepha, and occurs when such a vowel comes together with that the latter becomes glide. This is why this way referred to a 'gliding', represented as as a high vowel, in process is sometimes /VV/ -> [Vi], [Vu], [jV], [wV], (29) mi amigo -> mja-mi-go ('my friend') In the two of reduction, medial vowels 'placed' loses case a on - there is which becomes a vowel - typically, the first in a sequence of non-syllabic postlexically, that is, it is not the nucleus of the resulting syllable. Although it preserves its quality, it part of its quantity and intensity, and it becomes a lax vowel. The vowel resulting from this union is perceptually longer than in elision: /VV/ -> [(V)V], (30) 20 se Studies such acabo s(e)a-ka-bd ('it is over') Cabre and Prieto (2003:1687-1690) only use the term 'synalepha' in the following single metrical position - in written verse - or a single note of music, as if melismata did not exist - a melisma is a passage of several notes sung to one syllable of text. In this view, reduction, diphthongisation and elision would be three related phonological phenomena which could sometimes give rise to poetic and/or musical synalepha. sense: two as vowels in a V#V sequence occupy a 64 frequent type of synalepha and, Elision is the least occurs when which comes This is to adopt the onset of the previous vowel and indicates, it name its preserve particularly normal when the two vowels joined together /e/ + /e/:/VlV2/-» rhyme. own the are one, same - e.g. [VI]; [V2], (31) se acabo sa-ka-bo The realisation of four the completely deleted in favour of the following vocalic sound is a as synalepha as diphthongisation, reduction elision follows or general rules (Canellada 1987:54-55; MacPherson 1975:55-56), namely: a) If high vowels /i/ when one in the syllable, b) The or /u/ appear, of the vowels is /u/ as an more open onglide there or or vowel prevails may be diphthongisation. That is to III, it is pronounced, according to its position offglide. over the more the non-tonic closed one. c) The tonic vowel prevails over d) The element which in the second position prevails goes say, one. over the one in first position. These rules can, in turn, be reduced to two main ones and one exception which regulate diphthongisation in Spanish (Piera, personal communication), namely: a) The first type of diphthongisation is lexical, and it involves the glides [j], [w], U], [y]. b) The second type is postlexical and involves vowels. In this case, the must c) As any sequence of two adjacent only inviolable principle is that, if /a/ is involved, it always be the syllabic nucleus of the resulting syllable. an exception, there realised as are a limited number of cases of medial vocoids diphthongs already at the lexical level, namely: c.l) -AO: Bilbao, bacalao ('codfish'), participles in -ao cacao ('cocoa'), Curagao, and the (<-ado) and their nominalisations - e.g. pescao empleao ('employee'), pesao ('heavy') c.2) -OA: coalicion ('coalition') (vs. ko-a-la), coartada ('alibi') c.3) -EA: real ('royal') (vs. re-al, 'real') c.4) -AE: trae ('bring that here') 65 ('fish'), Although the phonological dichotomy long vs. short for vowels does not exist in Spanish, it seems obvious that, whenever two vowels are joined together under synalepha - particularly under diphthongisation and reduction -, the outcome will be a longer vowel. Actually, the length of a vowel made long under synalepha to double the undo length of synalepha proves simple vowel. Nevertheless, the fact that a can we can be up always that vowel length does not have phonemic value (Canellada 1987:54): -V.V-/-»[-V:-]/[-V-]. Although primarily considered speech, synalepha is also an linguistic phenomenon attested in normal a enormously productive device in the poetry of syllable- timed languages like Spanish. As these languages is based on a general principle, syllable count in the poetry of the principles of phonetics and euphony which the speech of the individual languages. In Spanish, and Portuguese, the adjustment of the is achieved by synalepha, by elision Synalepha and its realisations in 2.3. The verse or verse are as well as govern in Provencal, Italian line to the required number of syllables by hiatus (Preminger 1974:713-714). covered in the next chapter. prosody of English This section analyses the nature and characteristics of the English prosodic system. Section 2.3.1. constitutes an examination of the emergence and evolution of the English prosodic system starting with Old English. Section 2.3.2. deals with stress assignment in English, focusing the theories developed by Halle and Vergnaud on (1987), Giegerich (1992) and Gasiorowski (1998). In sections 2.3.3. and 2.3.4. I examine the interaction between stress, structure in English, as well as the vowel length and quality and syllable consequences of this interaction for the assignment of stress in this language. 2.3.1. From Old English to Present-day English The evolution of the stress assignment system in English is a controversial, intricate, topic in English historical linguistics. This chapter amounts to of that an as well as overview evolution, which will help understand the basic principles of Present-day English (henceforth PdE) stress assignment. Following the tendency of most Germanic languages, in Old English (henceforth OE) the primary stress of native words and, according to Halle and 66 Keyser (1971:88), also borrowed ones, was placed on the first syllable of the stem. A stem is the combination of the basic form of a word - the root - plus any derivational morphemes, which excludes inflectional elements; put differently, the stem is the form of the word to which inflectional morphemes can be added. In Lass's (1994:91) words, OE stress falls "on the first syllable of a simplex word, but on the first syllable of the lexical root in morphologically complex words whose first element is a prefix". (32) OE stem stress: scip ('ship') word ('word') cyning ('king') drincan ('to drink') cirice ('church') superne ('southern') Two observations in OE. On the just as one seem in order here with regard to the assignment of primary stress hand, in compounds the first element was always the strong one, in PdE (see Lass 1994:90; Halle and Keyser 1971:89). (33) OE compounds: meodo-heall On the other ('mead-hall') hand, OE prefixation was quite complex, with a considerable number of prefix doublets, where unstressed an old Germanic prefix could appear (Lass 1994:92). (34) OE prefixes: Stressed prefix Unstressed prefix 'aef-fanca ('offence') of- 'pyncan ('displease') 'an-,gin ('beginning') on- 'ginnan ('begin') 'bi-,genga ('inhabitant') be- 'gan ('occupy') 67 both stressed and At a glance, the principle at work first while verb prefixes were seems to be that noun prefixes were unstressed. Nevertheless, Lass (1994:92) lists significant exceptions to this conclusion, as we can a stressed, number of observe in (35). (35) Exceptions to (34): Verbs with stressed Nouns with unstressed prefixes prefixes 'inn-,gangan ('enter') be- 'bod ('command') 'softer-,swyrian ('inquire for- 'gifness ('forgiveness') after') 'and-,swarian ('answer') As Lass (1994:92) explains, there 'non-genuine' prefixation, resulting from (ia) (inn, asfter) joined to adverb then becomes If initially, we we with non-initial look at the cases so or independent and, (ib) an as such, stressed initially-stressed noun which in was (36) and-giet verb with an unstressed prefix which then (for- 'gifness < for- 'gifari). (ia) above, where prefixed noun that the resulting is, the first element In stress a will observe that certain independent roots, that an initially-stressed verb ('and-,swarian < 'and-,swaru), and (ii) an a noun verb, a noun or 'genuine' prefixation, resulting from becomes of prefixation in OE, namely (i) were two types prefixes behaved nouns nouns were as stressed though they were had the stress contour of compounds - the strong one: 'sense' spite of prefix-stress not being placed in accordance to the morphological category of the corresponding word in OE (see (35)), the Middle English (henceforth ME) period group was characterised, among many other things, by the of prefixed verb-noun doublets which were emergence noun carried the stress To make matters even more on the a indeed differentiated by the placement of the stress (37). As happened in the vast majority of the OE ME the of prefix, whereas the verb carried it on cases, in the root. complicated, this applied not only to Germanic words, 68 but also to French loans, and even to cases, taken Germanic roots with Latinate prefixes - in most the prefixes were no longer transparent in ME, and the prefix plus the root was as a inherited in PdE. stem. This situation was (37) Verb-noun doublets: As Verb Noun re-sit resit im-port im-port con-vert con-vert one can observe in (33), (34), (35) and (36) above, OE showed several degrees of already mentioned, "primary stress stress, crucially primary and secondary. As obligatorily lined was with the initial boundary of the lexical root" (Gasiorowski up 1998:141). Summarising, as Halle and Keyser (1971:97) point out, OE stress assignment responded to three main rules, namely the Initial Stress Rule syllable of the stem carried the main stress adjectives - and the Compound Rule - by whereby the first the Stress Retraction Rule syllable to the prefix in retracted the main stress from that first and -, - means some - which derived nouns of which the first element of a compound carried the main stress. I will now highlight the most important characteristics of the change in stress assignment system from OE to PdE. The key difference between OE and PdE stress assignment is that OE stress rules refer to the left edge of the root stress fell on the first syllable of the stem right edge of ,the word. That is to but-one syllable in the say, Stress Rule 21 In as -, this means that while PdE stress rules work from the PdE stress is assigned starting from the last- word.21 Moreover, in OE the location of primary polysyllabic words did not depend quantity-insensitive -, - on syllable weight - stress in OE primary stress was while it does in PdE. Lass (1994:93) summarises the OE follows: PdE, like in Latin, final syllables are extrametrical unless they are the only syllable in the word, in which case they are accented by default. 69 a) Starting at the left-hand edge of the word, look for left by a a binary foot c) Add at most left- to syllable bounded on the major category label. b) Construct The a reason one s [strong] additional of the w [weak], w. change in the stress assignment system of English - from right-handed - is not clear. It could have been due to language contact; nearly all the PdE words that do not bear stress borrowed from Latin In or Greek, either directly or syllable of the root the first on were via French. spite of these differences, OE and ME share several characteristics with PdE, something which favours the deep connection between the two systems. OE, just like PdE, builds bimoraic feet, that is, rhythmic units consisting of two weight units or morae.22 evidence. First, OE shows rule that deletes the neuter nouns another this idea, In order to support after light - after ending a a we can draw on three pieces of High Vowel Deletion (HVD) (see Lass 1994:98-99), -u stressed a (/u/) of the nominative and accusative plural of strong heavy syllable or after a stressed light syllable plus bimoraic foot. (38) High Vowel Deletion: BUT SINGULAR PLURAL scip scip-u word word-0 werod werod-0 HVD also affected the vowel Id in the After a same contexts, that is, after a bimoraic foot. light syllable, /i/ would remain, but would usually lower to Id. (39) preOE *win-i > OE win-e ('friend') (Lass 1994:98-99) 22 pointed out in 1.4.3.4., unit of sound that determines syllable weight. The syllable - if it is a short vowel or two morae if it is a long vowel or a diphthong. The coda of a stressed syllable represents a mora, but it is not clear whether the coda of an unstressed syllable represents a mora or not. As onset does never - a mora is a represent any mora. The syllable nucleus represents one mora - 70 (1994:100) explains, HVD reflected As Lass foot structure; a a weight-based constraint on heavy syllable + /i, u/ would be considered 'overheavy', while a light syllable + /i, u/ would be legal. Second, like PdE, OE has stressed light monosyllables in the next are not Minimal Word Constraint which implies that tolerated in OE - this principle is applied to PdE section.23 Third, equivalent to called a explained in section 2.3.1., in OE as a verse a stressed light syllable followed by an stressed heavy syllable is unstressed syllable, which is 'resolution';24 both of those syllables together constitute a bimoraic foot. The existence of bimoraic feet in both OE and PdE points to a historical relationship between the two systems. Before the change of the stress assignment system, there was an overlap between left-hand and right-hand stress. In the native vocabulary of OE, which stemmed from Old Germanic, three syllables long. In ME, native words underwent several in such a way would fall in the were special of vowel loss, either mono- or disyllabic. What this meant was identically whether the rules worked from left to right to left. On the other flood the processes than that, by the end of the Early Modem period, most of the native Germanic words to no root was ever more or that stress from right hand, when Latin and Greek loans with right-hand stress started language, speakers were able to reanalyse the native vocabulary as a of right-hand stress. case The Latin Stress Rule (Lass 1994:88) was subsequently applied to a vast amount of Latinate lexicon: Final syllables in Latin, unless they which are case they outside the are are the only syllable in the word, in accented by default, domain of are extrametrical; that is, they accent-assignment, 'invisible' to the accentuation rule. Therefore any disyllabic word, regardless of the weight of the first syllable, will be initial-accented. In polysyllables, 23 quantity-insensitive in that, in native polysyllabic words, stress is assigned to the first syllable regardless of its weight. Nevertheless, lexical monosyllables cannot be light, which means that, to a certain extent, stress and weight are correlated. Foot structure, on the other hand, seems to be weight-sensitive, too. 24 In Lass's (1994:101) words, "a sequence of two light syllables "resolves to" or is metrically/quantitatively equivalent to, one heavy one". OE is of the stem 71 accent penult if it is heavy. Otherwise accent the antepenult, the regardless of weight. The Latin system was characteristics which were thus right-handed and quantity-sensitive, two diametrically opposed to the Germanic system. There is controversy as to when the change from a left- to a right-hand system really happened. Minkova (1997) has shown that in ME verse French loans are predominantly stressed verse. In the initial syllable on except line-finally in rhymed - fact, those loans could be stressed either initially or finally, as required by the metre: (40) citee - cite; comfort - comfort', divers — diverse', present -present During the ME period, originally end-stressed loans from Norman French became nativised and received stress on their initial rules, something which indicates that etymologies Germanic stress taken into account for As Halle and either Germanic taken to have Germanic so To the stress sum up, Germanic system, Latinate system, mercy, tempest, virtue was or be assigned according to the origins of the Latin; nevertheless, sometimes they were origin when they were Latinate, and vice versa, assigned contrary to the rule. the stress assignment system of PdE is which was a hybrid between made complex during the OE period, and which started to gain influence in ME and ended up a a in becoming the stress. can be monosyllables, where stress and weight correlate, in native disyllabic words, which preserve word native borrowed major stress assignment system for PdE vocabulary. The Germanic heritage seen always Keyser (1971:102-103) observe, in later stages stress came to - were not prosodic matters. (41) abbot, barren, fortune, honour, word syllable, in accordance with the the stress as it changes, and in prefixed The Latin heritage, on was assigned in OE, nouns even when the structure of the and verbs which contrast in the placement of the other hand, 72 can be seen in the English vocabulary with Latin origin, where syllable-weight determines the placement of French or stress, and where stress itself is We must be aware, assigned from right to left. however, that the evolution from the OE stress system to far from simple, and that there the PdE system was were numerous stages that contributed to shape the system as it is nowadays. eighteenth and nineteenth centuries stand as intermediate The sixteenth, important critical points as regard the interaction between the Initial Stress Rule and the Latin Stress Rule, on the one and the Stress Retraction 2.3.2. Stress As Rule, on the hand, other.25 assignment in Present-day English already mentioned, stress has to do with the arrangement of syllables in rhythmic structures turn, according to their relative perceptual prominence. Those structures organised hierarchically. Syllables syllable is one that heads constituents such as are foot; feet a are, in arranged into feet, where the stressed are prosodic words. Within gathered into higher-order metrical a word, primary stress falls on the syllable that heads the foot that heads the prosodic word. For instance: (42) music: WORD FOOT SYLLABLE 1 HEAD SYLLABLE 2 STRESSED UNSTRESSED mu sic PdE is adverb - - a language, that is, stress must carry a every stressed syllable; in the lexical word - noun, verb, adjective case stress, one of those stresses will be the main or subordinated to the former with while 25 an acute secondary stresses Given that the main are on one syllable carries primary stress, and the others will be (Giegerich 1992:179) graphic accent that more than or - primary stresses are signalled the nuclear vowel of the corresponding syllable, signalled with a grave accent. topic of this dissertation is not stress in English, I am forced to leave the Keyser (1971) give a thorough account of the intermediate stages in the discussion here. Halle and evolution of English stress. 73 (43) lesson, abracadabra To that there a small extent, stress in are some English is phonemic. This is proved by the fact pairs of words that are segmentally identical but distinct in terms placement. This correlates, in turn, to the morphology of each word in the of stress pair; typically, • one of them is a noun while the other is • one a verb. 26 (44) NOUN VERB import import record record A syllable must satisfy certain structural requirements in order to be eligible to carry stress in English. First, stressed syllables must be heavy while unstressed syllables may be light or heavy. (45) Weight and stress: STRESSED accordion (V: UNSTRESSED or VC) English is, therefore, a HEAVY LIGHT honest (VC) erode (V) quantity-sensitive language, that is, stress is assigned according to syllable weight; weight is, in turn, closely related to vowel length and vowel tenseness. Those features that each of the terms in the term in each of the other the 26 are arranged into binary oppositions, in such oppositions would prototypically have a a way corresponding oppositions. Table 19 shows those oppositions, as well potential correspondences between each of them: The existence of these doublets goes back to ME, as we saw 74 in the previous section (see (37)). as STRESS Stressed Unstressed SYLLABLE WEIGHT Heavy Light VOWEL LENGTH Long (Short) Short VOWEL TENSENESS Tense (Lax) Lax Table 19: Stress and its related features in There is stress, a set of (Reduced) English. phonological rules governing this relationship between syllable structure and vowel quality, to which I will return in the next sections. In English, stresses, syllables and feet characteristics and rules. As are arranged according to pointed out in the last section, these rules directly from OE, and work for PdE as were a set inherited follows: a) The (bi)Moraic Trochee: As signalled for OE, PdE feet comprise two It follows from this that that is, a two an English foot can consist of either syllable with two rhymal X-positions, syllables with one of or two a morae. heavy syllable, light syllables, that is, rhymal X-position. On the other hand, those feet are trochaic, that is, the first element in the foot is stronger than the second element, that is, the first element is the head of the foot (see (42)). b) Word Minimality: As observed for OE, PdE stressed monosyllables must be heavy. This follows from the fact that heads seen, foot, and a an a syllable English foot requires two can only be stressed if it morae, as we have already in order to be built. When trying to summarise the rules of primary stress assignment in English, we come across quite an extensive number of different theories. One of the most important treatises is Halle and Vergnaud (1987), which devotes a chapter to the exploration of English word stress (1987:227-276). This volume inherits its theoretical and authored empirical apparatus from a previous treatise about English stress, by Halle and Keyser (1971) which, in turn, draws some co- of its theoretical principles from Chomsky and Halle's (1968) The Sound Pattern of English (SPE). As Halle and was the between Vergnaud (1987:227) acknowledge, "[a] major empirical result of SPE discovery of the central role played in stress assignment by the contrast 'strong' and 'weak' clusters - that is, between syllables with branching and non-branching rhymes". As mentioned 75 on several occasions throughout this dissertation, such contrast gives rise to the main rule of stress assignment in nouns, which bears a resemblance strong resemblance to the Latin Stress Rule (see section discovered was 2.2.1.1.). This by Chomsky and Halle (1968) and set a point of departure for contemporary treatises of English stress, which left behind the view that English stress was not predictable in order to assert the exact opposite. The rule states that, for nouns, "main stress falls on penult contains light - and on a non-branching rhyme the antepenult when the that is, when the penultimate syllable is - the penult otherwise" (Halle and Vergnaud 1987:227). (46) Canada, agenda, marina, tuna, henna, aluminum, conundrum, cerebrum, possum, venom Main stress in Halle and nouns does not normally fall on the last syllable because, Vergnaud (1987:227) indicate, this syllable is extrametrical, that is, it is 'invisible' to the rules of stress assignment. However, there several are exceptions to the rule of extrametricality, the first of which is formed by whose last as syllable has a long vowel groups of some nouns its nucleus. as (47) police, brocade, bazaar, attache, kangaroo A further group secondary stress on of exceptions to the extrametricality rule the last syllable - (48) Berlin, insect, decathlon There is and a a vs. further rule at work in mostly words with difficult to be accounted for. London, tempest, subject27 English, the so-called 'Rhythm Rule' (Halle Vergnaud 1987:234-235), which retracts the original main stress of a final carried syllable with a a branching rhyme to secondary stress. If there is words to which this rule 27 proves more - Some varieties of English applies are realise subject no a from previous syllable which originally such syllable, stress is not retracted. The lexically marked. as noun subject. 76 (49) designate, anecdote, telephone, jeopardise, elongate Of the numerous studies of English stress that came after Chomsky and Halle (1968), Halle and Keyser (1971) and Halle and Vergnaud (1987), I focus on two specific theories, namely Giegerich (1992) and Gasiorowski (1998). These two theories helpful for the are purposes of this study because they do not resort to a complex system of theoretical tools in order to explain the basics of stress assignment. The rules of (1998:143) as English falls stressed primary stress in English explained by Gasiorowski follows. Following Halle and Vergnaud (1987), primary stress in on the rightmost non-final stressed syllable. The rightmost non-final syllable must be heavy in order to be assigned primary stress; if it is light, antepenult, regardless of its weight. Primary stress the stress is retraced to the goes nouns are never beyond the antepenult. (50) November, disaster, example, agenda There are, exception is a group carry stress on would nevertheless, of trisyllabic the first syllable - exceptions to the above rule. The first many nouns - that is, not of Anglo-Saxon provenance on - which the antepenult, where the Latinate rule predict penultimate stress. (51) ancestor. calendar, sinister, minister, character, adjective, orchestra, talisman, protestant, Stress on the first messenger, lavender, cylinder syllable is particularly regular on trisyllabic nouns ending in [i:] [i]. (52) Normandy, monarchy, These irregularities can energy, tapestry, galaxy be grouped under the following categories: 77 ~ • In words with to • a light initial syllable, the second syllable becomes heavy due the contact between [st] or The a homorganic consonant. 'energy' class in RP (Received Pronunciation) shows because the It/ is not second • [+son] and pronounced. However, in rhotic varieties, where the syllable is heavy, the stress falls The rest render light second, ' a on the first syllable, too. irregularly stressed derivatives. (53) ancestral, cylindrical vs. sinistral, ministrant, Protestantism • For words ending in -ic, stress falls on the penult, even when this is light. (54) periodic, melodic, phonetic, Mathematics Exceptions to the rule • are The words underlined in rhetoric, catholic, lunatic. (51) and (52) are truly irregular, so they must be lexically listed. For most of the exceptions seen above, the oddity of the stress assignment to apparently light syllables could be solved by resorting to the concept of 'ambisyllabicity', whereby unstressed syllable can a consonantal segment working also function as the onset of as an the coda of the preceding syllable if this is stressed, thus rendering the latter heavy. The second primary stress on exception is a set the first syllable; these of long words cases are - four syllables - with the truly exceptional. (55) alligator, caterpillar, helicopter, caricature, dandelion There are various subclasses of these words: a) Greek loans in -lepsy, -plexy, -doxy, b) Latin loans in -gogy, -archy, -thermy, -mony, -versy. c) Idiosyncratically stressed words: -ary, -ory, -ery. (56) inventory, repertory, desultory, dysentery 78 -mancy, -urgy. Following Gasiorowski (1998:145), the explanation for this is that the second syllable in these words is closed with [+son]; it therefore, it can can work light syllable and, as a be unstressed. The first syllable, on the other hand, is light in most cases, although, means of ambisyllabicity. as happened with trisyllabic exceptions, it exception is The third a mixed group adjectives and can be made heavy by nouns with final primary stress. (57) NOUNS: degree,police, duress, ADJECTIVES: serene, effect, boutique, balloon, Carlisle, extreme, exact, bizarre, innate, Dundee, Madrid, Berlin, grotesque engineer, kangaroo There • are several explanations for the cases listed in (57) (Gasiorowski 1998:146): They end in 'autostressed suffixes' of French origin: -ade, -aire, -eer, -oon, - esque, -esse, -ette, -ee. • Many are geographical names in which there is a tendency to keep the native pronunciation - French, Spanish, etc. Summarising what a in the of one of their stress have heavy syllable. In the assigned to case we polysyllabic nouns, seen up to now, stress case of monosyllabic in PdE nouns, they must contain at least nouns is always they must be heavy; one heavy syllable, or light syllables must become heavy through ambisyllabicity. The rules of assignment for verbs are simpler. The major generalisation is that, for two- syllable verbs, the final syllable will receive main stress only if it has a long vowel; otherwise, the initial syllable will have the main stress. (58) achieve, employ, guffaw, ignite, maintain languish, manage 79 vs. arson, finish, giggle, Giegerich (1992) summarises the rules of stress assignment in English position of the stressed syllable, in accordance to the as well as nouns the morphological category of the word in question: I. Nouns • As a bearing final stress: principle, there are no final-stressed words in English that end in syllable. On the other hand, no a English word with final stress ends in light a lax Among the disyllabic examples, secondary stress occurs on the first vowel. syllable only where that syllable is also heavy. (59) bamboo • While nouns vs. balloon with final stress adjectives with final stress cases have is are are heavy; but unlike in comparatively quite nouns, common. rare in English, verbs and The final syllable in all such heavy initial syllables do not necessarily secondary stress. (60) obscure, *dbscure • Sometimes, secondary and primary stresses invert their positions, but only when both • syllables As mentioned are are heavy. above, there are a number of noun-verb pairs in English that distinguished by their stress patterns, the corresponding verbs final stress. (61) NOUNS VERBS digest digest survey survey torment torment II. Nouns bearing non-final stress: 80 nouns having non-final and the penult is stressed if it is heavy; otherwise, stress falls The • the on antepenultimate syllable. (62) horizon, potato, with two rhymal X-positions occupied by aroma, vowels (63) enigma, agenda, synopsis, with two rhymal X-positions occupied by a vowel and (64) a consonant. camera, retraced to the is light, so Exceptions (lexically listed): badminton, calendar capital, vertebra, with one rhymal X-position, antepenult. Exceptions: vanilla ambisyllabicity occurs to or so stress is madonna, where the penult make it heavy Giegerich (1992:189) observes that stress is not assigned only in accordance phonological conditions, but also following with an interaction of phonology and morphology. More specifically, he analyses the role of suffixation in the assignment phonological terms, suffixes of stress. In never make any with a case suffix (65) difference to the stress pattern of their base and the stress pattern - of the base derivational -ic, -ity, -ian -ee, -ette, -ese, -esque, A suffix that shifts stress in that it may be stress-neutral, in which -ing, -ed; derivational -less, -hood, -ly, -able, inflectional -s, in which can some - which may and are -ness are never - or case they stressed - stress-shifting, differ radically from that of the base can bear the main stress of the word. always heavy syllables instances does not necessarily do so in all bases attach to. (66) solemnity There vs. divinity are some cases where foot-level rules, which have to do structure, interact and override word-level rules, information. As an obvious with syllabic which deal with morphological example, heavy penults cannot be stressed if the final 81 syllable carries stress. In terms of foot-level rules, a foot aligned with the penultimate syllable has to be disyllabic, also including the final syllable. It cannot do that if the final syllable is itself a foot, like in (67). (67) nightingale, cavalcade, chimpanzee Monosyllabic nouns always constitute heavy syllables; quoting the Minimal Word Constraint, (68) bit, trisyllabic; as a do nevertheless also when monosyllabic lexical word cannot be light. camp At the level of the or a as we saw foot, feet principle, occur on every are formed from right to left, and they can be bi- lexical word must have a foot. Monosyllabic feet heavy penultimate syllables, but only where that syllable happens to be the first one of the word. (69) bamboo, champagne, July, balloon Giegerich (1992:201) concludes by theorising about Foot Assignment in nouns, which works according to the following rules: a) Assign foot to the final syllable if it contains a a long vowel, or exceptionally, if it is otherwise heavy. b) Assign a disyllabic foot to the penultimate syllable if it is heavy. c) Assign a foot to the penultimate syllable if it is heavy and initial. d) Assign a maximal di- or trisyllabic foot to syllables from right to left, and ensure any remaining string of that the word has at least one foot. These rules the apply from right to left: first the final syllable is checked by mle a); then penult is checked by b); and the default rule d) also assigns feet from right to left. As seen in this section, English stress is assigned by following two series of rules, namely phonological rules and morphological rules Last, there is a group - of words which must be listed lexically. 82 in the case of suffixes. 2.3.3. Stress and related features following sections I analyse the interaction between stress and vowel length In the (2.3.3.1.), on the one hand, and between stress, quality and tenseness (2.3.3.2.), on the other. These two levels of interaction have important assignment of stress in English and its classification as a 2.3.3.1. Vowel It is length. Long vowels display vowel the stress-timed language. length and reduction acknowledged fact that, in English, there is an consequences on a correlation between stress and statistical tendency to be stressed more often a the other hand, most unstressed vowels tend to undergo than short vowels; reduction. According to Hayes (1995:15) and Gasiorowski (1998:134), stressed vowels on always full in English; whether the are demonstrated. In fact, as Burzio sufficient condition for vowel converse is true remains to be (1994:112) observes, "stress is a necessary but not reduction, and [...] there exist a class of unstressed but unreduced vowels". (70) Halicarnassus, condensation In (70), stress are stress observe that the unstressed vowels in the syllables preceding the main we full, and do not reduce. This going vowel system of RP: b. that there are other factors, apart from itself, which contribute to reduction. Before a. means short long on with this discussion, it is i e as i: ei ai iu io eo A Reduced hi is not au ou a: 3: quality that shows specified for any D have u a look at the 3 oi Table 20: Vowel system The vowel necessary to u: o: U3 of RP (Gasiorowski 1998:130). up quite commonly in vowel reduction is hi. vowel features - except for minimal opening. As Giegerich (1992:68) emphasises, schwa is, in terms of its articulation, "neither high 83 nor low, neither front nor back. It is a articulators and is in this respect a vowel produced with a neutral setting of the 'minimal' vowel, involving, as it does, no displacement of the articulators from the neutral positions". In terms of its schwa occurrence, word-finally - can occur in all positions and all kinds of syllables monosyllabic, non-lexical word, it can - - word-initially, word-medially and open or closed (71). If it appears in a be the only vowel in that word. (71) /dAtdin, inf/a/(r)mdtion, us/d/(r) In are spite of the predominance of schwa in unstressed positions, although all schwas stressless vowels, not all stressless vowels are schwa. There are some other segments that can appear in unstressed position, as we observe in the list below (Hayes 1995:13): • Never stressed: • Variable: [o], syllabic [n], syllabic [m] syllabic [r], syllabic [1] [i, ou] / [i]/ [yu] rj ~ [yo] In contrast to the segments • V, and in prefixes. #, / above, certain vocalic segments Always stressed: [ei, [i:, e, se, a, o, a, u, ou, can never u] i] when not in the contexts above. Going back to the segments in unstressed positions, where schwa vowel can, and vice schwa with any be unstressed: can occur, no other In other words, there are no minimal pairs contrasting vowel other than/i/. This is the reason why Giegerich (1992:69) versa. points to the fact that, strictly speaking, schwa should not be called a 'phoneme' of English. (72) family > fam/dAy,fam/iAy. Let us go back now to (1994:112): "There exists a the observation quoted above, made by Burzio class of unstressed but unreduced vowels; stress is 84 a necessary but not sufficient condition to vowel reduction". This observation points to the fact that the correlation between stress and vowel one predict. Indeed, there would are length is not as transparent as several factors to take into account apart form stress, and most of those other have to do with the segmental characteristics of syllables. First of all, there is a distinction between open According to this, unstressed ones open and closed syllables. syllables reduce quite generally, while closed do not. (73) OPEN CLOSED (unreduced) America product parasi:te derelict economy manifest The or (reduced) exceptions to those cases seem to be those syllables closed by sonorant /s/, which frequently permit reduction. (74) information, carpenter, orchestrate The in that position of a syllable within syllable compare I. can be reduced a word also determines whether the nucleus Below (following Burzio 1994:117) or not. we final and internal positions: Word-finally: a) If the unstressed vowel is checked by obstruents - except /s/ remains unreduced. (75) parsnip, kidnap, lilac, Sherlock, nutmeg, ketchup b) If, in turn, it is checked by sonorants and/or /s/, it gets reduced. (76) balsam, utensil, apron, syllabus 85 -, it c) There are some cases where the vowel is checked by sonorants and/or Is/, but the vowel remains unreduced. (77) chaos, meteor, wigwam II. Word-internally: a) If the unstressed vowel is checked by obstruents - except /s/ it remains unreduced. (78) autopsy, architectonic, conductivity, enigmatic, hyperactivity b) If the vowel is checked by sonorants and is/, it gets reduced: (79) contemplation, affirmation, information, usurpation c) Like in word-final syllables, sometimes a vowel checked by sonorants and Is/ remains unreduced. (80) incantation, detestation, exorcise, inculpate The distribution in both cases exhibits what Burzio refers to as 'Arab-rule' effects (1994:119). That is, both in word-final and in word-internal position, "if the syllable bearing stress is light, then the unstressed syllable is affected by reduction, regardless of the quality of the postvocalic consonant, thus precisely carn[a;]p" (Burzio 1994:119). Let us have happens. a) Word-finally: (81) barrack, gossip, havoc b) Word-internally: 86 a look of some as in ar[o]b, versus examples where this (82) recognisable, resignation, adaptation There are, nevertheless, (83) ad[ee]ptation vs. cases that violate this generalisation: att[e]station, macr[v]n Finally, it is worth pointing that certain segments in clusters to license reduction than others. Ross are more likely (1972, in Burzio 1994:120) noted that coronal stops /t, dI are unlike other stops in not inhibiting vowel reduction,28 (84) myriad, invalid, period, lilliput where unstressed vowels are reduced, contrasts with unreduced unstressed vowels in: (85) ndm[ae]d, nlmr[v]d there is a difference between coronal stops and other stops which, aside from Arab-rule effects, generally block reduction. Final coronal stops also fail In any case, to inhibit reduction when sonorant or s as they occur as its first. Words such the second member of a cluster that has a as (86) elephant, element, catalyst, comfort with reduced unstressed vowels, contrast with (87) aardvark, asterisk, abelmosk (tropical plant), podunk (small, isolated town) (Burzio 1994:121) On the other hand, stops always inhibit reduction when combined with /t, d/ (Burzio 1994:122). 28 is a consonant articulated by closing the airflow in the vocal tract. A coronal consonant is by being articulated with the blade of the tongue. A stop defined 87 (88) cataract, insect, product, edict, concept, district All the of cases non stress-dependent reduction discussed here summarised in the rules that follow can be (Burzio 1994:124-125): 1) In the structure VCX, where X is not a vowel: a) Reduction of V is generally inhibited, because C requires vocalic support not - provided by X. (89) adirondfsejck, adj[e]ctival b) Reduction of V is permitted if either (i) i) or (ii): C has high intrinsic sonority, being either a sonorant or /si. (90) serfsfndipity, aprfaJn ii) C requires a low sonority downstep - being /t, d/. (91) connecticfdft c) Reduction of V is forced - plus or minus idiosyncrasies - by 'constant transition', when the preceding stressed syllable is light - Arab rule. (92) hammfdJck, reefsJgnition 2) In the structure VC1C2X, where X is not a vowel: a) If CI is relatively sonority-transparent, being maximally licensed if it requires a a sonorant or s, then C2 is low sonority downstep (/tf). (93) elephant and only minimally licensed if it requires 88 a high sonority downstep (/p, k/). (94) podunk. Minimal licensing - as with clusters Cp, Ck - results in both of (i), (ii): i) Relative infrequency. (95) podunk ii) Non-reduction of V - whose sonority is critical. (96) pod[a]nk b) If CI is relatively sonority-opaque, being licensed only if it either requires a a stop, then C2 is maximally low sonority downstep /t/. (97) abrup[t] or if it is intrinsically sonorous - sonorant, /s/. (98) participfl] Clusters stop-p, stop-k are thus excluded. The conclusion of this brief overview factors is that stress and vowel mediated the by a are vowel intimately related, but this relation is one being the structure of syllable where the unstressed vowel lies. seen in the quality and tenseness previous section, not only does stress interact with vowel length, but these two also correlate with vowel quality or tenseness. (1971:4) words, "[t]he location of primary stress in the distribution of tense vowels". In other words have a length and its determining series of other variables, the most important 2.3.3.2. Vowel As length on tense or a a tense vowel lax vowel as as 89 word is closely correlated with words, stressed syllables in polysyllabic their nucleus, while nucleus. a In Halle and Keyser's unstressed syllables can have either Table 21 reference to a. lax b. tense inventory of English vowels and diphthongs with shows the quality: i e as i: ei ai iu 19 e9 Table 21: The A 9 u oi au ou a: 3: ense D u: o: and lax vowels in U9 English. correspondence between the parameters of length and quality is striking. All short vowels are lax, while long vowels are always tense. As Giegerich (1992:101) observes, quality differences normally have a quantity correlate in the vowels of English. (99) /i'J is [+tense, +long] and III is [-tense, -long] In RP and GA (General American), [+tense] gives rise to [+long], while [-tense] gives rise to [-long], which If it is true that means that the feature [long] is redundant. long vowels have to be stressed and that the corresponding syllables will be heavy, then one could say that there is a triple correlation between syllable structure, vowel length, and vowel quality. SYLLABLE STRUCTURE VOWEL LENGTH VOWEL Heavy Long Tense Light Short Lax QUALITY Table 22: Correlation between parameters. As was indeed the with numerous. dialects of position case length, this correlation is not For instance, as so clear-cut, and exceptions Halle and Keyser (1971:13) observe, are some English have the characteristic feature that most vowels in word-final appear tense rather than lax: 90 (100) pit[i](not [i]), val/ju] {not [ju]) This means that "phonetically tense vowels subject to reduction are not totally without stress" (Halle and Keyser 1971:29). In (100), question lose vowels in a we even when observe that the degree of length, but their quality remains the same, so they become 'neutralised'. As which a sort of summary, we can relate stress, Giegerich (1992:142) establishes a rule by means of length, and quality or tenseness in English. Vowel-Length Rule (for RP and GA): a) Associate a [-tense] vowel with which such vowel consonant. Being can be one mora. light syllable a it is not eligible to so, This - carry means that the syllable in if it is not followed by a the stress of the corresponding word. b) Associate each element of occupies two morae, diphthong with a one mora. So, a diphthong and therefore, the syllable is heavy and eligible for stress. c) Associate a [+tense] vowel with two The corresponding syllable is, morae. again, heavy. It is eligible for stress. 2.3.4. Syllable structure and syllable weight Syllable weight depends primarily that on the length and quality of the syllabic nucleus, is, the vowel. In turn, stress is assigned according to syllable weight. Hyman (1985:5) notes the importance of the syllable weight concept in determining the placement of stress in different languages. As consists explored in the previous sections, English treats solely of a short/lax vowel long/tense vowel, treated or a as -VC a diphthong or a heavy. Schematically, rhyme defines a a as light, whereas short vowel plus -V rhyme defines a a syllable whose rhyme syllable whose rhyme has one or more a a final consonants is light syllable, and either a -V: heavy syllable. Giegerich (1992:146) explains syllable weight in terms of rhymal X-positions. According to this, stressed syllables have minimally two X-positions - that is, they have a rhyme with a are heavy -, while unstressed syllables single X-position - that is, they may be heavy or light. 91 may (101) alliance: /d *lai V- -VV light heavy 9ns/ = V:- -VCC = VC heavy English, like Latin, assigns stress in such always be stressed, while a way that a heavy penult will light penult will be skipped to assign antepenult stress a - regardless of the weight of the antepenult. (102) allocation vs. merriment It follows from what must have, at least, there are no one we have seen up to now that lexical polysyllabic words heavy syllable, while lexical monosyllables must be heavy; English lexical words of the form CV, where V is Function words, on the other hand, may (103) sea (/si:/), bit (/bit/) It must be observed vs. be light - with a a short vowel. heavy variant. the (/da/ heavy variant /di/)\ *bi (/bi/) that, in disyllabic words, the afore-mentioned Arab rule operates: if the first syllable is light, the second must have a full vowel as its nucleus. Summarising, syllable weight is "necessarily tied to the existence of a vowel length - or vowel tenseness - opposition" (Hyman 1985:6). As we have seen, heavy syllables characteristically attract stress, whereas light syllables receive stress only in the absence of an 2.4. eligible heavy syllable. Summary Chapter 2 focuses on the analysis of the prosodic characteristics of Spanish and English speech prosody, in order to find out whether there differences between the are any rhythmic two. Section 2.1. summarises previous research work carried out in relation to the study of prosodic typologies, paying especial attention to those theories devoted to the statuses of Spanish and English. The concept 92 on which most prosodic classifications syllables occur at studies depending on of languages groups 'isochrony'. One of the first researchers to come at which elements are perceived as the are isochronic. The two main syllable-timed languages, characterised by the fact that are equal intervals in time, and stress-timed languages, where stresses equal intervals. Other linguists who used isochrony were use (1945). The nuclear idea in Pike's theory is that languages term was Pike different based is that of are Jones the basis for their as (1950), who illustrated his statements with musical transcriptions, already pointing to the connection between speech rhythm and musical rhythm and Liberman among (1979), who used musical notation for the development of the grid theory, others. In any case, which prove measurable that physical isochrony does not really exist and that, rather than phenomenon, isochrony is existence of Pointon the isochrony hypothesis has been contradicted by experiments isochrony is a a perceptual illusion. Despite the fact that the controversial issue, (1980) admit that to the ears a even sceptical scholars such of the speakers of a as language like English, Spanish sounds syllable-timed. Given the lack of empirical support for the stress-timing classification, scholars such rhythm is not languages, a one as vs. syllable-timing Dauer (1983) have developed the theory that speech dichotomy but rather a scale or continuum, with two hypothetical perfectly stress-timed and the other ends of the scale. Theorists who support one syllable-timed, marking the the scalar view (Dauer 1983; Bertinetto 1989) claim that the rhythmic differences between languages result from a variety of components which interact in such a way that a language can be given a position in the continuum. Some of the components that determine the score of a language are quality, consonant quality and quantity, syllable duration, compensatory vowel shortening, syllable structure, relationship between syllable structure and stress, nature of stress, position of stress, intonation and tone. In sections 2.2. and and 2.3., a thorough account of the speech prosody of Spanish English is given. First of all, I examine their historical evolution with regard to prosody, and then I concentrate the two on a specific theory of stress assignment for each of languages. 93 The evolution of the Spanish prosodic system stems from the Latin Stress Rule. Classical Latin showed words of three or more preference for stress to fall a syllables, while light ones on heavy syllables in qualified preferentially as non- syllables. Latin is, thus, the best example of the connection between stress, stressable syllable weight, and vowel length and quality, which could be summarised by claiming that the Latin vowel system had Classical a direct impact on its stress system. In Latin, the position of stress, which was fixed in principle on the antepenult, the phonological distinction between long and short vowels, which in depended on turn mapped onto the distinction between tense and lax vowels and, therefore, on were the heaviness of syllables. As soon as vowel length stopped being phonemic and exclusively phonetic, stress itself became phonemic, became as it is in contemporary Spanish. In can Spanish, the relationship between syllable structure and stress placement be summarised into three main Spanish words end in a vowel is consonant on is the a characteristics, namely that the vast majority of vowel, that the unmarked stress position for words that end in penult and that the unmarked stress position for words that end in a the ultima. on According to Harris (1983), the rules of Spanish stress placement could be summarised If follows. Nonverbal items as they end in can end either in vowel, stress is prototypically placed a consonant, stress falls on the last possible, and indeed on a vowel as a consonant. numerous, stress will invariably fall on one as Spanish are an a are of the last three idea reinforced by prototypically quantity- (in Lipsky 1997:563) and by Trubetzkoy's idea that stress sensitivity to syllable quantity well a syllable. Although exceptions to these rules Hayes's theory that trochaic systems such as in the penult; if they end in syllables of the word. Spanish stress is not quantity-sensitive, insensitive or presupposes a tense/lax contrast In D'Introno et al. Rule A from - systematic vowel length contrast (in Roca 1997:621), (Hyman 1977:48). (1995), stress is assigned according to two basic rules: Closed Syllable Stress Rule: The first closed syllable (counting right to left) of the minimal word must be stressed. Rule B - Default Rule: Stress the second syllable. 94 primary stress assignment for Spanish We could summarise the rules of and nouns fourth adjectives follows: as as a if that word contains any be assigned to derivational morphemes, stress is assigned to the last only one dependent on a one of derivational vowel. Stresses at the level of the utterance in rather a Primary stress is assigned to the minimal word; morphemes. Last, sliding is inapplicable if the minimal word has suffix which contains of can never syllable from the end of the word; put differently, the three-syllable rule inherited from Latin is still at work. those general rule, stress Spanish are not completely free, but the overall rhythm of utterances, which require a certain number rhythmical 'leaning-points' at certain intervals. This entails that, at points where unstressed syllables discourse, which must come into are very numerous, means that, in play. There there will be need to increase the speed of another, certain a way or are two a processes of 'compression' compression devices in Spanish, synaeresis and synalepha, which have two corresponding 'decompression' devices, dieresis and hiatus. In English, following the tendency of most Germanic languages, the primary stress of native words in OE observations: in PdE and was placed on the first syllable of the stem, with two (i) in compounds the first element (ii) OE prefixation prefix doublets, where an was was always the strong quite complex, with old Germanic prefix could a one, just like considerable number of appear both stressed and unstressed. The key difference between OE and PdE stress assignment is that OE stress rules refer to the left of the stem say, - edge of the root - this means while PdE stress rules work from the that stress fell on the first syllable right edge of the word - that is to PdE stress is assigned starting from the last-but-one syllable in the word. Moreover, in OE the location of primary stress in polysyllabic words did not depend on syllable weight - OE primary stress is quantity-insensitive -, while it does in PdE. The left- to reason of the change in the stress assignment system of English - from right-handed - is not clear. It could have been due to language contact; nearly all the PdE words that do not bear stress borrowed from Latin or Greek, either directly 95 on or the first syllable of the root via French. were OE and ME share several characteristics with the PdE, something which favours deep connection between the two systems. OE, just like PdE, builds bimoraic feet, that is, rhythmic units consisting of two weight units or morae. Second, like PdE, OE has monosyllables verse a an a are not tolerated in OE. Third, as explained in section 2.3.1., in OE a stressed light syllable followed by heavy syllable is equivalent to stressed unstressed Minimal Word Constraint which implies that stressed light light syllable, which is called 'resolution'. During the ME period, originally end-stressed loans from Norman French became nativised and received stress their initial on (Lass 1994:88) Germanic stress rules. The Latin Stress Rule applied to a vast amount To sum Latinate system, subsequently was of Latinate lexicon. the stress assignment system of PdE is up, Germanic system, syllable, in accordance with the which was a hybrid between made complex during the OE period, and which started to gain influence in ME and ended up a a borrowed becoming the major stress assignment system for PdE vocabulary. The Germanic heritage in seen stress. can be monosyllables, where stress and weight correlate, in native disyllabic words, which preserve word native the stress as it changes, and in prefixed The Latin with Latin heritage, French or on a assigned in OE, nouns even when the structure of the and verbs which contrast in the placement of the other hand, can be seen in the English vocabulary origin, where syllable-weight determines the placement of stress, and where stress itself is To was assigned from right to left. small extent, stress in English is phonemic. This is proved by the fact that there are some of stress placement. On the other hand, stressed syllables must be heavy while unstressed pairs of words that syllables Gasiorowski stress in PdE monosyllabic may be light or nouns, is segmentally identical but distinct in terms heavy. (1998) articulates nouns are a theory which always assigned to a they must be heavy; in the must contain at least one heavy syllable, or one can be summarised saying that heavy syllable. In the case of polysyllabic nouns, of they of their light syllables must become heavy through ambisyllabicity. The rules of stress assignment for verbs The case are simpler. major generalisation is that, for two-syllable verbs, the final syllable will receive 96 main stress only if it has a long vowel; otherwise, the initial syllable will have the main stress. Giegerich (1992) summarises the rules of stress assignment in English nouns as the morphological category of the word in question. He also analyses the role of suffixation in the in accordance with the position of the stressed syllable, assignment of stress. In phonological terms, suffixes case they never differ never stressed make any can well be stress-neutral, in which difference to the stress pattern of their base, and they stress-shifting, in which or as the stress pattern of the base case radically from that of the base word with a suffix and can are may bear the main stress of the word. According to Hayes (1995, in Gasiorowski 1998:134), full vowels English; whether the stressed in as Burzio converse (1994:112) observes, "stress is vowel reduction, and [...] Burzio goes on to there exist as well as how the whether the nucleus in that The always is true remains to be demonstrated. In fact, a necessary but not sufficient condition for class of unstressed but unreduced vowels". explain how, in spite of the predominance of schwa in unstressed positions, and although all schwas schwa a are are stressless vowels, not all stressless vowels position of syllable can a syllable within a are word also determines be reduced or not. correspondence between the parameters of length and quality is striking. All short vowels are lax, while long vowels length, this correlation is not so are always tense. As clear-cut, and exceptions Syllable weight depends primarily on are was indeed the case with numerous. the length and quality of the syllabic nucleus, that is, the vowel. In turn, stress is assigned according to syllable weight. Hyman (1985:5) notes the importance of the syllable weight concept in determining the placement of stress in different languages. English, like Latin, assigns stress in such a way that a heavy penult will always be stressed, while skipped to assign antepenult stress - a light penult will be regardless of the weight of the antepenult. 97 3. Verse In the prosody introduction, I outlined the view that rhythm is manifested on multiple levels three of which are speech, verse and music those levels affected the other two. In - - and that the characteristics of each of chapter 2, I explored the differences between English and Spanish at the level of speech rhythm. Chapter 3 analyses the verse prosody of Spanish and English, emphasising those elements which hold direct a relationship with speech prosody. Section 3.1. examines the relationship between speech prosody and poetic rhythm and metre before introducing characteristics of verse some general prosody in English and Spanish in connection to the typological classification examined in chapter 2. In sections 3.2. and 3.3. I analyse the verse prosody of Spanish and English, exploring their historical evolution and specific characteristics. Section 2.4. is devoted to summarising the chapter. 3.1. Introduction: poetic rhythm and metre: their relation to speech prosody. Analysis of verse prosody in English and Spanish In section 1.2. I metre and pointed out that rhythm is constituted by two hierarchies, namely grouping. I also mentioned that, for using the word 'metre' in its broadest sense, reasons of terminology, I would be with the meaning of 'any kind of patterning', either consciously used and delimited by rules, unconsciously used and loosely regulated, as in the case in the as case of verse, or of normal speech. The two types of patterning are intimately related. In fact, as already mentioned, many theorists have verse may suggested that the origins of some of the metrical patterns used in be closely linked with the spoken language deriving not so much from abstract principles as on which they are based, from the prosodic patterns of actual phrases uttered in normal speech. As Abercrombie (1967:98) observes, "[t]he rhythm of everyday speech is the foundation of further and consider the idea that the two-fold, that is, not only are verse, in most languages". We could go relationship between language and metre is metrical patterns ultimately founded on phenomena of ordinary speech, but also the manifestation of such patterns is in terms of speech. This close relationship between commented upon by numerous verse and normal spoken language has been writers, particularly in recent 98 years. In T.S. Eliot's (1942:17) words, "[t]he music of poetry must be music latent in the a common speech of its time". The difference between language is passes into poetry, it is ordinary speech and no is that, when spoken verse longer used for simple communication; rather, it employed to create aesthetic effects. The aesthetic effects that lie at the poetic endeavour are never of intended to be apprehended straightforwardly by the observer's mind. On the contrary, poetic form - more specifically, poetic rhythm challenges people's minds with elements that interfere with the regular natural core course - of speech: The more prosaic a discourse, the more it loses its singing stress pattern, and is simply articulated dryly. Poetry has the opposite tendency. Thus, it must create its it is a discourse with its drawn out of perceive a own own end reality and put into sequence to [...] In this an regular subdivision of time way the listener will be imaginary time sequence; sequences, a measure discourse itself. Hence this marvellous deepest expression, when it is used make clear that he will inherent in the phenomenon, the fact that in its as a game, language spontaneously loses its arbitrary character, which otherwise rules it firmly, and it follows a law apparently alien to its contents. This law is now measure, cadence, rhythm (A. W. Schlegel 1963:103-104, in Todorov 1988:17- 18). In order for poetry to achieve those intended aesthetic effects, its rhythms are put on exhibition, skilfully organised (Malof 1970:1). In sum, poetic language may be considered As known as as a stylisation of ordinary spoken language. pointed out in chapter 1, the discipline that deals with poetic rhythm is 'verse prosody'. Verse prosody is also called 'metrics', turn, used in two senses, of versification At this or which is, in the study of the art (Halle and Keyser 1971:140; Duffell 1999:8). point it is In the field of verse one meaning either the art of versification, a term necessary to make an important terminological clarification. prosody, the word 'metre' is used with explained in the preceding chapters. In this second, 'metre' has two related meanings. On the 99 one a narrower sense narrower sense, than the the term hand, it refers to the basic pattern contained in defined a particular piece of verse, notion linked to the idea that matching between the result of the as a an verse is often abstract metrical pattern and language (Nespor and Vogel 1986:295). Modern metrics usually refers to this pattern the as 'template' of piece. Both traditional and modern metrists a term 'tension' for the phenomenon of variation from 1973:110), pointing to the fact that, the more a tension use template in a pattern the technical verse shows, the (Allen more complex it is (Halle and Keyser 1971:142). It is important to note that the abstract pattern of any poetic metre, that is, the template, can be violated to a certain extent before the reader perceives the abstract pattern the verse. a breakdown in the system; in fact, the tension between and the actual stress of the words provides variety and interest in The second meaning of 'metre' makes reference to the template plus a specified amount of tension. Jakobson (1960:365) introduced the term which best fits this definition of metre, called 'verse namely 'verse design', which differs from the lines of verse, instances', as well as from realisations of those, called 'delivery instances'. In accordance with the verse prosody, we can say narrow sense in which metre is used in the field of that this discipline is primarily concerned with the identification of the different metres and forms of at least from those that a are an us a set of rules that verse texts concentrate on added amount of tension. three types, are depending on - metrical from not, and determine when verse can be called so. Thus, a grammar actually attested in Let The backbone of metrics generative point of view - is to distinguish lines that metrics would be be verse. of generated only structural alternatives that could and precluded those that are not (Duffell 1999:7). the notion of 'metre' which refers to the template plus According to this notion, the metre of a line can be of which unit is taken to be the periodic element. The first type is 'syllabic metre', where the number of syllables in the line is the most important factor in order to consider it well-formed rhythmically. This metrical type corresponds to the versification of the so-called 'syllable-timed languages' in speech prosody. As explained in section 2.1., in these languages syllables so their distinct identity is never are isochronous, lost; equal timing of syllables naturally induces 100 syllable-counting in which, in turn, becomes its verse, 'accentual metre', which is timing.29 The second normally used in the versification of stress-timed languages. These languages present variable syllable duration. As versification does not syllables in a rely syllable-counting, but rather on type is on a consequence, the number of accented line and the contrast between strong and weak stress as forming feet - a definition of 'poetic foot' is given in the next paragraphs. Verse metrists distinguish third type, called 'accentual-syllabic metre'. As happened with speech rhythmic a typologies, accentual-syllabic metre is said to be syllabic metres, and and unstressed attention to the pays syllables in a more or a combination of accentual and less regular alternation of stressed line. This kind of metre would be used, although not necessarily, with the so-called 'mixed-type languages' in speech rhythm typologies, is, those which that neither syllable, nor stress-timed, which would hypothetically middle position in the timing scale introduced in section occupy a As on are was the case with English and Spanish as language typologies in reference to speech, I shall focus hypothetical models of accentual-syllabic metre and syllabic metre, respectively. I will concentrate these 2.1,30 on the methods of scanning poetry in languages, in order to analyse the differences between the verse prosody of both, and then determine whether there is any correspondence between the rhythmic typologies of speech and those of poetry.31 This analysis will also whether Spanish can language, in which be considered case syllable-timed language a serve to or a determine mixed-type it would show features both of syllable- and stress-timed languages. Historically, accentual metres verse is a clear example of accentual characteristics. As Halle and verse 29 had a were amongst verse, the most common but it also shows in Europe. OE some peculiar Keyser (1971:147) explain, "the Old English alliterative specified number of vowels with primary stress, but [...] required, in highlighted in chapter 2, it is a widely accepted view that perfect isochrony can never be production; in other words, isochrony is a perceptual phenomenon. 30 The correspondence between speech timing and metre is not always perfectly univocal. In stresstimed languages, the timing of speech can correspond either to an accentual metre or to an accentualsyllabic metre. In the case of English, while its speech rhythm became more and more stress-timed, its metre evolved from accentual in OE poetry - to accentual-syllabic - Chaucer, influenced by Italian and French poetry, was the first to use accentual-syllabic metre, in the fourteenth century (Adams 1997:5). 31 To scan a line or passage is to determine - by noting the pattern of stressed and unstressed syllables its normative, or typical, metrical pattern and deviations from that pattern. As was reached in - - 101 addition, that the consonant clusters preceding certain stressed vowels in the line alliterate".32 Apart from alliteration, another characteristic of OE line was composed of two-half lines, separated by English became Chaucer on, turned more stress-timed The classical tradition the more a pause or (see Bertinetto 1989). of the 'foot' as the fundamental a accented or as a measure one or more originals (Attridge are not synonyms in English, called attention to the fact that factors, namely the presence Abercrombie next some scholars, like Abercrombie (1967), have three sense that some syllables ■y c (1973:11 [ 1964], in Couper-Kuhlen 1986:55) defines 'foot' up to, as but not including, the stress-pulse". This definition points to the fact that, in addition to being by the principles of stress-timing and isochrony, English rhythm is characterised by the principle of rhythmic alternation. One of the most poem, they are important characteristics of feet is that, within the assumed to be isochronous; nevertheless, as speech, this assumption has been showed to be, rather than the reason for this lack of isochrony is that the lines of every which in turn consist of words 32 on of stress, the phonetic structure of the syllable, and the in time from the incidence of one stress-pulse conditioned The or stress and syllable quantity in English is dependent perceived to be longer than others. "the space unaccented 1982:6-7).34 Although position of word boundaries. English is 'quantitative' in the are of rhythm syllables, matching - however uncomfortably, but not equating, the 'long' and 'short' of the classical length stressed syllable and rhythmical unit, and measurable, patterned unit of poetic rhythm, the English foot is customarily defined unstressed 'caesura'. From accentual-syllabic, while the speech gradually consequently, the minimal unit of scansion.33 Being one that each approach to English metre inherits from the Greco-Latin notion consisting of verse was The native OE line was an same happened in normal norm, poem the exception. consist of feet, and, within them, syllables, which in their turn contain accentual line which consisted of four beats with a caesura, bound together by alliteration (Adams 1997:37). 33 Some metrists do not agree with the conception of the foot as the basic unit of scansion in English (see Giegerich 1985; Hogg and McCully 1987). This topic would deserve a deep discussion on its forced to leave for future research. that, in English, a strong syllable must be stressed, but a non-strong syllable may be stressed or unstressed. It is necessary to emphasise that long and/or heavy syllables are not always perceived as (primary) stressed, which leads to huge fuzziness in all the commonly-accepted definitions. own, 34 which I am We must take into account 102 phonemes. As explained in chapter 1, measurements have shown that deviations from strict foot isochrony depend precisely that deviations from syllable isochrony on are the number of syllables in a foot, and related to the number of segments in a syllable. However, and this is what makes the above assumption work, I emphasised that, in perception, such deviations often tend to be disregarded and, as a result, spoken like ordinary speech, tends to be heard verse, as more rhythmical than it really is. The names classical tradition. 'dactyl' (/ x of the four basic types They are of English feet called 'iamb' (x /), 'trochee' (/ x), 'anapaest' (x x) (Hamer 1966:9; Malof within the line, but never as bases; they x f) and 1970:29).36 Apart from those four basic rhythmic variations metrical feet, there are two further ones which can serve as (Adams inherited from the are called spondee (/ /) and pyrrhic (x x) are 1997:10-11)37. According to English 'classical' verse prosody, metrical patterns according to the predominant type of foot and the number of feet line of 'iambic pentameter' consists of five iambic feet. Similarly, 'trochaic tetrameter', iambic pentameter per are line. Thus, we may 'anapaestic trimeter', 'iambic hexameter' and named a speak of so on. The has been, since the time of Chaucer, the favourite of English poets. Its pattern allows for a great deal of freedom, although it also presents some important constraints (Halle and Keyser 1971:171). Halle and Keyser consider it the only basic type of foot in English verse, while the other feet are taken to be allowed deviations from the iambic pattern. The stanza, in turn, is a sequence rhyme and metre, which makes To mention some rhymed lines the up a are a 'couplet', and if they couplet'. Four lines rhymed abba 36 by a or a definite pattern of structural unit repeated throughout the work. of the most characteristic stanzas in pair is closed, that is, it does not followed of lines arranged in are English poetry, two successive in iambic pentameter and the carry over abab make into the following line, up a sense a of 'heroic 'quatrain'; three quatrains couplet, all in iambic pentameter, make up the conventional highlight the fact that in Greek poetry, these names corresponded to quantitative patterns. prosody has adopted Greek names, but they do no longer refer to quantity; rather, they refer to stress patterns, which makes them somewhat misleading (see Hamer 1966:10). 37 Malof (1970:37) argues that these variant feet, also called 'cadences', are "perhaps unnecessary", because they tend to obscure, rather than illuminate, metrical patterns. I must English verse 103 'Shakespearean sonnet'. A 'Petrarchan sonnet' comprises rhymed abba and for a sextet instance, cdcdcd with various arrangements of additional rhyme sounds The best way cdecde. For instance, the 'ballad stanza' contains four lines, of illustrating all the concepts explained up to here is presenting written in contemporary English, and then discussing its poem a metrical characteristics. For this purpose, - - trimeter.38 rhymed in iambic Song of two quatrains being unrhymed in iambic tetrameter and the second and fourth the first and third the scansion of or an octave (104) marshals a scansion of Rossetti's adapted from Adams (1997:21). This analysis is aimed at serving illustration for the concepts general principles from explained above; obviously, it is impossible to draw single a as an case study: (104) dead, When I am Sing sad no Plant thou Nor my dearest, oo, oo, for me; songs no roses at my oo, oo, oo head, 00, OO, 00, 00 shady cypress tree: 00, 00, oo Be the green grass above With showers and dewdrops wet; And if thou I shall not me oo, oo, wilt, remember, And if thou wilt, see forget. the oo, oo, oo(o) oo(o) rain; 00, 00, 00 nightingale 00, 00, 00, 00 Sing on, as And dreaming through the twilight if in pain: nor 00, 00, 00 set, oo, oo, oo(o) 00, 00, 00 oo, oo, haply may forget. An alternative beats. oo, oo 00, 00, 00 Haply I may remember, 38 oo(o), oo, oo, I shall not hear the And oo(o) shadows, I shall not feel the That doth not rise oo(o) oo(o) OO, 00, 00 analysis of the ballad stanza is to regard lines two and four 104 as instances of silent . in (104) is composed using The poem ballad stanza, constituted by followed with As we can unstressed quatrain of variable metre - a - a. - an in the first stanza, Both stanzas conclude, observe, there are some lines (1, 5, 7, 9, 13, 15) which contain an added syllable.39 The first quatrains in each stanza are a good illustration of the of feet used in English poetry. The basic pattern of the template, corresponds to the iambic foot (oo, oo) combined in feet, thus yielding an iambic trimeter; nevertheless, there (oo), and spondaic feet (oo), which work trochaic extension of the refrain, playing with the antithesis between 'remember' and 'forget'. different types the a song-like stanza which is quatrain of iambic trimeter, rhyming a sort of a by a as are also poem, groups some that is, of three instances of variations of the basic iambic design. Furthermore, while the first, second, and third lines of the first quatrains in both stanzas trimeters, the third are one is a tetrameter of the words in the language - see, for instance, the stress line 3, or 'if in lines 7, 8. This confers a certain well taking the foot in such and a way a cases. as variety and interest. In a sense any case, 'when' in line 1, 'at' in what seems to poem analysis of this is metrically arranged case in metrics for the There are Spanish study, we can verse reasons for prosody, most modern reference works in 39 is syllabic. over stress a verse a has a tendency in modern Spanish for timing and that vowel reduction does not Spanish (Duffell 1999:23-24). In spite of all that, Duffell (1999:45) that, from 40 verse believing that the essential unit of scansion in most perfectly regular syllable count, that there is occur cases. verse.40 is the syllable and not the foot, namely that much Spanish syllable timing to predominate main stress conclude that the stress-timed general reader state categorically that Spanish three main verse towards Spanish a in each foot is two in most of the rhythm of PdE is the basis of the structure of contemporary English now be clear is that, of isochrony is perceived. Each foot contains non-stress, and the number of syllables From the on degree of tension and complexity to the basic metrical unit of scansion, the as that Turning Apart from this, of the stresses within the lines does not always respond to the stress the arrangement the poem, as in both diachronic point of view, Spanish has an argues accentually-based metrics that phenomenon is also known as 'hypercatalexis' (Malof 1970:41). Some late twentieth century poets, like Marianne Moore, W.H. Auden, This Plath, carried out experiments using syllabic not work for verse English poetry (Adams 1997:64). Dylan Thomas, or Sylvia in English. Nevertheless, in general terms it does 105 syllabism due to the influence exerted by French literature and literary turned into theory verse the poets of other Romance languages. According to Duffell (2000:113), on rigidly syllabic only in the middle of the fifteenth century, when a became Spanish poet, Marques de Santillana (1398-1458), first articulated the skilled poets, writing for an who brought to perfection in Spanish the two types of syllabic the first poet verse for which he had French models: short-line metres by a syllables fixed - caesura. that educated public, had to count syllables. Santillana is said to be than nine norm and composite metres As - a consequence, over - those with lines shorter those with lines of ten the last six hundred or more, years divided it has become firmly established that Spanish versification is syllabic, because "generations of and audiences have observed the aesthetic code urged and cultivated poets exemplified by the medieval poet" (Duffell 2000:125). So French syllabism has prevailed in Spanish for social reasons.41 Traditionally, in order to scan a verse written in Spanish, three factors must considered, namely the final word of the line, synaeresis and synalepha, and be hiatus and dieresis. In what follows I shall explain each of those factors. The first factor has to do with the final word of the verse, is verso agudo ('oxytone line'), ('proparoxytone line').42 which is accentuated syllable counts 41 I am aware as on verso which determines whether the llano ('paroxytone line') An oxytone line ends with its final syllable. Spanish if it consisted of two syllables, as an oxytone verse or verso verse esdrujulo word, that is, one prosody prescribes that this happens in (105). that this assertion deserves further research and discussion. Nevertheless, for the purpose of this dissertation, suffice it to say that contemporary Spanish verse is syllable-timed. 42 As explained in section 2.2.2., depending on the syllable that carries stress, Spanish words can be oxytone, paroxytone or proparoxytone. This means that the stress falls on the last syllable, on the last syllable, respectively. With the sole exception of the adverbs (the equivalent to the English suffix -ly), Spanish words can have only one primarily stressed syllable. but one or on the last but two or three finished in -mente 106 (105) Caminante, 'Walker the nada tracks (7+1 syllables) mas. path and nothing else caminante, walker se your are el camino, y (8 syllables) tus huellas son no hay camino, not there is (8 syllables) path (7+1 syllables) hace camino al andar it is done path while walking' (Antonio Machado) A paroxytone a paroxytone penultimate (106). In this word, that is, which is the most accentuated on syllables neither added nor subtracted from the 'llano' are its line ends with case, one which is common one, verse. (106) No te 'Not I love you y de quererte a and from y living but because I love no quererte you to not de esperarte cuando and from mi pasa (11 syllables) quiero sino porque te quiero (11 syllables) Ilego living you I get (11 syllables) no te esvero waiting for you when I not wait for corazon goes my you you (11 syllables) del frio aljuego. heart from cold to fire' (Pablo Neruda) A proparoxytone accentuated word loses on a its line ends with a proparoxytone word, that is, one which is antepenultimate syllable (107). On counting its syllables, the last syllable. This is the least frequent of the three, something which is in accordance with the fact that, in Spanish, proparoxytone words are much less frequent than paroxytones and oxytones, in that order. 107 (107) Adoro la hermosura, y en corte and in modern aesthetics beauty 'I adore las viejas rosas I cut the old mas no amo roses los nor I am a del huerto de Ronsard; (13+1 syllables) of the orchard of Ronsard (15-1 syllables) afeites de la actual cosmetica. but I not love the ni soy un ave (15-1 syllables) la moderna estetica de makeups of current cosmetics esas (13+1 syllables) del nuevo gay-trinar. bird of those of the new happy-birping' (Antonio Machado) The second factor to be taken into account when to do with the section scanning Spanish verse has phenomena called 'synalepha' and 'synaeresis'. As explained in 2.2.3., synalepha, a very normal and usual device in Spanish verse, is the union of two or different words. more contiguous vowel sounds between two, sometimes three, Synaeresis is the union of two vowels in the normally pronounced as two same word which are syllables. (108) Ella dio al desmemoriado 'She gave una the forgetful one almohadilla de olor. A cushion (8 syllables due to the signalled synalepha) of odour' (8 syllables due to the signalled synaeresis and synalephas) (Jose Marti) The third factor has to do with hiatus and synalepha and synaeresis, respectively. vowels of different words word, which divides with two dots an dieresis, which are the opposites of Hiatus is the separation of contiguous (109). Dieresis is the separation of vowels within a single existing diphthong into two syllables. It is usually signalled placed above the weaker vowel of the affected diphthong (110). 108 syllables due (14 'the one who to me sings goes to church everyday' to hiatus) (Gonzalo de Berceo) (11 syllables) (11 syllables) (11 syllables due to dieresis) (11 syllables) (Ruben Dario) Summarising, in order to a particular poem the number of syllables in a verse-form used in in Spanish, the syllables of the first line must be counted without taking into account must scan any of the above phenomena. Second, the number of syllables adjusted by considering whether the line is oxytone, paroxytone be or proparoxytone, and whether synalepha, synaeresis, hiatus or dieresis (or several of those) affect the syllable count. When the syllable-count of the 'model' line has been determined, the other lines also can be scanned to yield the same number of syllables, taking into account the above phenomena. If, after making licenses, a said to be correct syllable-count is still not 'irregular' or achieved in 'defective', depending on any use of these poetic of the lines, that line is whether the irregularity is or is not intentional.43 Spanish names contain. The most syllable lines 43 In some -, of lines are commonly used determined by the number of syllables they names are octosilabo ('octosyllabic') general, much poetry establishes lines will deviate from the a - heptasilabo ('heptasyllabic') eight syllable verses -, - seven endecasilabo normative pattern, but does not adhere to it mechanically; norm as an aesthetic 109 or rhetoric device. ('hendecasyllabic') fourteen syllable verses as versos de and alejandrino ('Alexandrine') - - Verses from two to eight syllables are classified as versos de verses. ('minor art lines'), whereas arte menor classified eleven syllable - arte mayor of verses than eight syllables are more ('major art lines'). Although the rhythm in Spanish poetry is basically syllabic, when a verse is pronounced accents reproduced here as occur at certain intervals. Let have another look at (105), us (111), in order to find out whether this can be proved: (111) Caminante. El camino. y Caminante. huellas tus son nada mas. no hay camino. Se hace camino al andar. example, the underlined syllables In the above weakly stressed lines except this for the last recurrence poem all. As or not at one, stressed, whereas the rest we can see, stresses where there is of equally-spaced stresses one extra confers a the fact that those stresses are so much mediated on only interstress syllable. Obviously, highly rhythmical character to the syllables which fall in between stressed underlying rhythm is dependent not are fall at equal distance in all the when it is read aloud. Nevertheless, in Spanish poetry, the unstressed case, are as in Spanish speech, ones are never reduced, so the stresses themselves, but rather by the same number of syllables - the on in this three. Generally speaking, all lines require syllable. Moreover, lines of five accents. For instance, syllable and at least (octosilabo) has on one more on any one; into two syllables have with syllable but the fifth an an accent on auxiliary the sixth octosyllabic line one more on any hendecasyllabic line (endecasilabo), apart from the Alexandrine (alejandrino) is hemistichs the penultimate one or more the seventh syllable and at least a on one; an the tenth syllable, various distributions 1,6; 4,6; 4,8; 4,6,8; divided rhythmic accent heptasyllabic line (heptasilabos) has an accent on syllable but the sixth obligatory accent a or more a of seven 110 a are possible: 1,4,7; 2,6; line of fourteen syllables syllables each, and each hemistich - (hemistiquio) follows the rule for heptasyllabic lines (112). For this that it is plausible to metrists argue assume the existence of reason, some fragmentary foot a Spanish, but this has not been proved properly. structure in (112) Dichoso el arbol // que es apenas sensitivo, 'happy the tree y mas and la piedra dura, //porque esa ya no siente; more pues no as which is barely sensitive the stone hard because that anymore not feels hay dolor mas grande // que el dolor de ser vivo, pain bigger not there is ni mayor pesadumbre nor one than the pain of being alive // que la vida consciente. bigger sadness than the life aware' (Ruben Dario) Summarising, the main findings point to Spanish supported by constraints timed. This is using constructed a on a lyric poem basically syllable- poetic structure, that is to fixed number of syllables in each major determinant of how verse as say, a poem is line.44 Syllable count is thus the is structured. However, it might be the case that, in the phonetic realisation of the syllables in spoken poetry, the phonology of Spanish, including metrical constraints, interacts with a constraint similar to it the an rhythm of English isochronic stress. In any case, verse 3.2. Verse 44 plausible to English and Spanish are two say that verse, a languages prosodic systems. prosody in Spanish This section Spanish seems is really different from that of Spanish conclusion which reinforces the idea that with different higher level syllable-timing gives verse an overview of the historical evolution and characteristics of the prosodic system. Section 3.2.1. examines the prosodic characteristics (1944) highlights the fact that each language has a preference for a specific number in its speech utterances. In the case of Spanish, the preference range is in between five and ten syllables, with a predominance of the utterances formed by seven or eight syllables. Taking this into account, it is not surprising that seven or eight syllable lines or hemistichs have always been predominant in the history of Spanish poetry (Sosa 1999:37). Navarro Tomas of syllables Ill of Classical Latin verse as well as the changes that it underwent parallel to the language, until the various Romance languages and prosodic evolution of the systems emerged. In sections 3.2.2., 3.2.3. and 3.2.4. I explore the ruling rhythmic principles of Spanish poetry, namely isosyllabism, syllable counting and its related devices and stress placement. 3.2.1. Introduction: from Latin In neolatin quantitative verse to Spanish syllabic verse languages like Spanish, there exists from the versification system a common metrical basis derived used in Medieval Latin, which was, in turn, inherited from the rules of versification in Classical Latin. Classical Latin used quantitative of verse, according to which lines long and short syllabic values Latin often correlated with or, more were a model of designed in accordance to patterns accurately, vocalic values. Quantity in syllable weight, and the weight of a syllable determined whether it could be stressed not: or in this way, heavy syllables were normally stressed, while light ones where prototypically unstressed. This did not mean that a heavy syllable had to be stressed necessarily, but, rather, that the syllable in question qualified as stressable when language heavy syllables which were was put into verse; on the other hand, stressed when the corresponding word was some taken in isolation, lost their stress when the poetic composition required it. Let us have a look at following poetic fragment (Horace, in Gasparov 1996:86), glossed and the translated in (3) in section 2.2.1.1.: (113) Integer vi-tae / scelerisque purus Non eget Mau-ris / iaculis Nec vene-na-tis / gravida neque ar-cu sagit-tis Fusee, pharet-ra... In Classical Latin, the rhythm marked by the stresses determined how would be read; however, the real metre longs and shorts. We could quantity plays such as a say a poem corresponded to the arranged patterns of that in authentically quantitative languages like Latin, crucial role both in speech and verse prosody, while in languages English, where quantity is phonemic and is, therefore, 112 a key element in speech prosody, quantity does not determine stress and, in fact, gets relegated in favour of the latter when it between they become very are observed as of this, conflicts artistic devices in the former^ violent. In the latter, those conflicts get relegated fairly limited number of widely accepted poetic devices. As an to versification. As a consequence prosodic stress and lyric stress while in the latter to a comes regards poetic devices proper, synalepha - more specifically, elision - was extended practice in Classical Latin vowel followed by -m, appeared before verse, a especially when word beginning with final vowel, a a vowel - or or a -h. See (114), taken from Virgil's Aeneid (in Raven 1965:27): (114) conticuer(e) intentiqu(e) omnes 3rd pi. Shut up-past, ora chin-acc. pi. all-nom.intentus-nom. pi. tenebant have-past, 3rd pi. 'They all shut up and paid attention with their hands As would happen later in Spanish, this practice colloquial verse particularly of comedy, although Virgil also practices it; it lyric poets. Elision could would become was even in common 'hypermetric elision', and happen between lines the fifteenth - on we will was see their chins' common less in the common in that this practice century in Spain. This was called be observed in the following fragment from Virgil's can Georgicas (in Raven 1965:27): (115) Si if non tanta not so much iret peace-nom. go-imp.sub. 3sg.cold-acc. and 'If there wasn't so much peace caloremqu(e) heat-acc. Inter, Between, between the cold and the heat,' 113 frigusque quies and Latin poets - applied the with the word est, (116) The in which would lose its initial device called 'aphaeresis' reverse - a after a final vowel -e or -m. (e)st ('she uses') (in Raven 1965:28) usa opposite remained present Latin poets same process hiatus, could take place when process, before another vowel. Hiatus used it, as final vowel a was common in Homer, and Virgil in his Georgicas (in Raven 1965:28). In happened with elision, hiatus was more common or an -m in comedy than in lyric some any case, as verse. (117) Glauco/ et Panopeae/ et Inoo Melicertae. All this shows that the most such as Spanish, were already The transition from a common devices used in Romance languages in Classical Latin. quantitative system to a syllabic one started around the of some major changes in the language itself, which third century, as a consequence involved the loss of the common opposition between long and short vowels. For time, the two metrical systems cohabited. On the a period of hand, in spite of the one phonological changes in the language, the system of quantitative metrics survived through the Middle Ages and the Renaissance, thanks to continued short as - composing verse based the old opposition between long light - syllables. On the other hand, Saint Augustine, started to cultivated people; those hymns Medieval Latin compose were generalised, other devices In among hymns that could be understood by the first instances of syllabic were name In - and verse non- written in languages, called 'metrical', while introduced to compensate for the loss of syllabic those, the most salient device Romance was 'rhythmical'. As the syllabic system became was rhyme (Gasparov 1996:97). quantity does not play Correspondingly, these languages have not produced verse. heavy of the Christian Church, such the old quantitative system syllabic system acquired the quantity; some poets - (Gasparov 1996:88ff). At this time, two terms appeared to differentiate the two systems: the on cultivated authors who some fact, the essential elements in Romance 114 verse any have any prosodic role. instances of quantitative come to replace the role of the one on quantitative combinations of classical metre. Those elements consist, on the hand, of a formal disposition of the number of syllables, rhyme and stanza and, ordering of the rhythmic markers. the other, of the spite of their In common roots, the attitude of Romance languages has not entirely coincided with regard to the development and benefiting from the Latin subsequent transformations. If we focus inheritance and its French system was the earliest consolidated Romance system written around the tenth century consolidated later, in the twelfth century, by means specific languages, the - the first works were something which explains that its technique -, highly elaborated throughout the became on of the cultivation of rhythm and years. was see, - The Spanish system was subsequently enriched specifically for further details, Pensom (1998). According to Gasparov (1996:119-120), the evolution of the French and Spanish systems, which run parallel to the Italian system, can be related to three stages of increasing strictness. In the first stage, which coincides with the beginnings of the Middle be Ages, there was a vacillation as to what system, accentual adopted after the loss of length contrasts. The second stage the appearance of syllabic poetry rooted in medieval Latin or was verse, syllabic, would characterised by and the use of full rhyme. In the third stage, around beginnings of the Renaissance, the syllabic system and the use of full rhyme was consolidated, and blank verse appeared. By looking at the beginnings and evolution of Romance conclude that the character of Romance connection with the given that there is seems to have evolved in intimate phonological conditions of each language. Thus, for Spanish, no length contrast in the phonological system quantity does not play verse verse verse, we can shape and limits role in the structure of modem Spanish any are any more, determined by means syllabic verse. Rather, of the number and combination of syllables in the line, the distribution of accents, and, less importantly, the placement of pauses. 3.2.2. The principle of isosyllabism in Spanish verse The two main verse are syllable counting, other. This language principles that - means govern on that the the both the composition and scansion of Spanish one hand, and rhythmic accent placement, on the virtually perfect correlation between the timing of syllable-timing in the case of Spanish 115 - and the type of verse in the corresponding language that relies or, at exclusively least, not totally law of for Spanish, - on we the counting of syllables so.45 per line - verse, poetry using syllabic verse Syllabic syllable as verse verse is not always univocal perfect correlation between language and poetry, there a that is, In spite of the existence of clear exceptions to the quoted which this is indeed the case, that is, some the would predict syllabic - syllable-timed languages are languages in are rendered into for instance, Modern Greek. is governed by rules of composition and scansion which take their rhythmic basic unit. Lines are measured according to the number of syllables which they contain, and there is up of a regular number of syllables. This essential principle is called 'isosyllabism'. While some languages, such as a strong tendency for them to be made French and Italian, adopted the isosyllabic system from the early days of their literary tradition, some others, like Spanish, show a partial reluctance to accept the isosyllabic rule totally. Poetry written in Spanish follow two clearly different paths, depending the tradition from which it seems to stems. According to the first tendency, which is clearly strong and pervading in most cases, lines are on isosyllabic. The second tendency comprises non-metrical lines. This second tendency, which separates itself from isosyllabism, includes the defined as 45 'non-isosyllabic', where lines have Spanish poetry is by a poems non-fixed number of syllables. no means rigidly syllabic. In Duffell's (1999:45) words, Some modern reference works on poetics and metrics for the general reader state simply, but inaccurately, that Romance (or Spanish) verse is syllabic [...] This sweeping generalisation is a compliment to the influence that French literature and literary theory have exerted on the poets of other Romance languages, but it ignores the Spanish accentually-based metrics that has for centuries contested French influence and syllabism. 116 (118) Isosyllabic (7 syllables/line) Non-isosyllabic (indefinite number of syllables) Va la brisa reciente Vayan los mandados por los 'Goes the breeze recent 'Go por el espacio esbelta, through the y en space slender desi singing it opens a a some to Xerica and Onda e otros spring' a some deven aiudar they us must help for which Xericay los otros Alocad, a to Alocad Almenar, to Onda and others to Almenar los de Borriana primavera. una unos a some las hojas cantando and in the leaves abre los the orders que nos luego vengan the lens for Borriana then aca,46 come here' (Luis Cernuda) If rules of we follow Baehr isosyllabism, we (1970:37) and look at the history and evolution of the will observe that Spanish versification started being governed by those rules after the Middle Ages, when they were influence of the French and Italian writers of the fourteenth century. had introduced by Those writers acquired this regularity principle from Latin medieval poetry. This meant that, from that century depending on onwards, lines were whether they had the counted same as acceptable or unacceptable number of syllables in a specific composition. Before isosyllabism generally accepted, there was was a certain degree of tolerance towards subtle formal differences, which did not affect the essential of lines. In prosody rhythm fact, when syllables started being counted literally, a conventional kind of was imposed discrepancies became on verse, common, in such a way that certain vacillations and and the linguistic aspect of the poetic works minimised. Non-isosyllabism, the Middle Ages, both in folk versification and in literary poetry, and its gradual on the other hand, was elimination from the so-called 'cultivated 46 was indeed very common during poetry' only occurred throughout the From 'Cantar del Mlo Cid', in Duffell (2002:75). The system used to compose this poem remains a mystery for metrists, who cannot decide whether it is isosyllabic or isotonic (Gasparov 1996:135). 117 centuries.47 Nevertheless, fifteenth and sixteenth traditional poetry, existence in contemporary to point towards once in the kind of cultivated as in Spanish which deliberately sought isosyllabism appeared in 1240, in the first lines of the historical us well specific literary examples, following Baehr (1970:54-57), the now at first instance of poetry still observe traces of its inspiration.48 poetry that consciously shows a popular Looking as we can again to the fact that, at the isosyllabism, there was a Libro de Alexandre. Those lines allow poem same time that cultivated poetry turned popular kind of poetry which remained isosyllabic. The poetry of Gonzalo de Berceo - non- first half of the thirteenth century - represents the first culmination of the isosyllabic Spanish poetry in the Middle Ages. The continuation of the isosyllabic tradition, started by the Libro de Alexandre and by Berceo, is linked, at least until the end of the fourteenth century, to the 'cuaderna via', and therefore, with the so-called 'mester de clerecia' which did not towards belong to the 'mester de clerecia' had isosyllabism, although in a that achieved the best instances of a 49 But even similar evolution, tending less systematic, if more natural, isosyllabism was, the poetry way. The poet together with Berceo, Pero Lopez de Ayala (fourteenth century), with Rimado de Palacio. Towards the end of the fourteenth century, per 47 the so-called 'major art line', where the number of syllables line is regular, spread all over Spain. 'major art' poems were widely composed. They were compositions of syllabic size, with lines of two hemistichs (from 4 to 6 syllables each), two stresses per hemistich and arbitrary endings. It is said that 'major art' constituted a step back to a less strict syllabism, while strict syllabism kept pushing into the whole of the Spanish versification system. 'Major art' disappeared in the sixteenth century, and was replaced by a new type of long verse borrowed from Italian poetry (Gasparov 1996:138-139). 48 Most contemporary poetry shows non-isosyllabism - and lack of rhyme - as a stylistic marker. 49 The first forms of cultivated Spanish poetry written in the Romance language appeared towards the middle of the thirteenth century. Those works were written by clergymen and followed Medieval Latin models, with some French influence. Their topics were very varied and usually of a serious character religious, historical and novelesque. This corpus of literary works was to survive throughout the following century. The name 'mester de clerecia' - 'mester' means 'profession' and comes from Latin 'ministerium' refers to the group of clergymen that wrote those works, as well as In the fifteenth century, variable - - to the works themselves. The most famous writers of the 'mester de clerecia' were Gonzalo de Berceo Ruiz, Arcipreste de Hita. The most characteristic stanza of the 'mester de clerecia' was the 'cuaderna via', written in Alexandrine verses of fourteen syllables - seven plus seven - rhyming AAAA. The 'mester de clerecia' was opposed to the 'mester de juglaria', a popular kind of literature and Juan appeared in the twelfth century. The works of the 'mester de juglaria' were characterised by the secularity of its topics, their oral quality - it was never written down - and by the fact that they were anonymous - their authors and performers were called 'juglares', which means 'minstrels'. which 118 time, Spanish lines were classified into 'simple lines' At this moment in (versos simples) and 'complex lines' (versos compuestos). The former were, in turn, divided into two two to to subgroups, namely 'minor art simple lines', which contained from, eight syllables, and 'major art simple lines', which comprised lines from nine eleven syllables. Lines with 'complex'. This classification has according to Quilis (1984:54-55), than eleven syllables more a classified were as correlate in the phonetic system of Spanish: an average phonic group in Spanish is formed by eight to eleven syllables. century, Juan de Mena and Marques de In the second half of the fifteenth dodecasyllabic line. Marques de Santillana, follower of Santillana made popular Dante's, states theory of isosyllabism, this time under the Italian influence (Duffell a a 1999:53). Short lines started losing their original irregularity. Only certain irregular taste until the sixteenth century. In that Boscan and Garcilaso de la Vega implanted a a romances same century, kept a the poets Juan series of lines and stanzas of Italian precedence, something which meant the final victory of isosyllabism in the Spanish cultivated poetry As (Duffell 1999:37). Gasparov (1996:140) explains, from the sixteenth century to the of the nineteenth century, in there were two very end overwhelmingly dominant metrical patterns Spanish versification, namely the hendecasyllabic line combining syllabic and accentual elements - - used in cultivated poetry, and the octosyllabic line - used in folk poetry and song, with a purely syllabic rhythm. In sum, the development of the syllabic system in Romance languages stemmed from the natural evolution of the languages, as well as from their poetic traditions. 3.2.3. Syllable counting and its related devices Syllabic measurement in contemporary Spanish has given that achieving the same series of metrical devices put make lines - lot to do with isosyllabism and, number for each line is sometimes hard, there is a into practice to regularise syllable counts and thus isosyllabic. Also, Spanish poetry shows specific points in the line, depending an a on a tendency for stresses to fall at the number of syllables; according to this, in octosyllabic line, there will always be a compulsory stress Spanish poetry always counts the last syllable 119 as on extrametrical the seventh syllable - and then a series optional stresses of making lines isosyllabic that contribute to on their the first, third and fifth syllables. The on serve the same metrical devices of forcing stresses to fall purpose corresponding syllables, thus rendering the line metrical. Two of those devices, synaeresis and dieresis, happen at the word level, while the other two, synalepha and hiatus, happen at and synalepha follow comes to the natural a any level beyond the word. Given that synaeresis natural tendency of the Spanish language, the closer poetry pronunciation of Spanish, the more synaeresis and dieresis are likely to happen. There the certain mles which are regulate the use of each of the four devices at poetic level. According to Baehr (1970:42), in cultivated poetry form the Renaissance period, the following criteria are applied to the use of synaeresis: a) Synaeresis happens naturally in proparoxytones where the two vowels involved follow the stressed syllable. (119) heroe ('hero') b) It is less frequent if one of the vowels /a, (120) creemos, aereo e, o/ carries a full tonic stress. ('we believe') Synaeresis is less frequent when the vowel Id comes before the tonic vowel. (121) leon, real ('lion', 'royal') c) When in the group of vowels the second one is a tonic /i/ or /u/, poetic synaeresis is considered forced. (122) raiz, Whereas in sauco ('root', 'elder') synaeresis follows opposition with the common a natural tendency of spoken Spanish, dieresis is tendency of spoken language. In poetry, it has two 120 main functions, namely the general function of achieving the desired number of syllables in a line which is defective, or the specifically poetic function of more highlighting the artistic nature of the line where dieresis regulate the use of dieresis are appears. that norms the following (Baehr 1970:42-43): a) It is not normally used in those diphthongs which inseparable, namely /'ie/ and /'ue/, which /e/ and /o/, The are considered from vulgar Latin open post-tonic syllable, then dieresis is used more come respectively. (123) tierra, puente ('soil', 'bridge') If the diphthong is in a often. (124) cambio, injuria ('change', 'insult') b) There are some diphthongs that namely /ui/, /'ais/, /'eis/ - can be easily dismantled with dieresis, the last two in the second person plural of the present tense. In addition to take synaeresis and dieresis, there are other metrical devices that place within the level of the word (see Baehr 1970:43-44). The first do with the reduction of the aphaeresis - syllables in a - suppression of the first syllable of a word, suppression of a middle syllable, (126) redor for alrededor ('around') and apocope - has to word by means of its suppression; these (125) ciano for anciano ('elderly person') syncope group suppression of the last syllable of a word. (127) hi for hijo ('son') 121 are The second group in a word; these are has to do with the augmentation in the number of syllable prosthesis - addition of a syllable at the beginning of a word, (128) arradio for radio ('radio') epenthesis - addition of a syllable in the middle of a word, (129) coronista for cronista ('columnist') and paragoge - addition of a syllable, typically constituted by the vowel /e/, at the end of a word. (130) cantare for cantar ('to sing') These six devices, used artistically in the composition of poetry, belong to the level of colloquial or even uncultivated use The two devices which take hiatus. These two follow a of language in the study of linguistic registers. place between different words series of rules which first half of the fifteenth century. were are introduced in synalepha and Spanish after the Synalepha is the most important device used to help isosyllabism, and has proved to be idiosyncratic of the Spanish language; it is also called 'Syllable Merger' (Roca 1986:350). In spite of its extended important to note again that synalepha is Syllable Merger is vowels not a a common a string of adjacent carrying main stress" (Roca 1986:350). According to Roca (1986:355), occurs in a syntactic environment. Piera (1980:210) reinforces the idea that, in Spanish, synalepha is phonological a in poetry, it is phenomenon in Spanish speech. "resyllabification rule which coalesces Syllable Merger becomes synalepha when it becomes use process: "[sjynalepha is metrical rule The rules that a a side-effect of a phonological rule in Spanish, which then (property of the line)". regulate poetic synalepha are the following (Piera 1980:10): a) Adjacent vowels belonging to different words correspond to metrically relevant position. 122 a single b) All the remaining metrical positions correspond the line, up to c) 1, 2 can one to one to and including the rightmost stressed be overruled in specific cases the vowels in one. by hiatus, both word-internally and between words. Synalepha is can be considered the normal form in the oldest literary works, which supported by the fact that its poetic manifestations coincide with the realisation of the Spanish language. As a result, synalepha must have been used since the beginning of the most popular forms of Spanish poetry (Baehr 1970:59). One that synalepha, at least between unaccented vowels, would be suppose phenomenon in the thirteenth century, and that it fourteenth century synalepha appears in Berceo of synalepha as a As elision but was not a common As for cultivated poetry, thirteenth century. The disappearance of the 'mester - contributed to an even more salient natural device in Spanish. The introduction of Italian forms in the fifteenth and sixteenth centuries because this device is can definitely developed from the onwards (Navarro Tomas 1956:82). de clerecia' at the end of the fourteenth century use was very only supported the extended use of synalepha, utterly common in Italian, too (Baehr 1970:59-60). pointed out in sections 2.3.2. and 3.2.3., synalepha is not the rather, the latter is same as subtype of the former. In Latin poetry elision a practised. In Spanish poetry, like in Greek poetry, the use of such a proper device is optional (Balaguer 1954:14). Elision in Spanish only takes place when two alike vowels into contact, come that is, when the same vowel appears in to contiguous syllables in different words. (131) va a America ('she Punctuation marks and goes to America') caesurae do not (Baehr 1970:47). In Old Spanish, synalepha del), but this is not the general was case anymore, help synalepha from happening reflected orthographically {de + el = although cases where synalepha has spelling counterpart can be observed in the transcription of songs They are in popular poetry. way in which into contact, (ii) the placement of stresses and (iii) syntactical reasons. The appearance vowels get or a a of synalepha and hiatus is determined by (i) the proportion of 1 to 2 thirds in favour of synalepha 123 - in contemporary poetry there are 96-99 instances of synalepha for 1-4 cases of hiatus (Baehr 1970:48- 49): a) Synalepha always taken into account carries a occurs between two identical vowels. Stresses only in the case are that the second of the two vowels main stress; in such a case hiatus is preferred. b) Synalepha is compulsory when two non-tonic vowels get into contact. Conjunctions y and o ('and', 'or') considered to be vowels, and so If a tonic vowel gets follows into contact with - and their variants and u - are not they require the application of hiatus. non-tonic a e synalepha is regulated one, as (Baehr 1970:49): a) It is compulsory when the tonic vowel precedes the non-tonic vowel. (132) echo el cimiento ('he laid the foundation') b) It normally happens when the non-tonic vowel precedes the tonic (133) Subido he vor tupena ('I have climbed c) It is generally applied when two tonic vowels Synalepha is incompatible with the final compound lines. This general principle has called come in a sorrow') together. line or with caesurae only exception, constituted by the in so- estrofas de pie quebrado ('broken-foot stanzas'), which make the final vowel of the last word lines an pause up your one. on a line belong to the next line (Balbin 1962:79). The following (in Balbin 1962:79) constitute an example of this type of stanza and the corresponding realisation of synalepha: 124 (134) Quien se inclina 'Who himself inclines ja la muy fina to the very dulze flor fine de clavellina sweet flower of carnation' [...] (Fernan Perez de Guzman) (1970:80-81) explains that the configuration of synalepha between Balbin different lines presents the following characteristics from the fifteenth century: a) It arises between one is line and the immediately following line. The second usually shorter and its length is in proportion to that of the first one; the second line is b) It is a very a (verso de pie quebrado). 'broken-foot line' usual that, between the lines which enter synalepha, there is symmetry in the location of lexical stresses. c) The word affected by synalepha in the second line is exceptions - an 'empty word', that is, As clitic or a with few prepositional prefix. previously mentioned, hiatus is for synalepha what dieresis is for synaeresis, that is, the the a - reverse process. application of hiatus as Baehr (1970:49-51) formulates the rules of follows: a) Vowels HI and /u/, conjunctions y and which follow in between b) When the last stress in vowels, a are o - and their equivalents e and u -, considered consonants. line is preceded by a vowel, hiatus is usually applied. c) In a phrase, when two words hold the second one starts with grammatical word, there is numerous syllable is vowels a a a close morphological relationship, if stressed syllable and the first one is a tendency towards hiatus. This tendency has exceptions, namely (i) if the vowel that precedes the stressed an e, even at happen to occur as the end of a poetic line, (ii) when two alike explained above. 125 d) The application of hiatus and synalepha is not systematic when two stressed vowels norm, come into contact. In these cases, hiatus is although it only happens consistently when the stress of the second vowel is more obvious due to the Summarising, in Spanish poetry Q form and a sequence the sequence As applying - have rhythmic accentuation of the line. and also in PQ in which P ends in PQ tends to be associated with we usually the seen in this section, a a song -, when two syllables P vowel and Q begins with a vowel, single metrical position. isosyllabism in poetry is achieved by series of different devices to poetic composition and scansion. Those a devices have undergone a specific process of evolution throughout the history of Spanish, and currently follow certain rules. Of all them, synaeresis and synalepha respond to a natural tendency of the Spanish language, that of making two adjacent vowels belong in the same syllable; this is or more of the most salient one idiosyncrasies of Spanish rhythm and metre. 3.2.4. The displacement of stress in Spanish poetry The construction of based not only on rhythm in Spanish poetry, be it isosyllabic the counting of syllables, but also prosodic stresses. This relies on on or non-isosyllabic, is the periodic distribution of three fundamental principles (Balbin 1954:122-123): a) The centre of rhythmic movement is always the stanza. b) The distribution of prosodic stresses follows the principle of alternating stressed and unstressed syllables. c) Together with the main stanzaic rhythm, there is also a secondary rhythmic system, held at the level of the line. The previous statements point to the fact that the beauty of the line and the stanza depends to a great extent on often coincides with the is the placement of the rhythmic accents. Their placement position of lexical stress in individual words. At this point it important to take into account that stress is totally objective and determined by the rules of stress certain assignment of specific languages. In addition to the stressing of syllable in each word, the lines in each kind of poetic composition number of rhythmic accents accents must fact, there on specific syllables. This does not coincide with lexical stresses in order for the line are two mean to be a carry a that rhythmic acceptable. In specific devices that have to do with the deliberate shifting of the 126 position of stress within word; these a are systole - the stress goes back to the previous syllable, (135) impio for impio ('pitiless') and diastole - (136) the stress goes oceano On top forward to the next syllable in word. for oceano ('ocean') (Baehr 1970:44-45). of the number of lexical stresses found in fixed accents which must appear a line, there are a in that line in order for it to belong to number of a Spanish fixed rule for Spanish poetic composition is that all poetic category. For instance, a lines must carry an accent on the penultimate syllable accent'). The remaining accents look at the a are - acento estrofico ('stanzaic called 'rhythmic accents' proper. Let us have a following example (in Quilis 1984:35): (137) Pastor que con tus silbos 'Shepherd who with amorosos your whistles loving me despertaste del profundo sueno; me woke up from the deep sleep Tit, que hiclste cayado de ese leno, you en that made que out of that piece of wood cane tiendes los brazos poderdsos, the in which you put powerful' arms [...) (Lope de Vega) In the above stanza, - all the accents in the first and second lines are rhythmic accents additionally, they coincide with the lexical stresses in the corresponding words. As for the third and fourth third lines, the (hiclste) syllables are ones that fall extrarhythmic 127 - on the first syllable (tu) and they are on the lexical stresses which nevertheless do not fulfil any accent occurs next to a prosodically functional position. If rhythmic accent, giving rise to a an extrarhythmic clash, the former is called 'antirhythmic accent' (Quilis 1984:33-36). correspondence of stresses and rhythmic accents is The lack of a usual phenomenon in Spanish poetry. It is said (see Balaguer 1954) to have been inherited and poetry from the time when song 226) observes, if we same thing. As Balaguer (1954:223on poetry, change in certain lines of the may be a poem. it is not Indeed, similar device to that used in popular where poetic stress is subsidiary to musical accent. It could be said that the song, music and poetry run This theory about important point that any often manifested the one may displacement of prosodic accent more the take into account the influence of music surprising that the rhythmic accent the were of its first on a parallel, the more stresses get displaced. basic correspondence between poetry and penultimate syllable -, prosodic accent gets a accent a whether the time can rhythmic marker prosodic on or same way bar, always the that not every etymologic accent. the first beat of the bar; or be made to correspond to those made to map onto a on or X X X 128 moreover, in such bar, while weak position, thus giving rise to what 2/4: be observed, the last on complex (4/4, 6/8). Poetic same patterns, the main beat of the displacement': can secondary beat in the bar, depending signature is simple (2/4, 3/4), lexical stresses would fall X on strongly accentuated syllables, but templates according to different time signatures. As always corresponds to the main structures a a - this, the grids in Table 23 portray the metrical organisation of To illustrate always on rhythmic marker, in the rhythmic marker corresponds necessarily to there is another rhythmic marker situated syllables, which would correspond to the first beat in several musical the simple instance of verse has, apart form the final accent primary accent. Rhythmic markers usually fall not every song proves some we a way that some others would be have called 'stress 3/4: X X X X X 4/4: X X X X X X X 6/8: X X X X X X X X X Table 23: Musical metrical In the grids. (138), the organisation of stresses gets distorted when fragment is set into music. (138) (main beat = 1) (secondary beat/s 5/8 A esta yuerta hemos (1) con we 2) llegado (2) 'To this door = (1) have arrived intention de cantar; (2) (1) (2) (1) with the intention to sing denos licencia la nina (2) (1) (2) (1) grant us permission the girl vara ver (2) (1) de comenzar.50 (2) in order to try 50 'A esta puerta (1) and start' hemos llegado', in Manzano Alonso (2003:366). 129 a poetic leaving aside the possible influence of But song on the accentuation of poetry, there are hints that allow us to think that Spanish poets enjoy a certain degree of freedom in order to shift stresses around in accordance with the needs of the rhythm. This not may have been inherited from Latin poetry, where rhythmic accent did necessarily coincide with stress, as shown in section 2.1.1. Following this historical perspective, the displacement of stresses is by means exclusive to modem of accent This practice is also found in the nineteenth century, in contemporary poets as contemporary poetry. Indeed, there are many instances displacement due to rhythm need in the Siglo de Oro ('Golden Age') poets seventeenth century. - or no rhythmical arrangement in (Balaguer 1954:232-234). Let a us have a as well look at the fragment written in the seventeenth century (fragment quoted in Quilis 1984:193): (139) Lexical stresses jOh excelso Ah Rhythmic accents and synalephas iOh excelso oh torres coronadas muro, sublime wall ah towers crowned de honor, de majestad, de gallardia! with honour with jOh oh torres coronddas muro, gran rio, ah big river de arenas de honor, de majestad, de gallardia! majesty with elegance gran rey de Andalucia, jOh gran rio, gran rey de Andalucia. great king of Analusia nobles, yd que no doradas! of sands noble if not de arenas nobles, ya que no doradas! golden' (Luis de Gongora) The above stanza the penultimate stress tenth - a compulsory accent - second, fourth, sixth, eighth, and tenth the first and fifth they give rise to syllables of lines an accent on metrical syllable; this accent coincides with the lexical in the last word of each line. All the accents which fall syllable on - comprises four hendecasyllabic lines, with one - are and three on an even metrical rhythmic accents. The accents are antirhythmic accents in that clash. The only perfect instance in terms of the rules of the 130 hendecasyllabic line is line four, which carries rhythmic accents on the second, fourth, sixth, eighth and tenth syllables. Summarising, the most of displacement in Spanish poetry common types those which follow metrical needs, that is, those which have to do with us through isosyllabism to conclude that - as rhythm in Spanish poetry is achieved not only has long been thought organisation of rhythmic accents. These accents lexical stresses, that case seems to - but also through the periodic not coincide with may or may the requirements of the poetic template used; in the we have to talk about 'stress displacement', which widely used rhythmic device in the history of Spanish poetry. prosody in English following sections 3.3.1. I on they do not coincide, 3.3. Verse The depending be the most the need to rhythm (Balaguer 1954:228-230). re-establish the line's This allows are- are devoted to the study of English verse prosody. In section analyse the evolution of the English prosodic system, from OE to PdE, emphasising the parallelisms between the evolution of the language and that of its system of versification. Sections 3.3.2. and 3.3.3. examine the regulating principles of verse composition in English, 3.3.1. From Old syllabic The as well as the use of such devices as synalepha. English alliterative verse to Present-day English accentual- verse principles of English verse composition and scansion have undergone evolution from the first records in OE to current a radical practices. This evolution is intimately linked to that of the language itself, together with the influences from other traditions which have been Before the appearance suffered a incorporated into English throughout the centuries. of OE verse, the Germanic system of versification had major change from "hypothetical Indo-European syllabic verse" (Gasparov 1996:36), again as a consequence of the changes in the language, which ranged from the shift and intensification of stress to the reduction of unstressed syllables. In spite of the difficulties in reconstructing Old Germanic agree verse, scholars that it evolved "from syllabic into tonic [or accentual]" (Gasparov 1996:36) in the three Germanic areas, namely Scandinavia 131 - and later Iceland, too - Britain and what is nowadays contrast between stressing and by stress so is equivalent to the other group of syllables of syllables also united by group a (regardless of the number of syllables in the groups)". Britain, the OE period witnessed the flourishing of a - applied to any poem from mid sixth century to early twelfth century encoded the verse series of principles which had to be - half lines joined together into long lines by number of unstressed so was a that aimed at being labelled 'metrical'. As poetry was composed in hemistichs were - highly complex system of verse composition, called 'alliterative verse'. This system OE any on other syllable (regardless of its length, any on)", the latter is regulated by the fact that "any one stress In lines taxonomy, the main syllabic and accentual metre is that, while the former relies "any syllable is equivalent to basis that united Germany.51 Following Gasparov's (1996:37) - general rule, OE also called 'verses'. Those half of structural alliteration. The means syllables in between stressed a variable and, ones was as a result, thus, the surface basis of length of hemistichs and lines. Alliteration was, composition. Its working principles the following: in the second were hemistich, the first metrical beat should alliterate with either beats in the first hemistich. Let us one or two observe how alliteration works in the fragment from Beowulf - the segments which alliterate are metrical following in bold: (140) (from Gasparov 1996:39): Beowulf wees breme / - Scyldes eafera / In bleed wide sprang — Sceldelandum in.52 spite of the fact that alliteration has been analysed as the most idiosyncratic feature of OE poetry, some metrists (McCully and Anderson 1996:44) regard it superficial feature, and not as an metrists, OE alliteration in accentual languages, and Romance and later 51 was a The information that so essential building element in OE poetry. For those stylistic marker, used to distinguish verse from prose its role could be compared to the role of rhyme in English poetry. Gasparov (1996:38) we as a have about Old Germanic verse AD. 52 Translated by Heaney (1999:3): Shield had fathered through the north. 132 a goes argues back only famous son: as that Old Germanic far as / Beow's the ninth century name was known languages did not develop rhyme instead of alliteration because of the suprasegmental characteristics of those languages. They had beginning of the word, in speech and, as a so that roots were more a strong stress at the prominent than the reduced suffixes, result, in poetry. Rhyme works in the opposite way, as it is applied to the coda of the last stressed syllable in the word onwards, somehow shifting the prominence from the first to the last segments. Compare (140) with the following fragment from King Horn (c. 1225), the first courtly (rhyming segments romance in English in bold): are (141) (from Gasparov 1996:182): A song ychulle ou singe OfAlloff>e gode kynge Kyng he by weste, wes Pe whiles hit yleste The -53 comparison between (140) and (141) points at both alliteration and rhyme verse-building devices which employ similarity of sound ends; on the other hand, a fact that is linked to the opposition between syllabic metre. A second observation that, although OE verse had on a alliterative verse, made will see position of the unstressed syllables in the next was was a variable number of purely accentual, because the regulated to some extent, as we paragraphs. Metrists have for years that existed in by Duffell (1996:211), is fixed number of stresses and syllables, it would not be accurate to state that it number and their basis and mark line observe that, while alliteration binds hemistichs into we lines, rhyme binds lines into stanzas, accentual metre and as as OE.54 In tried to give a clear-cut account of the types of lines spite of the ongoing controversy, the most widely accepted taxonomy of verse types is the one established by Sievers in 1885. Sievers used a set of stress patterns as a starting point, and claimed that the designs of stressed and 53 Translated by Weston (in Gasparov 1996:182): A song I now will sing/ OfAllof, crowned king; / He reigned in the West / While he with life was blest. 54 In an insightful article, McCully (1996:42-58) revises Sievers's theory and explores the possible verse-demarcation devices in OE poetry, highlighting the importance of the relationship between stresses and syllables at the end of the verse. 133 unstressed syllables within each hemistich could be organised into five types through E. Cable (1974, 1991) expands Sievers's theory, basing it stressed and unstressed a are syllable each. According to this, a substitute for one or a prototypically but not exclusively occupied by syllables series of unstressed the first two from two to five - which could /x/x B x/x/ C X D //\ E /\ every /\ X X X / verse types. hemistich in OE will correspond to one of the patterns above. In the following example, from the poem of - syllables in the half-line.55 Table 24: O 3 According to Sievers, one half-line will have four syllables; the exceptions to A stress the counting of property of early Germanic verse; as mentioned above, positions, which this will be A syllables to form positions. It is important to note that simple syllable-counting is not Cable counts on - syllables unstressed we can see the scansion of several OE half-lines Beowulf - seventh-tenth century AD - ones - both in terms of the counting Cable's theory, indicated with / for stressed syllables, and - and the indicated with numbers. counting of positions - x for Sievers and Cable's theories, (142) marshals the classification of some lines from Beowulf according to Sievers's taxonomy. 55 theory are the concepts of 'dip' and 'lift' (Beare 1957:105-106). The former was a position (\) typically occupied by a single syllable, but sometimes occupied by more than one weak syllable. The latter was a strong position (/) typically occupied by a main stress, but sometimes by a secondary stress - in types A, B, C in Table 24 - or a stressed plus an unstressed syllable. Crucial to this weak 134 (142) (from Cable 1996:10): Type of line Line in Beowulf Scansion 12 A / X X 1 XX X fordam he X 4 / X under wolcnum weox B (8a) 3 2 3 4 / X / manna 1 2 3 4 XX / \ (2645a) mcest (482a) X Poet hie in beorsele 12 3 11 \ D (31a) 4 x leof landfruma 12 3 4 / x / \ (564b) scegrunde neah A theory of OE metre would not be complete if it only accounted for the stressed and unstressed positions in each hemistich and line. The four possible combinations of stresses within each hemistich rendered something much more interesting than simple metrical patterns, namely intricate melodic patterns (Table 25). In Cable's (1974:94) words, 'given four metrical positions and certain constraints upon those positions, contours', of which the last three a model of metre will produce exactly eight were not permitted in OE poetry. 1: 1 \ 2 / 3 \ 4 2: 1 / 2 \ 3 / 4 3: 1 / 2 \ 3 \ 4 4: 1 \ 2 \ 3 \ 4 5: 1 \ 2 \ 3 / 4 (6): 1 / 2 / 3 \ 4 (7): 1 / 2 / 3 / 4 (8): 1 \ 2 / 3 / 4 Table 25: Melodic contours in OE hemistichs. 135 The actually implications of this approach seems to As Cable formula, far beyond metrics, in the go sense that it downgrade the importance of stress in favour of pitch movement. (1974:96) puts it, "the metrical basis of OE poetry a contour of pitch drawn from eight, to which words are a set the melodic was of five contours of a possible set of fitted according to stress rules". In this view, metrical patterns are subsidiary to melodic paradigms, used by poets to decide upon which words could fit to a specific paradigm. Cable (1974:96) himself used musical notation illustrate the melodic paradigms found in Beowulf. Despite the plausibility of this theory, in later works Cable (1991:55) restates it, arguing that the melodic paradigms "were the result of a metrical paradigm and not the engine that is clear is that, in OE, metre interacted with which should not be Unfortunately, English songs no OE songs preserved with music from the middle of the twelfth century, classical a natural are St. Godric's, probably which nevertheless have echoes of the important elements in OE metre versification, the wrong placement of a stressed position, unstressed syllable, would make the whole The Germanic tradition of alliterative after the Norman English stress, measures, breath verse rooted in a verse verse experienced a new one placement of unmetrical. did not stay the same in Britain hand, the Germanic inheritance continued though relaxing the principles of OE alliterative which still relied verse. half-lines, each generally of two stresses, and unstressed wrong types of verse emerged in this period, each of them different tradition. On the However, half-lines carried More the series of innovations, helped by both Germanic and be cultivated in the form of accentual verse, alliteration. or Conquest, in 1066. Between the tenth and the fourteenth centuries, Latinate influences. Two in OE. were foot divisions, and inferred pitch contours. Under the rules of OE groups, two was meant to have survived in musical alliteration, quantity, resolution and suspension of resolution, to something Anglo-Saxon verse form (see Baltzer et al. 1991). We know that the most an way, surprising if we take into account that OE poetry be recited, chanted or sung. notation. The earliest melody in driving it". What was a Lines many were on alliteration, still composed of half-lines were linked by greater number of unstressed syllables than specifically, the first half-line almost always had two or more syllables before and after the first stress. In (143), extracted from Sir 136 Gawain and the Green Knight, we can observe that the syllables before the first stressed one, which was presence of two unstressed legitimate in ME, would have made the half-line unmetrical in OE: (143) (from Cable 1996:9): X / X And he X into come and the / X granted and hym gafe A number of models X forms indebted to both Medieval Latin and new Anglo-Norman play at the end of the twelfth century. The Latin quantitative foot English accentual foot were connected through French, which was purely syllabic. A hexasyllabic line in French, characterised for having six syllables and rhyme at the end, could respond, in English, to five types of metre, depending whether the beat fell every third syllable same would on - alternate syllables - either odd- or even-numbered beginning either with the first, second, happen with a heptasyllabic line, and alternative scansions rendered in or on or on -, third syllable. The so on, as we can see in the (144). (144) Trochaic (/ x): Fear no/more the/ lightning/flash (Shakespeare, Cymbeline) Iambic (/ x): In any case, classical Latin Fear/ no more/ the llght/ning flash the classical names for those five types of metre, which in responded to length contrasts, were kept in English to designate stress contrasts, hence the denominations 'trochaic', 'iambic', 'amphibrachic' (see Beare The second structural 'dactylic', 'anapaestic' and 1957:33).56 important innovation is rhyme, which starts to be used as a principle, borrowed probably from Medieval Latin, but also due to the 56 Early prosodists not only applied the foot taxonomy borrowed from Latin to English verse, but also length with stress, a confusion that has survived until the present day. It was only in the eighteenth century that prosodists such as Bysshe {Art of English Poetry, 1702) and Johnson {Dictionary, 1818), rejected this idea and started to dissociate length and stress (Beare 1957:34). identified 137 influence of French, particularly certain Anglo-Norman forms (see Preminger 1974:355; Gasparov 1996:41). At the beginning, internal rhyme breaks the short alliterative couplet, until eventually it supplants alliterative long line into alliteration altogether. In general terms, the introduction of rhyme marks accentual to verse the number and In a accentual-syllabic verse, where both the number of syllables and spite of the Romance influences, alliteration continued to be come to a revival of this principle, in be called 'Alliterative Revival'. There Layamon's Brut (ca. 1200) poems or the which anonymous poem are a of English medieval poets, to conform to alien to their native accentual system. the French are also a movement some remains of The Owl and the Nightingale completely different with regard to internal organisation (Gasparov 1996:182). The latter is was widely tradition, like, for instance, the interaction between the Germanic and the Romance (ca. 1200), two a principle throughout the thirteenth and fourteenth centuries and, indeed, the fourteenth century witnessed the part shift from position of stresses in the line become relatively fixed. cultivated structural which has a Let a us good example of the attempt, model - the syllabic one - on which observe (145), written following principle of syllabism: (145) (fromBeare 1957:33): po stod on old stoc par bi-slde par po ule song hire tide It is not until the fourteenth century Chaucer shifts considered eleven that the attention of poets such as completely towards strictly French and Italian forms of composition, more syllables elegant and cultivated. Chaucer imitated the hendecasyllabic line per like Dante, Petrarca, line -, which was - being cultivated by French and Italian poets Boccaccio, but he transformed it into the iambic pentameter, a truly accentual-syllabic kind of metre composed of five feet, with prototypically two syllables (unstressed eliminated - - stressed) in other words, per foot. As a result, only double-timed feet any were triple time line was kept. The following example (146) of iambic pentameter lines is taken form the General Prologue of the Canterbury Tales: 138 (146) (from Gasparov 1996:184) Whan that Aprille The with his shoures soote droghte ofMarch hath perced to the roote And bathed every veyne in swich licour, Ofwhich vertu engendred is theflour, (...) After Chaucer's major shift toward Romance models, the modem period sixteenth and seventeenth centuries - alliterative forms, which nevertheless witnessed the definitive persisted up - decay of accentual and until 1550. The iambic pentameter transformed by Petrarchan French and Italian of Chaucer and his successors, influences, had by then become the most widely used pattern of English verse. Shakespeare used the iambic pentameter line extensively, and indeed carried it a step further, complicating it by the addition of stresses which increased the density of lines. In (147), extra stresses are indicated in bold: (147) (from Guthrie 1991:90) Be thou the tenth It can in such an a way Classical Latin to ten times more in worth that, around this time, English be said currently is, that is, grows muse accentual-syllabic type of that Elizabethan poets verse gradually becomes what it verse. The Romance influence adapt the patterns of even manage to English, replacing quantity with stress as the marker of the pattern. Throughout the seventeenth, eighteenth and nineteenth centuries, poets like Donne, Milton or models by English poets, and finally arrive at the Present-day accentual-syllabic Browning contribute to the definitive adoption of Romance and Latin model of composition and scansion. 3.3.2. Stresses and syllables in contemporary English verse Present-day English verse is built specific number of stresses per upon the basic principle that there has to be line, each of which gives rise to and respective foot. The placing of stresses in the line responds to is thus fixed. This Shakespeare, as seen principle was a governs 139 as (148b): its specific pattern, and extensively used by Renaissance poets such in (144), quoted below a as (148) Verse instance Verse pattern a. Iambic To rear/the co-/ lumn or/ the arch/to bend (x /) (Pope, Epistle to Burlington) b. Trochaic (/ x) Fear no/ more the/ lightning/flash (Shakespeare, Cymbeline) This entails that, whereas in pure accentual verse "the stresses must all be true speech-stresses" (Preminger 1974:771), in accentual-syllabic following a promotion of'or', The second identical. If the mind, metrical pattern, will adjust the stressing to fit via the strategies of promotion and demotion of syllables metrical verse, we line have two an - observe, in the iambic instance in (148a), the otherwise unstressed syllable. principle states that the number of syllables look at (148) above, syllables, and the per foot must be observe that all the iambic feet in the first we same goes for the trochaic sequence, with the exception of the last foot, which has the second (weak) position empty (flash - 0). With these two principles at work, English verse is considered accentual- syllabic and not only accentual. Nevertheless, the type of accentual inherited from Germanic verse, can also be found in verse which was English. It is the type for much popular folk poetry oral poetry and also for the English and Scottish ballads. As Preminger (1974:770) points out, in such compositions, isochronous in timing (although stresses do not fall accentual-syllabic verse), the phenomenon known and reveal its direct as origin in The fact that the system syllabic depending in English. On the occurrence of four 'dipodism' or use are on fixed positions like of metrical rests, and the all characteristic of such close association with of English verse can verse song. be both accentual and accentual- on the form of composition, has to do with the traditions merged one hand, the native tradition of Old Germanic metres requires the syllables carrying major stress in each line, while the number of syllables in the line is relatively less constrained. On the other hand, the Romance metres imported in the Middle Ages regulate the number of syllables strictly. From the fourteenth century onwards there was a 140 clear proliferation of Romance poetic forms, since the English imported Romance morphological base which had been built verse upon forms to map onto a shifting Germanic and Latin strata (see Preminger 1974:779). Table 26 marshals the in different stages possible combinations between syllables and stresses in the evolution of English versification: Position of stress Fixed Accentual-syllabic Fixed Number of syllables Free Free Syllabic verse verse E.g.: Present-day English E.g.: Medieval Latin, French Accentual Free verse E.g.: Old English verse E.g.: Present-day English Table 26: Verse systems. The accentual-syllabic metre adapted to English first by Chaucer and again by Wyatt and Surrey from French and Italian syllabic models centuries of its constraints on use in England principally stress, and with constraints both a during the first syllable-counting metre with secondary only later developed into on was a really accentual-syllabic syllabic counting and also on verse, stress placement. As Preminger (1974:786) well indicates, the requirement that each line had to have definite number of a syllables entailed the adoption of specific devices in order to adjust the number of syllables in the line. Of those devices, the most salient one was synalepha, first imitated from French and Italian by Chaucer, and then cultivated by latter poets, 3.3.3. such as Milton, Donne or Jonson. Synalepha in English: a comparison with Spanish In sections 2.2.3. and 3.2.3. I showed that is not completely free since it depends each utterance, intervals. there must be a Spanish stress at the level of the utterance on the overall rhythm of the utterance; for number of rhythmical leaning points at specific This entails the existence of certain compression devices, the most important of which is synalepha. So, in Spanish, synalepha is primarily phonological rule with the a a predominantly rhythmical function, and manifests itself in prosody of everyday speech. 141 Let us summarise the principles that regulate synalepha in Spanish, section 3.2.3. The realisation of of colloquial. There is also a occur, a continuous rise or a continuous fall in The domain of synalepha words, but sometimes it extends under synalepha is five, but the most There are three types a occurrence single syllable prominence. is the syllable over and register, phonological condition for the synalepha in Spanish, namely that the two vowels combined into form in synalepha is regulated by two non-phonological conditions, namely tempo, which must be fast for synalepha to which must be as seen across words. It normally covers two three. The maximum of vowels joined together common synalepha of synalepha groups two (Canellada vowels. 1987:54), namely diphthongisation, (149) mi amigo -> mja-mi-go ('my friend') reduction, (150) se acabo -> s(e)a-ka-bo ('it's over') and elision. (151) se acabo -> sa-ka-bo As for the combinations which possible to establish non-stressed an can give rise to synalepha in Spanish, it is acceptability gradient, whereby synalepha syllables is always realised, (152)y estaban -> jes-ta-ban ('and they were') (153) [a] prevails over [i] in mi amigo -> mja-mi-go (154) si unian -> sju-ni-an ('they united') 142 among vowels of synalephas while always among vowels of stressed syllables are more complex and do not occur. e-choel-ci-mien-to ('he laid the foundation') (155) echo el cimiento (156) Subido he por tu pena Su-bl-doe-por-tu-pe-na ('I've climbed your sorrow') (1980:210) observes, synalepha in Spanish is not only As Piera device, but a line)". As of the works very and "phonological rule [...], which then becomes a a prosodic metrical rule (property device used for metrical composition and scansion, synalepha a productively for Romance languages such Portuguese, where "the adjustment of the verse Spanish, Proven9al, Italian as line to the required number of syllables is achieved by synalepha [...] by elision or by hiatus" (Preminger 1974:715). The status of synalepha in English is not that of a phonological device which metrical function. Rather, synalepha in English is metrical device then acquires with non-phonological status. In Napoli's (1975:403) words, "[sjynalepha is a optional convention in English verse". This fact points to a a an structural difference between Spanish and English in terms of suprasegmental phonology, but also in of segmental phonology, because it explains the relationship between vowels terms in the two form a languages. In Spanish, the normal tendency is for consecutive vowels to diphthong; vowel system, the other hand, diphthongs on are synalepha as an Generative metrics - rules that as as a independent a dichotomy between those languages phonological as well as a metrical rule, and those exclusively metrical device. sees relate as the result of the joining together of two presented, thus, with synalepha encoded which observe rules' diphthongisation, and diphthongs behave specific characteristics, not single vowels. We which have entities separate from the but rather the result of the joining together of two vowels. In English, consecutive vowels avoid units with are not synalepha as one abstract patterns 143 of the so-called 'correspondence to surface lines - and makes it interdependent with the constraint against stress maxima appearing in W positions (Napoli 1975:401). When Halle and iambic pentameter which we Keyser (1971) proposed in English, they defined synalepha assign may a or separated by single metrical position (S, W as "a metrical convention by incorporating at most two vowels a sonorant sequence (immediately adjoining to series of correspondence rules for a a sonorant consonant) within a line of verse X)" (Napoli 1975:402). or constitute Related to this is the idea that certain segments a more powerful syllable boundary than others. In Youmans's (1996:191) words, "syllabic boundaries are weaker when nuclei when separated by single [+sonorant] segments and stronger are separated by [-sonorant] segments". According to the sonority scale, thus, the weakest boundaries [1, r], nasals [m, n, glides [j, w] and the segment [h], followed by liquids the are q] and voiced fricatives [v, 6, z]. At the other end of the scale the voiceless fricatives and the stops, and therefore prevent single poem which constitute powerful syllable boundaries, synalepha from occurring. It is obvious that where the number of are poetic syllabification can be understood syllables would remain constant, written in a or as an as an abstraction, idiosyncrasy of each given language. The difference between the first languages and the second group is that the former have a group of series of phonological principles encoded in their system, which regulate the possible realisations of synalepha, and rule out those of the cases which do not respond to language. In the latter, synalepha is just a a phonological tendency device ad hoc, with much weaker regulating devices, which never correspond to the phonology of the language. In fact, the latter do not show and the actual scansion of instance and its The verse correspondence between the metrical pattern instance, or between the scansion of a verse pronunciation. In Halle and Keyser's (1971:171) words, assignment of syllables to positions is, of course, strictly metrical. It does not be slurred are a a not imply that the syllables assigned to or elided when the verse instructions for poetry single position should is recited. The recitations. They principles of verse construction whose effect verse a is indirect. 144 on correspondence rules are, rather, abstract the sound of the recited The history of synalepha in English is one of idiosyncratic adaptation of a foreign device, which in principle responded to the segmental and suprasegmental characteristics of Romance languages. In English, synalepha is indeed feature of seventeenth and eighteenth century prosody - a common Milton and Pope used it conspicuously -, where it is used to maintain the syllabic conformity of lines. Indeed, many same sixteenth and seventeenth century poets thought it word twice in a single line or in adjacent lines, a once beauty of verse to contracted, use the in full. As once Preminger (1974:786) indicates, compression and decompression devices could not always be signalled by spelling in sixteenth to seventeenth century English Preminger (1974:787) also points out that it that was not verse. until the time of the Romantics English poets started to admit to the familiar decasyllabic line extra unstressed syllables which did not allow of contraction, whereas poets of the sixteenth, seventeenth and eighteenth centuries - non-dramatic poets, at containing extra syllables whole not to have written lines any rate - - seem on the the feminine ending excepted -, which could not be resolved phonologically. As The most stressed happens in Spanish, English also has rules for the realisation of synalepha. general tendency is for synalepha to assign two unstressed syllables to a (158) 'Tis virtue alone, where or my mind is set.51 nothing that knits friends. assign two fully stressed syllables to synalepha joins together stressed weakly single position. (157) Wholly in heavenly things It does not or an unstressed a or single position. There are instances weakly stressed syllable and a fully syllable. (159) But though I have found thee, and thou my (160) For that is first required, own. a man 57 be his thirst has fed. Examples (157), (159), (161) by Donne (Sonnet 17), (158), (160) by Jonson {Epistle to a Squib), (165) by E.B. Browning {Sonnets from the Portuguese, XXI), all borrowed from Fabb (2002). Examples (162) by Rossetti {Song), (163) and (166) by Donne {Sonnet 17), (164) by Becquer {Rimas). and 145 (2002:8) gives Fabb of a clear account of synalepha using syllables from the metrical pattern to the every syllable in a theory of projection instance. As verse a general principle, line projects onto the grid and is symbolised by an asterisk: • a 1 58 (161) she Since loved I (2002:9) also elaborates Fabb that whom a paid hath her last debt series of rules which regulate non-projection, is, realisation of synalepha in verse lines. They go as follows: Non-projection rule a: Do not project of the line which is unstressed a a syllable at the (right-hand) end weak in stress and which or comes after strongly stressed syllable. (162) And if * * thou * wilt * re mem * * ber 0 Non-projection rule b: Optionally, do not project on a vowel, when that syllable precedes a a syllable which ends syllable which begins on a vowel. Optional rules like b apply where necessary in order to bring the number of syllables to the right number. This is the equivalent of Spanish synalepha. The difference between rule is Spanish and English with regard to this specific non-projection that, in Spanish, the application of synalepha is the unmarked option, as it corresponds to a phonological rule of the language, while in English, it is the marked option, brought forward as an exclusively poetic artifice alien to the phonology of the language. Compare examples (163) and (164): (163) And her soul early into heaven ravished, 58 (2002) develops an overall theory of scansion. This first projection principle is complemented series of other principles, so that, in the end, a grid with asterisks and brackets as the main visual Fabb by a tools is built. 146 (164) Del salon en el angulo angle dark' 'Of the hall in the In (164), verse we see a oscuro. synalepha realised metrically in order for the line to conform to pattern, which regulates that the line must have a specific number of syllables; this is the same (163). However, while the pronunciation of the synalepha in for (164) responds to phonological rule of the Spanish language a everyday speech would be realised in the - one - that synalepha in (163) is only metrical a convention, which does not correspond to the pronunciation of those two words any a - or other two words with similar segmental structures - in connected speech. Non-projection rule its nucleus or one which has c: of the as Optionally, do not project a syllable which has following sonorant consonants: [1], [r], [m] its nucleus the weak vowel schwa followed by or as [n], one of (Fabb 2002:10): these sounds (165) Say o ver gain, a **Q** Fabb himself (2002:45) and * once ** o ver a gain * Q * * observes later in his book that The vowel e, which is often cut off and in yet syllables before r, as by an apostrophe in the word the, dang'rous, gen'rous, &c. ought to be preserved in the pronunciation, because the syllable it forms is as to admit of being sounded with the preceding syllable, increase the number of syllables to This confirms, once metrical pattern, not the ear, or at so short so as not to all hurt the harmony. again, that non-projection rules have to do with the abstract with the pronunciation of verse instances. In the examination of the phenomenon in question presented by Halle and Keyser (1971, in Napoli 1975:401-403), instances like (165) above would also qualify as synalepha in English. In Spanish, synalepha vowels without any of synalepha intervening consonants; in Spanish. 147 no can only happen between other segments can enter the process Non-projection rule d: Optionally, do not project monosyllable grammatical preceding monosyllable (i.e. preposition, article, a syllable which is grammatical another pronoun, a auxiliary verb, modal verb, conjunction, etc.) (Fabb 2002:27): (166) To nature, to hers and my good is dead. In Fabb's to (2002:1 Iff.) view, the choice to project project is related linguistic representation, and not to the performance of the line. Projection has the nothing to do with pronunciation: the syllables which pronounced metrical pronounced. So synalepha is or not representation of the that the line fits the metrical poem. terms of scansion, a are not seen as a projected can be fact about the abstract In other words, projection is a way of ensuring grid which is built from the linguistic representation. Independently of this, there is also the or not to a choice to decision has to be made on pronounce or not to pronounce. In which syllables to project and then grid is construed from the projected syllables. If the grid fails, the process starts again by projecting different syllables. conclusion, synalepha is In in languages others as diverse encoded -, instance extra syllables not projected. a project as many some syllables as among required by the metrical pattern, and leave languages, such a as Spanish and Italian, synalepha is was not even a metrical rule. One could clearly observed in the methods of composition of English poets, for instance Chaucer. Later and hypothesise that, in English, correspondence rule until the Middle Ages, when influence of Italian writers started to be Rossetti it has been attested phonological rule which is realised phonetically in everyday speech, which then becomes synalepha - Sanskrit, Persian, Arabic, Italian and English, as verse primarily universal metrical device correspondence rule which has the function of making the as a However, in a on, in the nineteenth century, poets such as Browning further adapted Italian poetic forms to English, thus increasing the Romance influence on English Poetics and Metrics. 148 Summary 3.4. Chapter 3 is devoted to the analysis of certain metrical patterns used in speech. This makes uttered in mean that derive from prosodic patterns of phrases verse even more sense manifestation of metrical patterns does not There is evidence that the origins of verse. if consider the fact that the we themselves is intended for spoken language. This speech prosody and verse prosody be equated, given that can verse has verse prosody is self-conscious in formal terms, while speech prosody is not. For this a conscious aesthetic dimension which the study of specific reason, verse prosodies common can shed speech lacks. In this some ways on the rhythmic more broadly, the light principles that underlie the corresponding speech prosodies and, sense, in which different languages parameterise and articulate the general rhythmic capacity. Verse the prosody is also known as 'metrics', which is usually taken to refer to study of the art of versification. As happens with speech prosody, metre subclassified into three main types, establishes the depending on the element that can recurs be and rhythm of a specific composition. The first type is 'syllabic metre', where each line is made up of a specific number of syllables. This metrical type corresponds to the versification of the so-called syllable-timed languages in speech prosody, whose rhythm, as explained in chapter 1, is based on the isochronic repetition of syllables. The second type is 'accentual metre', where stresses fall at regular intervals in the line. The third type is called 'accentual-syllabic metre', which is a less combination of accentual and syllabic metres, and pays attention to the regular alternation of stressed and unstressed syllables in of metre can be related to the rhythm is based on the a more or line. These two types speech prosody of stress-timed languages, whose recurrence Section 3.1. examines the of stresses at equal intervals in time. relationship between speech prosody and poetic rhythm and metre in depth, paying particular attention to the languages analysed in this dissertation, namely Spanish and English. Sections 3.2. and 3.3. explore the historical evolution and specific characteristics of Spanish and English verse prosody. As to the explained in section 2.2., Spanish is correlation between a syllable-timed language. According speech prosody and 149 verse prosody advocated in this dissertation, Spanish verse is syllabic. The syllable constitutes the main rhythmic unit in Spanish verse, which can be proved by the fact that much Spanish perfectly regular syllable count. In fact, the names 'octosyllabic', 'hendecasyllabic', and - of syllables are an - has a 'heptasyllabic', explicit reference to the number that they contain. Spanish metrics has been marked by the contact with other The evolution of Romance traditions. into so on of Spanish lines verse Originally, Spanish had an accentually-based metrics that turned syllabism due to French influence. In spite of this, accents in Spanish arguably at occur specific intervals. Throughout history, poetry written in Spanish has followed two second one clearly different paths. The first tendency is isosyllabism, while the comprises non-metrical lines. This second tendency, separated from isosyllabism, includes all the poems defined as 'non-isosyllabic', where lines have a non-fixed number of syllables. Isosyllabic poetry series of metrical devices in order to regularise uses a syllable counts. Curiously, those forcing stresses to fall metrical. Two of those while the other two, same metrical devices also serve the purpose of their corresponding syllables, thus rendering the line on devices, synaeresis and dieresis, happen at the word level, synalepha and hiatus, happen at any level beyond the word. As analysed in section 2.2.3., synaeresis and synalepha follow the tendency of Spanish to make contiguous vowels belong in the same unification. Other metrical devices that take aphaeresis, syncope and apocope, syllable in of syllable a process place within the level of the word are and prosthesis, epenthesis and paragoge. Although Spanish stress is determined by rules of assignment, the lines in each kind of poetic composition syllables, which does not stresses mean in order for the line to carry a are systole stress goes - the stress be acceptable. In fact, there a are two specific specific devices shifting of the position of stress within back to the previous syllable forward to the next syllable in lexical stresses found in appear goes on that rhythmic accents must coincide with lexical that have to do with the deliberate these number of rhythmic accents line, there a are a in that line in order for it to belong in 150 a word. Last, - on top a and diastole word; - the of the number of number of fixed accents which must Spanish poetic category. Summarising, Spanish of Spanish verse a significant amount is composed and scanned by counting the syllables in each line. In order for lines to be number of prosody is syllabic in that verse isosyllabic, several devices used in order to adjust the are syllables, namely synaeresis and synalepha tendency of the language - which respond to a natural and dieresis and hiatus, together with several other - Apart from isosyllabism and despite the marked syllabic character of devices. Spanish, there is another principle at work in Spanish of stresses at recurrence verse prosody, which is the roughly regular intervals. This is helped by the devices quoted above. As observed in section make think of its us 2.3., English is a stress-timed language, which would prosody verse as accentual. Nevertheless, English accentual-syllabic. The characteristics of PdE hybridity, inherited from Romance period of a - one. a is prosody have to do with its fruitful interaction between the native traditions and the Old Germanic verse evolved from syllabic into accentual. The OE from mid. sixth century to beg. twelfth century - witnessed the flourishing highly complex system of verse composition, called 'alliterative verse'. At the end of the twelfth century a and verse verse number of new forms indebted to both Medieval Latin Anglo-Norman models English accentual foot The second came into play. The Latin quantitative foot and the connected through French, which were important innovation was was purely syllabic. rhyme, which started to be used as a structural principle, borrowed probably from Medieval Latin, but also due to the influence of French, particularly certain Anglo-Norman forms. It was not until the fourteenth century that the attention of poets such as Chaucer shifted completely towards strictly French and Italian forms of composition, considered cultivated. It it can be said currently is, that is, an that, around this time, English verse gradually became what accentual-syllabic type of verse. Present-day English a verse specific number of stresses per is built upon the basic principle that there has to be line, each of which gives rise to and respective foot. The second principle states that the number of syllables be identical. The foot tradition. English feet elegant and more as are a unit of scansion was per its foot must inherited from the Greco-Latin said to be isochronous, although, 151 governs as happened in speech prosody, isochrony is perceptual. The names of the four basic types of English feet have also been inherited from the classical tradition. As for the devices which are used in English in order to adjust the duration of feet, synalepha is a salient one. However, synalepha in English is not a phonological device which then acquires a metrical function. It is, rather, a metrical device with non-phonological status. This has to do with the fact that, in English, vowels avoid diphthongisation, and diphthongs behave characteristics, not a verse pronunciation. In a correspondence between the metrical pattern and the actual instance, or between the scansion of poem. Independently of this, there is also to instance and its pronounce or not to the syllable in question. conclusion, synalepha is a universal metrical device attested in many languages, which works in English of a verse Synalepha in English has to do with the abstract metrical representation of the pronounce independent units with specific the result of the joining together of two single vowels. In fact, as the latter do not show scansion of as making the verse as a instance project correspondence rule which has the function as many pattern, and leave extra syllables not projected. 152 syllables as required by the metrical 4. The In relationship between speech, verse and music chapters I, 2 and 3 I explored the connections between speech prosody and prosody. In this chapter I examine the triple relation between speech, verse verse and music, in order to discern whether there exists any kind of correspondence between speech prosodic typologies, the interaction between prosody and musical rhythm. Section 4.1. analyses verse prosody and musical rhythm, arguing that the verse characteristics of the former have deals with the main important introduction to the theoretical as an methodology which will work as section 4.2. is devoted to the summary applied to metrical analysis, as the other. Section 4.3. introduces the main folk song. Section 4.3.1. establishes main characteristics. criteria Section 4.3.2. applied to the corpus of a seven subsections contained in one hand, and to text-setting analysis, definition for folk song and summarises its explains the transcription and classification analysed in this dissertation. Section 4.3.3. songs and and Section 4.3.5. explains the song. song corpus, reasons for the choice of a namely Manzano Alonso's (2003) Cancionero de Burgos Kennedy's (1984) Folk songs of Britain and Ireland. In section 4.3.6. I present some preliminary conclusions extracted from the analysis of mismatches in the corpus of 444 Spanish folk IV and V), while section 4.3.7. contains songs and 239 English folk a from it. Section 4.4. is intended extend the theories as a songs (see also Appendices deep analysis of two specific folk extracted from the above-mentioned corpus, as to song, complements this by exploring the general characteristics of text-tune relationships in folk specific folk on object of study in this dissertation, namely briefly introduces the topic of the importance of lyrics in the study of folk section 4.3.4. corpus and critical review of an article dealing with the on principles the fundamental bases for the analysis carried out in this dissertation. Each of the OT the latter. Section 4.2. subject of this dissertation, namely text-setting, framing its study within OT. This section is intended and research consequences on well as the conclusions that complement to sections 4.2. and 4.3., as songs emerge it seeks explored in the former and then applied in the latter by checking them against art song. that of section 4.3., in this case The internal structure of this section is parallel to always referring to art 153 song. 4.1. The interaction between As has been prosody and musical rhythm verse argued in this dissertation, verse and music share speech lacks - "whereas music characteristically functions (Lerdahl and Jackendoff 1983:7) - verse, like music, shows the auditor's as art, language does not" a dominant instrumental dimension; in other tool for transmitting meanings 'outside' itself, while clear aesthetic dimension, that is, its main goal is calling observer's attention towards its form. Structuralism held the view or and music that poetry a as a quality which natural where 'language' stands for 'ordinary spoken language'. As already stated, speech has words, it is normally used a are autonomous in that their meaning does not depend on the exterior world: So far, this is simple and straightforward enough: practical language finds its justification outside itself, in the transmission of thought interpersonal communication; it is language, on a means and not an or in end [...] Poetic the other hand, finds its justification (and thus its entire value) in itself; it is its own end, and no longer a means (Todorov 1988:12). Beyond the assertion that both poetry and music always accepted and Poetry is it are a even forms, poets have vindicated the inherent musical quality that poetry has: composition of words set to music. Most other definitions of indefensible, music may, leaves are art or worse, and does, music, or at vary; least an metaphysical. The amount or quality of the but poetry withers and 'dries out' when it imagined music, too far behind it (Pound 1917:90-91). Rhythmic devices articulate the flow of time in aesthetic dimension that both poetry is an intentional way, hence the and music share: "Though the material of poetry verbal, its import is not the literal assertion made in the words, but the assertion is way the made, and this involves the sound, the tempo... and the unifying, all- embracing artifice of rhythm" (Langer 1953:260-61). As Langer observes, the element that makes poetry of rhythm as an a real and music such a close pair is rhythm. The pre-eminence abstract property of measuring time not only shows the existence of affinity between poetry and music, but it also proves that poetry is the only real analogue for music in the realm of human language: "the principal definition of 154 objects lies in their rhythmic articulation. More than musical articulates time and verse and music is bidirectional. Composers overwhelming predilection for using poetry, rather than for their vocal works. If one has one dimension, rhythm emphasises its forward movement" (Barry 1990:65). The link between find that any of the very a look at the different prose, as have shown an- the verbal material existing vocal genres, one will few forms where poetry is not used is in the canticles of the Christian tradition of sacred Western music. regards the most characteristically vocal As especially overwhelming in songs, which constitute as of poetry is genre. In a traditional lyric match and a much as its meaning: the patterns of suggest a metre, the rhymes define a melodic phrase and the phonetics inspires stress melodic shape. This is also true for opera, libretti written in verse. language and music in of lyric at least a use phrase. Thus, songwriters value the reinforce the musical articulations of bar and properties of pure sound of the eminently lyrical an the poetic articulations of metre and rhyme of song use genres, text a into a In fact, like a genre one opera a dramatic in which genre composers often of the most intriguing relations between is the one which specific musical construct. Depending emerges on from the setting the specific operatic subgenre, the setting of texts is utterly different. Just to mention two of the most important subgenres of opera, 'recitative' is a type of vocal writing, normally for a single voice, which follows the natural rhythms and accentuation of speech and its pitch contours; 'aria', of a on the other hand, refers to essential an treatment is The metre. they as in are while recitatives work of expression for passages in as well as a as a a a or part musical tune. Arias are vehicle for dialogue, that libretto for which lyrical connecting link between arias. rhythmical hierarchy that mediates between poetic text and music is based significant similarity between poetic metre and musical metre is that on the alternation of strong and weak events, which English poetry - or syllables - of text-setting has to do with chosen poem an means inappropriate, The most process is either independent larger work, which results from the setting of poetry to the dominant elements in opera, is, a song, a as can be stresses - in Spanish poetry. According to this, the matching between the metrical pattern of the and that of the corresponding musical composition. Nevertheless, there essential difference between metre in poetry 155 and in music which has to do with the arrangements English there of strong and weak positions. As discussed in chapter three, in are a number of possible arrangements of stressed and unstressed syllables that result in make no a variety of feet. Unlike poetic foot divisions, musical bar lines distinction between iambic-trochaic, or avoids this anapaestic-dactylic, metres. Music apparently unreal dilemma by automatically inserting bar-lines, which always fall before the main beats (Attridge 1982:20). pointed out in section 1.1., the metrical organisation of As musical a composition is signalled in the time signature and measured by the bar lines. When poets set verse to music they must take this into account: "Stressed and unstressed syllables in 'sung' forms line the musical metre" most up in mandatory ways (Cureton 1992:46-47). According to this, the musical quality widely used to reinforce linguistic stress is metrical position. Primary counts, is, the first and third pulse in 4/4, and the first that with strong and weak beats in in 3/4, one are usually made to correspond to stressed syllables; secondary counts, that is, the second and the fourth 4/4, and the second and third beats in 3/4, beats in stressed or unstressed coincide with a hierarchical made to correspond either to syllables. Last, submetrics, that is, positions which do not count, correspond to unstressed In sum, text are and music are syllables. associated with metrical patterns, which are organisations of strong and weak elements. The condition association of text and music is that their metrical patterns superimposed be on be congruent. One must a controlled level of tension between stress patterns and poetic metrical schemes, which is important not only for poetical analysis explained in section 3.1., good poetry always shows and 4.4., tension In what that, in as implications for can be a the a device used follows, I shall focus a musical setting on purpose on - as was certain degree of tension as we - but will discuss in 4.2., 4.3. in vocal music. the analysis of text-setting in two languages observed in this dissertation, show different rhythmical characteristics, both ordinary speech and in of this the the other without conflicts of metrical positions and stress. Nevertheless, there is always also because of its on verse. According to the hypothesis stated at the beginning dissertation, the different speech rhythms of Spanish and English condition verse sound and prosody in these two languages and, in turn, the different approaches to rhythm in English and Spanish poetry may be 156 seen to have a correlation to of musical setting of the two cultures. As the differences in the processes I introduce the theoretical framework that will later songs. This framework is related to OT, explored for metre, 4.2.1. - on the guide the empirical analysis of specifically to the OT constraints hand, and for text-setting, one Morgan (1988) "El acento dislocado Janda and first step, - on the other. Thus, section pues cantado - castellano" Morgan and Janda (1989) "Musically conditioned stress shift in Spanish and revisited" although not really framed within OT, summarises the contents of the -, only two articles which, to 4.2.2. folk poetry?" -, section 4.2.3. section 4.2.4. -, English folk verse" folk verse" - observed in are as - - - verse and - Kiparsky (2006) "A modular metrics for analysis of OT metrical and grouping constraints song lines, while sections 4.2.5. -, 4.2.6. - - are - as Hayes (in press) Dell and Halle (in press) "Comparing text-setting in French and English songs" constraints and Hayes and MacEachern (1998) "Quatrain form in constraint conflict" "Text, tune and metrical form" sung Hayes and MacEachern (1996) "Are there lines in and section 4.2.8. devoted to the English "Text-setting musical knowledge, deal with text-setting in Spanish. Section my Hayes and Kaun (1996) "The role of phonological phrasing in - chanted verse" the more a - and 4.2.7. specifically focused mainly, but not exclusively, in English on songs. - Halle (forthcoming) the study of text-setting Section 4.2.9. summarises principle findings and conclusions that will guide the empirical analyses in the following sections. 4.2. The The to text-setting problem in the existing literature study of verse and music, the evolution of was as well as of their interaction in an interdisciplinary methodological apparatus which used tools borrowed from music in order to analyse speech prosody. On the other hand, generative linguistics became the basis for some a one of the most groundbreaking theories (Bernstein 1976; Lerdahl and Jackendoff 1983), which in turn added fruitful theoretical and precursor to has contributed suprasegmental phonology. Metrical Phonology (Liberman 1979) the first to devise of tonal music song, methodological elements to the original, thus becoming OT (Prince and Smolensky 1993). Since the appearance of Prince and Smolensky's (1993) Optimality Theory: constraint interaction in Generative Grammar, OT has gradually gained ground as one of the major theoretical 157 frameworks in segmental phonology. More recently, it has expanded to cover the marginal field of suprasegmental phonology. In the last ten often important musical papers on the relationship between linguistic prosody, rhythm have been written by linguists such as years verse several prosody and Hayes and Kaun (1996), Hayes and MacEachern (1996, 1998), Hayes (in press), Halle and Dell (in press), Halle (forthcoming) and Kiparsky (2006). The cited characteristics. First, setting they all deal, to varying degrees, with issues related to text- differently, with the conflicts that arise when trying to align text and or, put Second, with the exception of Halle (forthcoming), which approaches the tune. matter from frame. their a purely generative perspective, all of them use OT as their theoretical Third, they are primarily concerned with folk verse and song as the object of empirical study and, with the exception of Dell and Halle (in press), which folk presents a comparative analysis between analyse folk verse and deal share several important papers song songs in English and in French, they all in English exclusively. The fact that most of these overwhelmingly with English verse and songs papers makes their theoretical claims powerful enough. Therefore, it is crucial to analyse materials in other biased and not languages in order to be able to check the validity of the general theoretical assumptions made by OT in the fields of metrics and text-setting. Fourth, they share the main working theory underlying text-setting studies, namely that, in vocal music, rhythmic structure three tiers of grouping, (ii) verse metre come into play: (i) linguistic prominence and and grouping and (iii) musical rhythm and metre. The questions which naturally arise from such relation between the three tiers of a theory are, among others, (i) what is the rhythmic structure? (ii) which constraints applied to text-setting in different languages? (iii) are are they ranked equally in different languages? An OT approach to text-setting clearly aims at examining whether the setting of verse to music to responds to language prosody, constraints this may are kind of underlying rhythmic constraints prosody and music, as well as a common checking whether those differently from language to language, and what implications have for the study of prosodic investigation. Vocal music contribute to discussed ranked verse some can, thus, deeper knowledge and understanding of issues related to the long- dichotomy between stress- and syllable-timed languages and, 158 more generally, the generally neglected consequences of rhythm in language variation and change (see Schliiter 2005). going Before which tend to is that of recur with the discussion, it is in every necessary to object text-setting, objects each with its - own structure, a as "a composite which linguistic object - text - and a tune". The relationship between those two objects is mediated by a concept how lines of clarify two concepts linguistic analysis of text-setting. The first key concept 'song', defined by Dell and Halle (in press) combines two musical on explored by Halle and Lerdahl (1993), which has to do with linguistic texts arranged in time against are a predetermined rhythmic pattern in sung verse. Text Tune weak, strong weak, strong beats syllables Table 27: As can Text-setting mediating between text and tune. be observed in Table 27, both text and tune share the characteristic of being arranged hierarchically into two main levels, namely and weak units alternate - and a grouping level bigger constituents. Text-setting operates metrical level where small units - on a grouping and tune grouping, on where strong are arranged into those two levels simultaneously, facilitating the interaction between text metre and tune metre, text - on the text-setting. In OT terms, a series of CON regulating how text metre and tune metre metrical have a hand, and the other. As pointed out by Hayes and Kaun (1996:10), the existence of a series of constraints (CON) is central to of one are can any OT study applied at the metrical level, thus interact to render a well-formed composite. This interaction will be different in different languages, and will direct impact how text groups on the ranking of CON at the grouping level, which regulate and tune groups are arranged to yield 159 a well-formed composite. In ranking of CON at the grouping level will have direct consequences on the turn, the ranking of metrical CON. following sections, I review the most important aspects in the above- In the paying particular attention to the set of metrical and grouping mentioned papers, CON which relevant for the are analysis of text-setting in different languages. By reviewing the most recent literature in the topic, I will be in the position to highlight both its values and its flaws in order to then be able to offer a more comprehensive approach to the matter at hand. Morgan (1988) "El acento dislocado 4.2.1. Janda and cantado pues - Morgan and Janda (1989) "Musically conditioned stress shift castellano" and in - Spanish revisited" study of text-setting from The a linguistic point of view started to be given serious eighties, after the publication of two consideration in the late Morgan (1988), and Morgan and Janda (1989). After research what on the an by Janda and papers exhaustive process of topic, Janda and Morgan (1988:151) list the first observations of they call "stress-shift" in Spanish music. In 1887, Eduardo de la Barra (1887:19, in Janda and Morgan 1988:151) mismatch in national a song, anthem. was the first to point to a stress-beat namely that of the preposition contra ('against') in the Chilean Almost a century later, the Real Academia Espanola's new grammatical sketch (1973:66(7), in Janda and Morgan 1988:151) briefly mentioned the mismatch their 1988 phenomenon in Spanish defined in that text article, Janda and Morgan try to develop analysis of such the song, a a as an "anomaly". In systematic, non-impressionistic phenomenon in order to determine the linguistic consequences of systematic mismatch between stresses and beats in Spanish music. They elaborate a fully comprehensive list of mismatches found in (1988:153), written from a variety of songs since the fifteenth century, original from every century twenty-two countries, and from a variety of musical styles. According to the authors (1988:160), there are two main consequences phonetic and phonological/morphosemantic and Morgan a consequences. Within the latter, Janda (1988:160) distinguish two major effects of musically-conditioned stress-shift. On the there is of stress-beat mismatches, namely one hand, stress-marked distinctions violation of the conditions on are levelled; on the other, syllable structure and stress proposed by 160 Harris (1983), according to whom antepenultimate stress is not supposed to happen in words whose penult contains a branching rhyme, which is possible in vocal music. Morgan (1988:161-168) put forward four hypotheses for the Janda and phenomenon: a) Null hypothesis: whether or not a language allows musical stress-shift is a parameter of linguistic variation which must be arbitrarily stipulated (1988:161). b) First hypothesis: the notions of'syllable-timing' and 'stress-timing' play a crucial role (1988:161-162). This cannot be fully proved, given that there is shift in stress-timed languages, too - for instance, Brazilian Portuguese, Catalan, Russian. c) Second hypothesis: the crucial consideration in the musical stress-shift in a presence or absence of language is the relative availability of certain word- types which facilitate or at least accommodate the matching of texts to tunes (1988:163). This refers, for instance, to the realisation Lusitanian Portuguese and French, or the or not of schwa in of final vowels in Italian. apocope d) Third hypothesis: the fact that stress is predictable, and thus recoverable, makes stress-shift stress minimal a possible device, pairs are much even in a language like Spanish, where more common frequency of polysyllabic words forces the than in English. Also, the appearance of stress-shift (1988:167-168). Morgan and Janda's (1989) in 1988 and, more paper revises and extends the observations made importantly, re-defines the role of speech prosody in musically- conditioned stress-shift. The authors (1989:274) make a slight change in their approach to the topic and assert the importance of the syllable (if not syllable-timing) in the process of stress-shifting in One of the most song. important differences between linguistic and musical stress is, according to Morgan and Janda (1989:277), the lack of cues for the latter. They allege that pitch, duration and loudness do not work that the only cue for stress in music, and is timing itself: "ordinarily unstressed syllables which coincide with the musical downbeat stressed as cues syllable has a are not perceived as being stressed as musical/rhythmic accent which is at least 161 long as as the normally pronounced". In spite of the observed plethora of mismatches in Spanish Morgan and (1989:282) point to the fact that "the correspondence of musical stress to Janda linguistic stress is demonstrably desirable linguistic order to secure even in Spanish, given both the other it and the rejection of certain deviant stresses by some speakers". A more important than integrity" (1989:282). The difference between the employed in of this has to do with the question "of whether [...] the correspondence of musical phrases to real syntactic units is accentual are (synaeresis and dieresis, for example) which processes further consequence on song, English and Spanish prosody seems to have an effect quantity and quality of mismatches allowed in both languages. As Morgan and (1989:284) observe, Janda In order to sing an English text successfully, it is important only that intralexical stress not clash with the metrical pattern whatever division consideration positions erasure subdivision of the [...] Monosyllables as necessary. tree structure with or of texts can melody is under be squeezed into strong In Spanish [...] even whole we are or weak able to ignore metrical within words. Text/tune of all lexical stresses and of the tune at matching begins proceeds mechanically by left-to- right, one-to-one association of syllables with notes of the tune. The conclusion drawn from these two papers has to do with the prosodic conditioning of text-setting. As already mentioned, the main rhythmic unit in Spanish syllable, while the main unit in English is stress. is the 4.2.2. Hayes and Kaun (1996) "The role of phonological phrasing in sung and chanted verse" Hayes and Kaun's (1996) to metrics and, more importantly for the fairly straightforwardly to The authors chanting a purposes of the link phonological structure paper, to text-setting. As the (1996:4) point out, "[t]he ideas of metricality and complexity authors from paper represents an attempt to are and chanted verse". (1996:2) start by stating the fact that rhythmic beats in singing or isochronous in nature. This statement is, nevertheless, quite inaccurate musical nature of sung carry over point of view, for two main reasons. On the one hand, the isochronic rhythmic beats in singing, just like in instrumental music, depends 162 on the musical Thus, for instance, saying that Gregorian chant shows and genre. era 'isochronous beats' would forms of contemporary world. On the other certainly be incorrect, and the art-music, well as would apply to certain certain styles of folk song across the hand, the authors do not really explain what they mean by By this, 'chanted verse'. as same we assume that they are referring to a type of flat, inexpressive declamation of a line in order to make its positions fall onto the metrical positions of an abstract grid. In this respect, chanted verse different in nature. The former would be the result of deleting from sung verse - accompaniment, and and sung verse are everything but the beat that is, neglecting its melodic contour, expressive devices, so on, while the latter would be the result of the interaction of all the afore-mentioned musical and textual elements. What the authors this paper deliverance of a line By 'metricality' they corresponds to Hayes and Kaun's (1996) metricality of are setting a - paper, a as a text formed any in a specific way is a of human choice, which implies that "the a spontaneous when singers improvise text-settings, they prove that every a artistic decision" (1996:5). are selecting from the well- Hayes and Kaun make the strong universal ability. The argument used by the authors individual has text-setting abilities is that "music implies that the remaining verses must verse of a multi-verse song", be set by the singer at the moment of performance. This is arguable, since the skill of a singer to set depends purposes pointed out, they only accept binary settings -, they acknowledge publishers often provide the text-setting of only the first tune specific possibility of sharp variation in the settings yielded by assumption that text-setting is which a is always binary. In other words, the degree of options in that specific tradition. (1996:5) to the extent to which pre-designed grid, which, for the variability in text-setting usually represents In this respect, mean specific text-setting is its degree of well-formedness. Although the eliminating the informants that testing in is the degree of metricality of specific instances of predominantly chanted, rather that sung, verse. authors are on many more variables than an a text to a particular innate ability comparable to that explored by linguists for natural language. As mentioned above, Hayes and Kaun study text-setting as a phenomenon independent of melody, leaving the issue of how the tune influences text-setting for further research. As already pointed out, it seems 163 misleading to think that text-setting at two occurs complete if we isolate are namely the metric level and the melodic pitch interact in such and metre distinct levels, a way I think that one. that the analysis of text-setting will not be of these two components. In fact, interval and pitch itself one important for the metre, and the melodic contour of songs is decisive sometimes the well-formedness of a setting. As Tovey (1979:212) observes, in order to judge even strength of modern musical rhythms, the with the full muscular strong beats of the bar constitute only one form of accent, which may be easily counteracted by the length of harmonic have intuitions about they give hypothesis that native speakers of a specific folk tradition text-setting, Hayes and Kaun choose some piece of research are a group a of informants to ethnomusicologists of that time, such to observe that the they authors as checking. The informants used in - probably middle class Karpeles, who imposed a a group on an of Cecil Sharp. In this respect, I would also like acknowledge the biased character of the original grid - collection of English and American folk collected and transcribed by ethnomusicologists, such were confronted we are collected at the beginning of the twentieth century by songs since are of twentieth-century exposed to lines extracted from Americans a group lines in order to be chanted. As readers, with the doubt of whose intuitions the authors this by its height, by its colour, and by incidents in the accompaniment. In order to test their whom a note, as songs, Sharp and otherwise 'shapeless', i.e. not necessarily metrical, chain of pitches. It is a well-known fact that traditional singers do not always - in fact, most of the time they do not - sing metrically - hence observation in relation to my previous Hayes and Kaun's statement that all music is organised isochronously. If Hayes and Kaun accept that the original transcriptions of the which they use respond to Sharp's own intuitions when he transcribed them, which clearly differed from the original singers' intuitions, I wonder whether certain about the song adding, for instance, piano accompaniments to sung an in otherwise an one can be validity of the authors' strong hypotheses about the native nature of text-setting abilities. Sharp adapted folk - songs purely vocal musical for a salon, middle-class audience, - and the consequent metrical tightening genre, which would most probably have been ad libitum fashion. Evidence for this 164 can be found in the various sound archives collected by people such as Alan Lomax and Peter Kennedy in the second half of the twentieth century. As pointed above, another important observation in relation to Hayes and experiments has to do with the requirement that all informants set the lines in Kaun's question to "ternary a binary rhythm. Real intuitions purpose of the binary metrical grid a paper. thus clearly biased, to the extent that gently corrected" (1996:8-9), that is, only those informants scanners were who set the lines to are were The fact that the authors had to 'correct' contradicts the their assertion that "no obvious education" (1996:8) were is, whereas some of the non gender a binary rhythm, as or a musical musically opposed to a ternary musically-educated informants would have found it problematic to differentiate between succeed in their effects of informants some found in the text-setting patterns. Surely, educated informant would have known what one, taken into consideration for the a binary and a ternary rhythm and, therefore, to binary settings. (167) and (168) illustrate the difference between a binary and a ternary rhythm. (167) Binary rhythm: the odd beats of each binary unit even are stronger beats; the first beat of the highest unit is always the strongest than the one. 2/4: X X X 4/4: X X X X X X X (168) Ternary rhythm: the first beat of every the other two are weak. 165 group of three is strong, while 3/4: X X X X X conclusion, In intuitions. The authors informant is needing to resort to If we seem to really intuitive informants to prove a paper is checking innate just too affected. Also, they or a specific seem to use the kind of field work. used, the settings yielded by the informants are that the reader visualises the text-setting structure, the weight syllables, the stress pattern and the phonological structure (P-structure, of stressed a 1984), which comprises the hierarchisation of the borrowed from Selkirk term that this theory which they have previously articulated without actually any way a say judging whether the chant given by be refer to the notation notated in such really cannot we components into words, clitic groups, phonological phrases and intonational phrases (see Hayes and Kaun's example in 1996:8). There number of key are a ideas that the authors explore in order to judge the degree of metricality of the settings yielded by their informants. The first central idea in relation to the well- full and truncated with are ill-formedness of text-settings explored further in settings or a later - text-settings is the distinction between truncation is related to paper -, a concept by Hayes and MacEachern (1998). Truncated those in which grids do not have all their strong metrical positions filled syllables, which makes them salient. There settings, of which the two that (i) SALIENCY59 A setting with so that the Kaun no are most are four main classes of tmncated often used in English song are the following: syllables between the third and fourth strong positions, syllable in the third position is sung quite long. Hayes and (1996:9) call this the 'Green O' truncation and exemplify it as follows: 59 OT has adopted the typographical convention of capitalising the names of constraints. I text-setting from an OT perspective. this convention in the sections that deal with 166 am adopting (169) XX XX XX XX XXX A- XX xxxxx xxxx XX leaves the mong XX XX so O green A setting with the fourth strong position unfilled, in such (ii) line-final syllable is substantial pause. '3') lines, with a detectable in the sung very long, or a way that the alternatively, followed by a Hayes and Kaun (1996:10) call these 'three beat' (or reminder to the reader that the silent fourth beat remains timing: (170) XXX X XX XX XX XXX xxxx xxxx sowed I the seeds of XX xxxxx love Hayes and Kaun associate the non-filling of metrical positions in the grid with the concept of 'extrametricality', thoroughly explored in verse prosody studies. Nevertheless, it is essential to bear in mind that this concept, which depends theories of prosody, cannot be equated - as it is in this paper - on to musical feminine endings, which have the shape S W, just like extrametrical structures, but where the W beat is completely metrical. The second nuclear idea in the paper is related to the MAXBEAT CON and, more specifically, with MATCHSTRESS (1996:10), whereby the rises and falls of stress within the line studies of are matched to the rises and falls of the metrical pattern. In all text-setting in English, the correspondence between linguistic stresses and musical strong beats is absolutely essential when it formedness and, therefore, the reader with a comes to judge the well- quality of a specific text-setting. Nevertheless, knowledge of folk songs can predict, there 167 are many as any exceptions to that correspondence between stresses and beats, something of which Hayes one-to-one and Kaun I will are aware. As some of these exceptions in the next pages. complement to the idea of truncation in combination with of sort a explore MATCHSTRESS, Hayes and Kaun (1996:11) adapt the so-called Zero Provision rule - used in musical context a when a metrical by Lerdahl and Jackendoff (1983) - as follows: position is unfilled, or equivalently - - phonemic continuation of the preceding syllable, treat it by filled by the as being filled level that is weaker than that of a stressless overt syllable. a stress This constitutes a systematic explanation of the fact that any unrealised position must obviously belong in the weakest metrical level in the grid. This means filled that position, The no unfilled position as we can see novelty of this can a stronger beat than a preceding in examples (169) and (170) above. paper is the overt link that Hayes and Kaun (1996:13) prosody, draw between metrics and musical structure, on the other. In fact, metrics of the be assigned on the one hand, and language P- they clearly state that P-structure influences the language in question. They summarise this hypothesis by observing that in all metrical traditions there is a requirement that line boundaries coincide with high-level breaks in P-structure, that is, the correspondence between speech, and tune grouping and musical - phrases put more precisely, the correspondence between P-phrases, lines - must be as high as possible. As will be observed in Hayes and (1996, 1998), Dell and Halle (in press), Halle (forthcoming) and Hayes MacEachern (in press), the concept of Tine' is central to requirement can vary study of text-setting. While this song, high. With regard to the interaction between P- rules refer to P-structure in phonology do", and that, a just the priori, the principle known applies to text-setting processes, too. As its as same ways tends to be free at the verse. that rules of "beginnings free, endings name explicitly indicates, this principle states that the correspondence of linguistic material to written the agreement (grouping) and MATCHSTRESS (metre), Hayes and Kaun (1996:14-15) state that "metrical strict" any in strictness, they also observe that, in folk between P-structure and metrics is structure verse a metrical pattern beginnings of units, and strict at the ends, something tested for However, its validity in the field of song needs to be tested. 168 Although Hayes and Kaun deal exclusively with metrical issues, leaving aside all other variables, they mention durational effects in text-setting, where phonemic duration is said to be determined by weight level stress - a phonological variable particularly the stress distinctions that determine vowel reduction segment identity within - - lower vowels tend to be longer In P-structure. this foundational rule schema for respect, - -, and location of the syllable Hayes and Kaun (1996:16) establish a text-setting, called 'Syllable Duration Rule'. Syllable Duration Rule: reflect the natural phonetic durations of syllables in the number of metrical beats they receive. way: the appears the According to this, P-structure relates to syllable duration in the following higher-ranking the right edge in P-structure at which more an given syllable lengthened. In this respect, "a high level P-structure right edge (such it is that of a I-phrase) should induce substantial lengthening and induce a on as the preceding syllable, strong preference for that syllable to receive multiple beats" (1996:17). Hayes and Kaun base this idea on previous studies by Ladd and Campbell (1991) and Wightman et al. (1992). Summarising, Hayes and Kaun (1996:24) defend the view that P-structure is a principal determinant of phonetic syllable duration. Owing to the Syllable Duration Rule, syllables in English text-setting matched to appropriate numbers of beats. This since its total validity would imply that value note in music, just as no no seems to are characteristically be too strong short vowel would long vowel would be set to ever a a statement, be set to a long- short-value note in music, something which is clearly not the case. Let us have a look at one of the numerous exceptions found in Kennedy (1984:348): (171) The observe that In (171) to correspond to we cuc-koois a a a pret-ty bird,he sings as he light syllable like the, with crotchet - a quaver - schwa as its nucleus, is made the equivalent of two metrical positions heavy syllable like -koo in cuckoo, with correspond to a flies a long /u:/ as -, while a its nucleus, is made to the equivalent of one metrical position. On the other hand, 169 it is generally true that the ends of musical phrases tend to show longer note-values other position - in this example, than any flies, is assigned to a we can see that the last syllable of the line, minim, which is the longest note-value found in the phrase. Hayes and Kaun analyse MATCHSTRESS. All of those defined in Dell and Halle a number of settings which represent exceptions are an exception to related to the concept of 'mismatch' as (in press), which Hayes and Kaun (1996:36) call 'lexical inversions'. A lexical inversion is defined as "a word internal /oo/ sequence that is mismatched against the metre" (1996:36). As mentioned above, the focus of Hayes and Kaun's study is "quasi-evenly" bracketed binary structures, such iambic pentameters. in the third mentioned rule of are or fourth foot" line-initially, and of the remainder, most (1996:36). This seems to confirm the afore¬ "beginnings free, endings strict". Nevertheless, in folk happens at the instances where the lexical inversion musical those of According to their observations, "the great majority of lexical inversions in iambic pentameters occur occur as phrase, for this matter. Let us have a very end of song a line there - or look at (172), recorded in Kennedy (1984:653): (172) t=3j=j■xrJ i1 r O As we can see in an example of a who-will mar-ry my fair-la-dy?0 (172), the last lexical word of the line in question, lady, constitutes lexical mismatch, since its stressed syllable, la-, falls single metrical position, while the unstressed syllable -dy falls and it is made to duration seems The Kaun occurs to correspond to two metrical positions. In this be reflected on a strong case, weak, position, neither stress nor Hayes and Kaun predict in their model. explanation to the lexical mismatch phenomenon is linked in Hayes and (1996:42) to the afore-mentioned concept of 'truncation', which typically at the ends of lines. As syllables do not make positions. As the as on a a sense, already pointed out, in since there is result, unfilled positions an are surplus of positions gives the text setter musical setting, extrametrical isochronous, constant flow of metrical freely tolerated at the ends of lines, and some 170 a freedom not available to the spoken verse poet. Parallel to this, there seems to be a higher degree of mismatch freedom at the ends of lines than at the theories of verse prosody beginnings. All these facts make the are not same as conclude that us theories of musical prosody. A macro- theory which explains the interactions between the two is clearly needed. As their last sung verse observation, Hayes and Kaun (1996:43) insist, is performed in isochronous rhythm, with the following a) Substantial constraints imposed in are once consequences: requiring sung verse again, that match-up of a syllable durations with beat count. Non-isochronous spoken natural verse permits great flexibility of syllable durations. b) The metrical pattern for sung c) There is no such thing Since I have extrametrical syllable in as an already manifested conclusions, I do not consider it would like to refer to an strict indeed interesting the disagreement with some of the above observations. However, I a study of the from southern Arizona reveals that in certain any case, a edges of lines sung verse. by Fitzgerald (1998), where paper beginnings of lines (1998:31). In related to the my necessary to repeat my metre of Tohono O'Odham songs musical traditions the is in principle continuous. verse are not lax at all but, on the contrary, revision of typological claims based seems to be necessary, as well as a on criteria separation between requirements of purely poetic lines and those of lines set to music. Hayes and Kaun (1996:43) end sung verse most up summarising their findings in OT terms: in candidate lexical inversions settings that slide the lexical inversion are into over a said to be beaten out by rival text- matching position. Although the originality of this approach is unquestionable, the fact that and chanted a very verse are regarded equal points to as clearly cultural phenomenon, such whatsoever in this paper, certain something which linguistic circles. But there is would benefit as immensely from a no folk can an song. excessively positivistic view of There is no cultural allusion actually be academically beneficial in doubt that broader, sung more some of the phenomena observed interdisciplinary perspective. For instance, what Hayes and Kaun call 'lexical inversion' is in fact a stylistic marker of English folk song, and, therefore, an This is indeed not universal and not a folk song stylistic marker. For a idiomatic way of building allowed mismatches. specific to English music in general; it is simply linguist/musician, escaping the cultural side of any 171 study folk on folk song, illustration to this text-setting in general, is misleading. As verse, or point, Hayes and Kaun the term 'Scotch snap' for use type of mismatch where an S position is displaced of a Nevertheless, the syncope. Scotch snap - a an specific in musical terms, the realisation, of that term is musically inaccurate, since use is too quick to sing anything to it, as it is a a dance figure. (173) A more accurate would be 'Lombard name sometimes used as a century opera and, therefore, - This paper what a synonym offers an text-setting really is, for Scotch snap - a syncope although this term is figure used in eighteenth- vocal figure. a original, although not scientifically convincing, view of as well which mechanisms as work when singers of are at specific music tradition set lyrics into music. 4.2.3. Hayes and MacEachern (1996) "Are there lines in folk poetry?" Hayes and MacEachern's (1996) metres and stanza structures of their nuclear paper is "part of but also of folk verse. line in folk verse, described in In a program to study the as part as a structural element not of their conscious intentions lines typographically as units on the page - - the existence and structural importance of the In order to prove a corpus of 1028 Southern Appalachian folk later paper by Hayes and MacEachern (1998). methodological terms, this analysis of the paper songs resembles Hayes and Kaun (1996). The is based on the metrical grids proposed as rhythmic analysis by Liberman (1979) and Lerdahl and Jackendoff (1983: chaps. 2-4). For each accommodate the notated lines use the authors analyse method used in their basis for research hypothesis about the existence of the line the evidence for this is that poets arrange songs a English folk verse" (1996:1). The authors start with only of art-verse, where literate poets a -, rhythm' song in the rhythm of the along filled positions, of truncation remains corpus, song. as seen Hayes and Kaun establish Once again, silent beats a grid which are can observed and in examples (169) and (170), and the concept essential since "line-final truncation specified levels of metrical constituency" (1996:2). 172 serves to highlight hypothesis for the existence of lines put forward by Hayes and The (1996:3) is the following: MacEachern We claiming that lines exist; that is, that they are structural units of the system. possesses its own We are authentic are also claiming that each line distinct stretch of grid, and that a song-specific choice is made for what this stretch is. None of these claims is obviously true, the grid is essentially continuous. because in the audible form of a song What the authors intend to show is that the virtual chopping of the quoted continuous grid is not arbitrary but follows specific rules. Hayes and MacEachern's (1996:3-4) hypothesis relies large breaks in the P-structure". In this to a way, on each intuitive line of a the "presence of song corresponds phonologically unified string. According to the authors (1996:4), if almost any song one takes and aligns it with the correct "infinite" rhythmic grid, it will be possible to cut the grid into perfectly equal pieces such that the beginning and the end of each verse. piece will coincide with Each line of each each line in observe that case syllables, usually bounded by that line's associated musician's perspective, would be easily explained by saying that stanza corresponds to the repetition of the so grids tend to be filled-up to the the grid starts with even conclude that lead to same musical phrase. As seems to if this means to very be a some edge at the beginning of lines, in weak position, be it with stressless a strong "Shady Grove", which or stressed position. After analysing the four verses are arranged to start on a strong metrical drop certain syllables at the beginning of the line, they "[1]ine-initial function words are dropped in all cases where it would unintelligibility, probably because English phrases characteristically start with stressless function words" common a that the grid starts with of the traditional position as a separate for the beginning of the line actually to abut the line boundary". They pressure not line is a observation, the authors (1996:4) point out that "there second which a a large break in the P-structure of the associated quatrain could plausibly be thought of Intonational Phrase. The text of grid, which, from a dropping of the (1996:6-7). An example of this would be the pronoun I at the beginning of lines. The opposite phenomenon, that is, the insertion of an extra weak syllable to fit the grid, can also take place. Such is the case of the archaic verbal prefix 173 a-, which is extensively used in folk song in order to fill certain beats at the end of bars. However, Hayes and Kaun (1996), there is start with the these syllable fall on according to a line - line" defined a any case, "a sum, organised song particular grid will typically avoid placing the syllables of a linguistic by phonological phrasing more, there is - outside the grid a strong allocated to that space tendency for the linguistic line to point precisely coinciding with the beginning of the line grid. Again, musicians may not be so sure of anacruses found not in grid. In redistribution of the positions to make the strong a the first strong position in the grid. In (1996:7). What is begin at observed in small minority of lines that might conceivably a strong position, thus failing to fill the left edge of the exceptions always allow for as about this statement since it overlooks the high number only in folk song but also in art I will song. come back to this 4.2.6., devoted to Dell and Halle (in press). As mentioned in be more Hayes and Kaun (1996), the right edge of the grid seems to problematic than the left edge. Hayes and MacEachern (1996:7) wonder why there is not similar pressure for lines to end at the ending of the line grid. The explanation given for the lax nature of the right edge of the grid in this (1996:7) goes back to Hayes and Kaun (1996), who tends to match the number of beats duration of a assigned to a argue that, since paper sung verse syllable to the natural phonetic syllable and line-final syllables tend to be phonetically long, it will be logical to place line-final syllables not too close to line ends. Nevertheless, this statement seems to be an overgeneralisation since there are longer syllable duration does not necessarily correspond to and vice versa. Let us have a look at plenty of examples where a higher number of beats, (171) again, repeated below as (174): (174) P'i J i ^J r1J "J i J 1JJ r i? i The cuc-koois a pret-ty bird,he sings as he flies Hayes and MacEachern complement their explanation with the a short allusion to theory of "rhythmic cadences", developed further in Hayes and MacEachern (1998), which allocates the end of the line grid to silent beats, a hint that serves to highlight the constituency of the quatrain. Summarising, the length of line-final 174 syllables, together with the action of rhythmic cadences, would tend to block the filling of the right edge of the line. For these reasons, Hayes and MacEachern (1996:7) conclude that "we cannot expect the ends of lines to abut the ends of their, grids". Summarising the first part of the paper, the authors (1996:10) observe, again, that the distinct behaviour of songs using grids from the both in their characteristic line for the line as a beginnings tacit unit of folk turn out to MacEachern be a statement a reality composition. Secondly, the method of verse seems to be the most precise for reinforced by the fact that rhyming positions overwhelmingly line-final. Actually, as observed in Hayes and (1998), the organisation of silent beats at the ends of lines follow certain fixed patterns. MacEachern infinite pattern, possible line endings, indicates an inspecting the alignments of grids and P-structure locating line boundaries, same once seems to In spite of the strength of their hypothesis, Hayes and (1996:10) recognise that, regards line ends, "[n]either of these as diagnostics locates the line boundary with the precision obtained from inspecting line beginnings, but they virtually never diagnose In the second major section of their a different parse into lines". paper, Hayes and MacEachern (1996:10) analyse lines in the metrical hierarchy, where they "form only one layer in a quite deep constituent structure that embraces whole stanzas". They point out that the vast majority of songs analysed in their study could be assigned layers are a structure where all the binary: [quatrain [couplet [line [hemistich [dipod [foot [beat b] f [b b]] d] h [d [f [b b] f [b b]] d [f [b b] f [b b]]]] 1 [h h]] c [1 [h [d [f [b b] f [b b]] d [f [b b] f [b b]]] h] 1 [h [d [f [b b] f [b b]] d [f [b b] f [b b]]] h [d [f [b b] f [b b]] d [f [b b] f [b b]]]]]]] Hayes and MacEachern point out that the evidence used to argue structure coincides with the evidence for the existence of lines: metrical for such a "agreement of constituency with phonological constituency, the patterning of silent beats at the ends of constituents, and rhyming" (1996:11). The hierarchical structure avoid the "infinite grid" seen element of the theory of metrics, above allows us and makes it possible to attribute the similarity of line grids to "shared constituent structure" to (1996:11). 175 as an Following these observations, Hayes and MacEachern (1996:13) accept the existence of exceptions while certain songs associated with those clear-cut statements. They observe that, show statistically dominant partitioning, there a can be one or two places in which P-structure and line boundaries disagree. The authors argue that the exceptional involve what they call the "theft" of positions by cases linguistic line either from the preceding that such cases are their statistical corpus in their songs the following metrical line. Their view that they corpus only 107, that is, 10.4%, involved those positions. On the other hand, the boundaries fluid to cases of songs in which the line defy the establishment of a grid norm for line division really rare. Another argument used to qualify the existence of exceptions is that were there is never a limitation takes virtually us as Among the 1028 Southern use. "thefts" of metrical were so specific exceptional and non-threatening to their theory is supported by rarity in the Appalachian folk or a observe this how much linguistic material than the last strong more never on reaches even may be stolen: "Leftward theft beat of the preceding line, and rightward theft the first strong beat of the following line" (1996:14). Let phenomenon in (175), taken from Kennedy (1984:503): (175) There was In j an old IJ man Jjljijljlg and he lived in the east And his (175), the syllables And his, which belong in the second line of the lyrics, are belong in the last bar of the first musical phrase. However, we "stolen" and made to could argue that, given that the first line begins with been stolen from any beginning of the beat - or, song so positions' basic types a case anacrusis, which cannot have the grid is structured from the very that the first syllable(s) of each line corresponds to the last exceptionally, to the last two beats - of each 3/4 bar. The authors line, other position, in this an can (1996:15-16) be characterised of 'theft'. The first go on to more one explain that the patterning of 'theft of precisely. According to them, there are two is the theft of a strong position from the preceding pattern that is consistent throughout all the lines of the song in which it occurs. The second type is the theft of a weak 176 or medium-strength position from the following line; this type is sporadic, within the even song in which it occurs, primary function is to accommodate specifically problematic texts, such lines with feminine can handle - endings such is the lines with or case of more (175) above. As syllables of this stanza we can see made to follow are linguistic initial pretonic syllables than the grid in (175), the beginning of the second line of the first stanza does not coincide with the The first two as and its on beginning of a bar. the last beat of the previous bar. In Hayes and MacEachern's terms, those two syllables, both weak, are 'stolen'. question of why theft of strong positions only To the occur in the leftward direction, Hayes and MacEachern (1996:16-17) argue that, in English at least, "it is only in the leftward direction that the positions are there to be stolen". This idea is linked to that of truncation, whereby lines, with the that syllables from the following line those empty the first consequence a number of beats positions. On the other hand, position of grids empty, so as left unfilled at the end of are can be used to fill already observed, English seldom leaves theft from this position sounds less natural. Concluding this section, Hayes and MacEachern (1996:17) repeat their idea that "characteristic correspondences between linguistic phrasing breaks and particular metrical locations indicate that the line is metrical constituent of the sung-verse an authentic, well-defined grid". They acknowledge the existence of counterexamples to their claim, involving theft of positions in both directions. However, given that interline thefts are statistically unusual compared to the mainstream cases conclude that and that the number of beats that may be stolen is limited, they they have strong evidence in support for the line as a neatly bracketed structural unit. The very last section of the line levels", which refer to the more than one metrical level" paper is devoted to what they call "ambiguous possibility to parse higher constituents into lines "at (1996:17). Hayes and MacEachern claim that the basic principles of metrical form in English folk verse are relatively abstract, and involve "requirements for strict identity of constituents of equal rank, isochronism of rhythmic beats at all levels, and branching" (1996:19). At such a a strong preference for binary and ternary level of abstraction, they accept that there is 177 no such thing 'line' as a as a formal primitive. Nevertheless, there certain intuitions about following (1996:20): what the lines are, such as the a) The line level are seems to be the level at which the 'echoing' of metrical bracketing with P-structure bracketing is the clearest - an idea borrowed from Attridge (1982:107). b) There lines. seems to be Restricting with 16 terminal for every a rough maximum for the internal metrical complexity of our attention to binary metres, we find that positions typically suffices to provide a a metrical grid rhythmic location syllable in the line. c) Stanzas tend to be long enough to permit at least four Tines'. Hayes and MacEachern's (1996:20) corpus-based conclusion is that the lines diagnosed in their paper "serve important additional structural purposes in defining rhyme schemes and patterns of line-final empty positions". Finally, they point at the addition of two difficulties to their where P-structure boundaries more than one level of the study, namely a discussion of the unusual disagree with line boundaries, and the binary hierarchy could serve as where cases the line level. As cases a major conclusion, they state that neither of those two difficulties threatens their results. 4.2.4. Hayes and MacEachern (1998) "Quatrain form in English folk verse" Hayes and MacEachern's (1998) paper constitutes an attempt, loosely framed within OT, to find out exactly what rules regulate the patterns of truncation at the end of lines in English folk verse. The concept of 'truncation', central to defined as already observed, was also Hayes and Kaun (1996) and Hayes and MacEachern (1996). Truncation is by Hayes and MacEachern (1998:473) as the "nonfilling of metrical positions at the end of lines". According to the authors, there truncation patterns are for English folk quatrains. The function of truncation, 26 possible as observed by Hayes and MacEachern (1996) is to render certain layers salient in the natural constituency of the quatrain, namely the line, the couplet Given that it is not or the quatrain as a whole. possible to render those three constituents salient at the same time, the constraints that regulate saliency will be in conflict amongst themselves, well as with metrical filled with conflicts CON, which require, by definition, that all metrical positions be syllables and stresses. The outputs of the multiple among as constraints - the solving 178 process ways of solving these has to do with different prioritisations of the constraints themselves truncated is the - group of 26 well-formed quatrain types. The authors start able to by stating the problem. They observe that native speakers of the well- ill-formedness of a given quatrain. Truncation English are poses a problem with regard to native intuitions. In assess or a nursery rhyme quatrain, to quote one of the most widely used verse quatrains in English, each line is felt to have four metrical isochronous beats, but often the fourth beat is 'silent', that is, it is "observed in the timing of the recitation but not aligned with the other hand, certain beat assume an on innate ability (1966), which concluded that children's not so languages resemble immediately judged ill- brief account of previous work done one verse types. from unrelated, geographically verse types another in a very They quote Burling striking way, something which is noticeable in art-verse. In their verse, a topic of native intuitions with regard to folk distant are the part of native speakers to judge the metricality of on Hayes and MacEachern (1998:474) give the syllable" (1998:473). On by native speakers (1998:474). As in Hayes and Kaun (1996), the authors formed lines. combinations (*3434; *3444) a analysis of the quatrain as the main compositional unit of English folk Hayes and MacEachern (1998:475) emphasise the idea, already observed in Hayes and MacEachern (1996), that the fundamental basis of folk hierarchy. This rather a means that a quatrain is not simply a sequence is verse a binary of four lines, but pair of pairs, with the structure [quatrain[couplet[line line]][couplet[line line]]]. This pair-of-pairs structure is reinforced by rhyme. To represent the structure of the lines scanned in their study, Hayes and MacEachern use, once again, the grid, which not only represents the alignment of syllables and beats, but also the saliency of certain beats with respect to others. In order to distinguish the different types of quatrains, the authors observe that the ends of lines work for each of them. MacEachern (1998:476) define the use of 'rhythmic cadences', from the musical tradition. In musical terms, a rhythmic or For this purpose, 'cadence' is Hayes and a term an borrowed ending within musical form. A rhythmic cadence is thus "the characteristic grid placement of the final syllable or two of the line" (1998:476). The authors list the 179 most rhythmic cadences in English folk common in Hayes and Kaun verse, as seen (1996) and Hayes and MacEachern (1996): a) 3, where there are no syllables after the third strong position of the line. It is- the first truncated cadence quoted in Hayes and MacEachern (1996) b) 'Green O' (G), which fills the third and fourth strong positions, with no syllables between them. It is the second truncated cadence quoted in Hayes and MacEachern (1996) c) 4, where the fourth strong position is filled, but not in the fashion of G, that is, there has to be at least one syllable between the third and fourth strong positions d) 3-feminine (3f), where the third is the last strong position filled, but it is by an extra weak position. followed Cadences, since as pointed out in the paper (1998:478), they mark the end of constituents. In in every stanza, songs, essential element in these cadences that is, the cadence tends not to shift from for instance, if the first stanza ends off with the same cadence to mark their end songs a one are song usually repeated type to another. So, '4', the rest of the stanzas will present points. empirical basis of this The main English folk are an paper is the examination of a large compiled in the early twentieth century in rural corpus areas of of the English-speaking world: rural England (Karpeles 1974), Newfoundland (Karpeles 1970) and the Southern Appalachians (Karpeles and Ritchie). The materials from the Appalachians, which had already been used in Hayes and MacEachern (1996), once again, the reported in the ones purpose of the investigation paper. In order to and MacEachern 'common most-carefully analysed for the were, classify the quatrains found in the analysed (1998:478-483) use a corpus mixed taxonomy. For of songs, Hayes some metres, like the metre', they use the taxonomy inherited from the study of Church Hymns, while for others they simply make up names that fit their purpose: a) 4343 metre, called 'common metre' according to the traditional taxonomy, referred to as 'couple-marking' quatrain type in this paper b) 4443 metre, called by the authors 'quatrain-marking' type 180 c) 3333 (GGGG, 3Df3f3f) metre, referred to known as 'short metre' in Church as 'line-marking' construction, Hymns studies d) 4444 metre, called 'metrically replete' quatrain in this known as paper, traditionally, 'long metre' e) 3343 (GG4G, 33G3 > 44 ('semiquatrains'), 333f3) metre, called 'short metre' 'long-last construction'. or Together with these clearly-defined metres, Hayes and MacEachern find two other quatrain categories, namely f) Quatrains with 3 different cadences: 3f343, G343, 3f3G3 g) Quatrain types with free variation: F(or G)3F3. The classification of the corpus to four groups of CON. The first relates to the concepts can ends of group refers to CON on their analysis in relation on saliency domains, and of cadentiality and saliency. The idea of a rhythmic cadence applied to speech be of quatrains relies well as as to music, meaning a kind of slowing down at the phrasal units. This is what phoneticians call 'final lengthening', explored in Hayes and MacEachern (1996) in the field of folk song. an idea Turning this tendency around, Hayes and MacEachern (1998:483) posit that final lengthening is cue for phrasehood, that is, rhythmic cadences group are a kind of constituency marker. The hypothesis is that "the ranked according to their ability to induce the perception of a ending (line, couplet, quatrain)" (1998:483). According to this, the cadentiality ranking is represented as follows, from most to least cadential: 3 » 3f » G » 4. SALIENCY has to do with two main conditions final a rhythmic cadence of that constituent is more cadences, and (b) all of its non-final cadences salient quatrains in the analysed corpus are (1998:485), namely (a) the cadential than all of its non-final uniform. Under this definition, the would be 444G, 4443f, 4443, GGG3f, GGG3, 3f3f3f3f. The salient couplets would be 4G, 43 f, 43, G3f, G3, 3f3. As for lines, Hayes and MacEachern (1998:485) see them all as "(vacuously) salient". The authors • analysis of the corpus according to this type of constraint leads the (1998:486) to make the following observations: A major goal of the metrical system of English folk the verse is to render salient major structural units: line, couplet, quatrain. This is done by placing the final syllables of lines in appropriate arrangements of cadentiality. 181 The most • heavily cadential line endings truncated: to serve the cadential function, are also the most metrically they must fail to fill quite a few positions at the end of the grid. The second (in press), defined major CON is PARALLELISM, also mentioned in Dell and Halle as follows: two couplets with the What this entails is that "the cadences same structure form a quatrain. ending the units of the maximal analysis of a quatrain must be identical" (1998:488), a principle that is never violated in well- quatrains. Parallelism is, in turn, induced and reinforced by rhyme in that the formed cadences of the maximal analysis usually rhyme with each other. The third main group subdivided into the of CON refer to long-last constructions, and it CON LONG-LAST, PREFER PARTIAL can be LONG-LAST COHESIVENESS and TOTAL LONG-LAST COHESIVENESS. The LONG-LAST implies that, in CON a sequence longest member should go of of unequal length groups last. A quatrain is a - like 3434 -, the long-last construction if (a) its second couplet is salient by the all-or-nothing definition, and (b) both its first and second lines are more salient important remark criterion will be account was (by the gradient definition) than the third line (1998:489). An seems in order here. Quatrains satisfying the LONG-LAST experienced as line + line + couplet. This is important if we take into that, in principle, quatrains of the type 3343 would not satisfy it if every line experienced as such. On the contrary, what Hayes and MacEachern suggest is that the first two lines (3-3) will be experienced will be taken to form a of the single constituent, a as such, while the two last couplet. In this way, ones (4-3) the last constituent quatrain will indeed be longer than the first. At first glance, it is problematic to accept this change in the constituents' length and shape in order to fit some theoretical statement. LONG-LAST is directly linked to the degree of cohesiveness of a given quatrain. In Hayes and MacEachern's (1998:489) words, "a couplet, if salient, is cohesive inversely to the saliency of its first line". According to this principle, 43, 43f and 4G are fully cohesive couplets, G3 and G3f are less cohesive, and finally 3f3 is the least cohesive one. 182 The fourth major group of CON are metrical CON, which have to do with the requirement that all positions in the grid be filled with syllables and stresses. This group can be subdivided into three main CON (1998:492): a) Fill strong positions (FILLSTRONG): fill the strongest positions in the line b) Avoid lapse (*LAPSE): avoid sequences in which no syllable is placed in the of the four strongest positions in the line interval between any two employ G in feminine endings, 4 c) Match stress (MATCHSTRESS): elsewhere. The last main CON dealt with CORRESPONDENCE (1998:492), which states that "in domains must be invariant As mentioned study are by Hayes and MacEachern is that of STANZA across a song, the set of salient stanzas". above, the theoretical tools and methodology employed in this those of OT. In order to give ranked, and the quatrains in the OT account of the analysis, the CON an corpus are are analysed in accordance to the different possible rankings. Last, the ideal candidate in each ranking is chosen, and finally, the absolute best candidate is chosen. The empirical evaluation of the a) scanning of all the is divided into the following steps: songs songs b) stanzaic pattern of rhythmic cadences for each c) filtering of criteria to know what we song consider quatrain. Total: 627; 1028 a songs. The results are as follows: there are 203 metrically replete verses (4444); 188 couplet marking (4343); 38 couplet marking (4G4G); 35 quatrain marking (4443); 29 quatrain marking (F3F3); 26 12 without refrain invariant across ill-formed (not predicted), 14 of them with refrain, are (1998:25). Refrain stanzas, that is, it gets The conclusions observed attested ill-formed in certain songs as be defined as the verbal material that is repeated through the by the authors are song. surprising: there are some quatrains, and also unattested quatrains which nevertheless appear (1998:497). Some refrain examples sound fairly well formed where analogous non-refrain examples such can 4G43, which are seem ill-formed musicological point of view, it is rather lame. Furthermore, there as stanzas easy to see 183 but well-formed the reason as are structures, refrains. From for this: whereas stanzas a are generally structured into two length and rhyme in such from one another song to a way that a specific stanza could be easily transferred refrains tend to be one, rhyming schemes, which makes them in their they belong. In this are of two lines each, where each line has the same groups much sense, more more irregular both metrically and idiosyncratic to the which it is only natural that the structural requirements of stanzas tighter than those of refrains. The authors try to and mention a possible explain the memory lapse occurrence of non-predicted quatrain structures the singer's part. As on a less individual-bound explanation, they suggest that minor unmetricalities add unexpectedness to seem, song to a song that "adding of regulate song refer to as a a certain zest of (1998:498-499). However unscientific this argument might certain zest" is actually one of the main principles that composition and performance - what Hanson and Kiparsky (1996:295) "interest paramenter": "maximise the aesthetic interest of the verse". If every song responded totally to the principles of metricality the corpus analysed by Hayes and MacEachern would be nothing else than monotonous paper, a would be inexistent. As examples of what they call "minor unmetricalities" the authors quote *43f4G, *3f3f3f4, *434G, plus refrains 4344, Second, they try to explain why certain predicted structures, as is the case 433f3. GG4G, 3f3f43f, 3f3fF3f, 33F3, 3f3F3, F3fF3f, do not appear in the analysed They accept that there might be corpus. quoted structures are attested in nursery The last section of the paper a stated in this uninspired repetition of metrical schemata, where the artistic dimension inherent to song with as theoretical frame such provides a way as of taking a set grammar as inherent goals of the system resolution of the conflict in the chosen as a of corpus, given that the rhymes (1998:499). (1998:500) is devoted to highlighting the role of OT in studies of this kind. explicit well a gap raw According to the authors, OT structural preferences and turning it into an natural account of the variety of quatrain types. The are in conflict and each outcome represents a particular by assignment of priorities. More importantly, it makes it possible to rule out certain forms without actually formulating a constraint against them. The ranked set of CON with which up is the following: 184 Hayes and MacEachern (1998:502) end PARALLELISM (ii) PARTIAL LONG-LAST COHESIVENESS (iii) STANZA CORRESPONDENCE (iv) COUPLETS ARE SALIENT (v) TOTAL LONG-LAST COHESIVENESS (vi) FILL STRONG POSITIONS (vii) AVOID LAPSE (viii) QUATRAINS ARE SALIENT (ix) LINES ARE SALIENT (x) MATCH STRESS (xi) PREFER LONG-LAST The the (inviolable) (0 (inviolable) (inviolable) general categories of quatrains according to the analysis carried out are following (1998:504-505): a) well-formed, well-attested quatrains, such as 4343, derived from statistically likely rankings of CON b) well-formed, poorly attested quatrains, derived from fully legitimate but statistically unlikely rankings (3f3f43f) c) marginal quatrains, derived from ranking certain CON slightly outside their normal range of strictness (333f3) d) ill-formed quatrains, poorly- or unattested, not derivable within the system (3434). As out that a major conclusion to this English folk rhythmic cadences verse are paper, Hayes and MacEachern (1998:505) point is tightly patterned at the level of the quatrain: the various arranged in non-random, essentially strategic fashion. They emphasise the importance of cadences in order to induce perceived bracketings, which are then employed to structural ends: i) the enhancement of metrical constituents at various levels and ii) the placement of long elements last. Last, they praise the theoretical model of OT, which makes possible the use of primitive analytical ingredients to capture complex descriptive results. The general feeling of a musically-trained linguist is, that in order to build up a as already pointed out, comprehensive model of folk verse/song analysis, it is 185 devise necessary to a phrase/cadence, since musical theory that describes metre, pitch, quantity and we are dealing with song. Hayes (in press) "Text-setting as constraint conflict" 4.2.5. Hayes's (in press) paper constitutes point of view of OT. In this metrical CON in case, a deeper exploration of text-setting from the Hayes focuses on the interaction between the Hayes and MacEachern (1998) and pays special attention to four of them, namely: a) matching of stress to strong position b) avoidance of long lapses c) avoidance of extreme syllable compression d) alignment of phonological phrase boundaries with line boundaries. The pervasive idea repeated in these papers is that people have productive text-setting abilities, which they show by finding the best possible resolution between conflicting metrical principles in order to come up with a well-formed text- setting. The stated resolution between metrical principles involves "the sacrifice of certain principles in order to realise others with Hayes introduces his Lerdahl 1993), chanted verse" verse as seen paper a higher priority" (in press). by defining the "text-setting problem" (Halle and in section 4.2., which "arises in the context of (in press). I have already pointed out that equating is not accurate since there is much more to sung verse sung than a sung and and chanted succession of beats. Once again, the chosen representation for text-setting is the metrical grid, following Liberman (1979), Lerdahl and Jackendoff (1983) and others. As Hayes (in press) explains, "in a a grid, units arrayed in rows depict series of isochronous beats on hierarchy of levels, and the columns indicate the strength of individual beats". So, in this model, grids are very explicit in that they represent both constituency and saliency. The paper emphasises the idea that text-setting, just like natural language, involves well-formedness intuitions setting tradition, "if we are (in press). This entails that, for willing to abstract away from a already noted. The "modest amount of free variation" 186 specific text- modest amount of free variation, text-setting is predictable" (in press). This statement as a seems too to which simplistic, Hayes refers is precisely what makes certain folk song traditions so definitely what makes the study of text-setting such different from certain others, and an intricate topic. According to Hayes, the representation of the text-setting process would be as. follows: < Metricalgrid. Line x Table 28: As a sort of Lerdahl Theory -> Output -> Text-setting Text-setting process. background review, Hayes explores the first attempt made by Halle and (1993) to build up a metrical grammar for text-setting. Their grid looked w s as follows: s S row w m X M row X W row X w X X X X Table 29: Hayes does not agree X X X X X Text-setting metrical with the utility of the fourth he eliminates it from his own X in Halle and Lerdahl's grid, secondary accents in music, so grid representations. Halle and Lerdahl base their system on the so-called Syllabic Distribution Algorithm (SDA), which consists of a set of mapping rules and of when and how the rules X grammar. row which would differentiate between main accents and in the m w a group apply. The algorithm works by taking a of principles number of steps analysis of a specific instance of text-setting (Hayes in press): a) The line in question is scanned from left to right. b) When the a stressed syllable is found, it is made to occupy grid. c) The algorithm iterates, thus repeating steps (a) and (b). 187 the first S position in d) Whenever the assignment of a stressed syllable to S skips stressless syllables, a these stressless of specific rule is applied: the algorithm counts the number syllables before the stressed syllable that of stressless over a sequence mapped, then was maps syllables one-to-one, right-to-left onto the highest grid level having enough marks available to accommodate them. e) Any trailing stressless syllables preferring the highest grid from left to Hayes (in press) sees an on (EW) row. can - a variety of reasons. First, the SDA, based as far to the left Third, if there consecutive stressed spilled are more over enormous (in press). This is not desirable, - when there an Extra causes the are lapses. A possible so serve the purpose Hayes suggests an of "rescuing the alternative. According (in press), in order to determine why the algorithm sets syllables into the strong positions. Hayes states are if this leaves problems would be promoting and demoting stresses incorrectly, it is crucial to know its ultimate goal. Such which even left-to- into the next line. Fourth, if there systematically, something which would only him possible, than four stresses, the SDA syllables, the algorithm creates solution to avoid these to as on by the (undesirable) creation of could be achieved additional stresses to be metrics" accommodate them, but this time the line. Second, squeezing the stressless syllables than three are more Weak later that important flaw in this algorithm: it does not set all lines right mapping, places the stresses gaps row right. correctly. This could be due to large mapped to the remaining positions, again are a a so many lines goal is getting the stressed need to account for other goals, identified with the constraints with which he is dealing in the paper. Such goals will be avoiding lapses, aligning beginnings and ends of Intonational Phrases and ends of lines, and avoiding the extreme compression of syllables. So, the next step in the paper is to 'translate' these intuitive constraints into OT language and rank them in such CON are the a way that the system can yield well-formed text-settings. The following (Hayes in press): a) A stressed syllable "wants" to be in S position b) If not, M preferred over W c) An empty syllable is preferred following the scale W - M 188 - S. . d) Directionality we - refer to the want to Hayes (in press) adds three e) *LAPSE: f) never more never more LAPSE > as no well. a row EW position than the established number of syllables as unless CON to the above list, namely: three empty positions in These CON would be ranked necessary, phenomenon that S units tend to be long TILL EXTRA WEAK: g) *RUN-ON: * in Halle and Lerdahl (1993) is not as seen follows per line. (Hayes in press): *FILL EW > *RUN-ON > *NULL IN S > *STRESSLESS IN S > *STRESSLESS IN W > STRONG IS LONG > STRESSED IN W > *STRESS IN M > *NULL IN M. stating his theoretic-methodological apparatus, Hayes (in press) After presents his results, obtained after devising and testing the ranking as a computer program on native 364 lines. 70.6% of the outputs matched the favourite setting of the speakers, whereas 23.1% matched the SDA. In this respect, Hayes's alternative seems to complete the SDA in order to make the system effective in predicting the well-formedness of text-settings. Hayes's conclusions have to do with what he calls a "need for more Metrics" (in press), which will yield better results. The specific aspects that should be studied are the following (in press): a) Stress involves multiple levels. b) Importance of both word and phrase boundaries. c) Text-setting involves duration matching: the number of grid positions assigned to natural a syllable (all else being equal) tends to match that syllable's phonetic duration, as determined by pre-boundary lengthening and syllable weight. 4.2.6. Dell and Halle (in press) "Comparing musical text-setting in French and English songs" One of the terminological issues addressed in the to do with the definition of in which two ways: on the positions; and on under discussion in 4.2. has 'song'. Dell and Halle (in press) define it independent elements, namely one papers a text and a tune, as a composite interact in two main hand, in terms of the alignment of stressed syllables and strong the other, in terms of the alignment between the grouping of linguistic and musical constituents. 189 Throughout the analysis of Dell and Halle's (in press) and Halle's (in press) papers we observe that they rely heavily, both theoretically and methodologically, Lerdahl and Jackendoff s (1983) groundbreaking A Generative Theory of Tonal. upon fact, when Dell and Halle refer to the two Music. In interact, they are - over are 'time-span reduction' and 'prolongation reduction'. As (in press) explain, alignments similarities between units (1983), namely the 'metric hierarchy' and the 'grouping the other two Dell and Halle or text-settings make explore prominence in language in they simply state that this prominence is achieved by they use as synonyms. accent and stress in and that stress is of two formal of constituents into hierarchical structures. others and the arrangement which use language and music, namely the relative prominence of certain Dell and Halle do not terms in which text and tune simply re-naming two of the four main hierarchies developed in Lerdahl and Jackendoff hierarchy' ways Of means course, anyone one of the language, together with tone and ways a of stress way; or accent, two a synonym in which prominence can for the latter be manifested in combination of both pitch and stress thorough analysis of accentual typologies, see in fact, familiar with the study of language knows that the former is not only systematic a - for a section 1.4. In order to avoid terminological mistakes, it is crucial to point out that, when Dell and Halle (in press) talk about the manifestation of language", they such as are French and prominence in "stressed or accented syllables in necessarily referring to the action of linguistic stress in languages English. Leaving linguistic inaccuracies aside, Dell and Halle (in press) state that, in order for groups on text-setting to be successful, "linguistic the one hand, and on groups must match with musical the other hand certain stressed syllables must fall on strong beats". So, using Lerdahl and Jackendoff s (1983) terms, both the metrical and the grouping hierarchies of a given text and the alignment to be considered well-formed. As mismatch between are not given tune must coincide in order for an obvious consequence alignments. So, as a any general rule, "[t]he edges of linguistic coincide with those of musical groups" (in press). Again, Dell and Halle precise in their explanations of linguistic terms. In this case, of the term 'linguistic groups' by saying that they it vagueness of this, linguistic constituent structure and musical constituent structure will render ill-formed groups must a 190 use they solve the as a synonym of 'prosodic constituent'. If language, one must grid imposed on the existence of prosodic constituents in one asserts on any external purpose. In other- also accept that this existence is not dependent language at a specific moment for a particular words, the linguistic prosodic hierarchy is independent of the musical metric and grouping hierarchies. Following the Metrical Phonology tradition initiated by Liberman (1979), Dell and Halle (in press) use the metrical grid to represent the alternation between strong and weak beats in music, and explain that the columns in the grid correspond to metrical make the positions. As a representation musician, any more one has the feeling that the grid does not really explicit; in fact, it completely eliminates most of the melodic information contained in music notation. Let of notation for one of the compare us the two kinds examples chosen by Dell and Halle (in press): (176) Grid representation: G4 X G3 X G2 X G1 X X X X X X X X X X X X X X X X X X X X X X C A G A Bb A ne pleu- re pas Jean- net- Musical notation X X X X G F te representation: ftljlJ- J J p \^m Ne There is pleu - re pas Jean - net - te nothing in the grid that cannot be grasped by analysing the musical notation. What is more, the latter does not neglect the interaction between metre, pitch and cadence, while the former privileges metre over the other components of musical experience. Text-setting must surely have to do with tune which a as implies that the grid is valid only partially when the well- particular composite is analysed. 191 well or as with metre, ill-formedness of In the first main section of their paper, difference between the traditionally-called 'syllable-timed languages' and 'stress- languages', though they refrain from using these terms and simply explain that timed "in Dell and Halle briefly introduce the languages in which certain syllables preference for songs to stressed than others, there is are more a associate stressed syllables with strong metrical positions" (in press), something that they call "stress-to-beat matching" (in press). In these languages, such as English, certain stress-to-beat mismatches environments. The reader thus wonders which mismatches not, and on what grounds this gradient morphologically or with the to do Before occurs. are are prohibited in all allowed and which are Are these mismatches phonologically, syntactically motivated? Is it the that this gradient only has case prosodic characteristics of the language in question? going on to explain what they consider define a stress-beat mismatch, Dell and Halle (in press) crucially important concept in their theory, namely the a concept of 'tactus'. They adopt Lerdahl and Jackendoff s (1983) notion of the tactus as the regular beat into which listeners naturally fall when they clap their hands to that tune. The notion of tactus Halle to assume. is, nevertheless, much more An idea the neither Lerdahl and Jackendoff embrace is that the process problematic than Dell and nor Dell and Halle seems of perceiving the tactus is culturally determined, as proved by ethnomusicological research carried out in different parts of the world (see Gray 1998:617-627). As Gray (1998:620ff.) shows, if exposed to Ugandan traditional songs Ugandan people would clap tactus at an - falls on the on on one mean their perceived opposite beat to the Westem-perceived tactus. Let which starts on (177) Piece that starts on the upbeat an old man and he lived in the 192 east us have a look also called 'anacrusis' the upbeat (from Kennedy 1984:503): fljJUJjiJ.Uijjjij There was - the downbeat: on was the main downbeat of each bar, while the upbeat, which would example of a musical piece which starts and then Western audience and asked to clap to the music, they would probably define the placement of the tactus the a n And his (178) Piece that starts IVe on the downbeat (from Kennedy 1984:606): nob-by nob-by coat got a coat and a (177) and (178), the tactus falls at the level of the crotchet, which is signalled In both by 4 in the time signature. The definition of tactus is essential for Dell and Halle's of what a stress mismatch is in Let S and s be two English: syllables occurring in and the other unaccented adjacent within the (in press) explanation same any order, (s). A stress mismatch one occurs accented (S) if S and s are line and the metrical position associated with s belongs to the tactus, but not that associated with S. So, for Dell and Halle, Kaun a mismatch would be roughly the same as what Hayes and (1996) call "lexical inversion". This by Hayes, identifies lexical mismatches We could say all as a just like all the and text in song. one can observe Actually, mismatches a papers co-authored sign of ill-formedness in text-setting. that this identification is valid, but only to European traditions composers paper, a certain extent. In almost certain degree of mismatch between music are also common in the works of art-music of all times. In his outstanding analysis of the early of the fifteenth- songs century composer Dufay, Boone (1999:82) states that, in the process of imbrication or overlapping between musical and poetic rhythms, there is always a "seeming disjunction between poetic and metrical (musical) stresses [...] commonly prompted by the presence of formulaic rhythmic modules in the musical setting". A certain degree of disjunction is in fact monotonous As more and necessary, as songs from becoming inartistically predictable. explained before, 'accented' means 'stressed' for Dell and Halle. What is important is that in this definition Dell and Halle suddenly introduce the line a verse prosody concept - as mismatches. The line is not an a - essential structural unit in the determination of stress one of the (1989), whom Dell and Halle quote. It draw it prevents prosodic constituents established by Hayes seems to clear difference between theories of 193 be the case that this speech prosody, paper verse does not prosody and In that regard, the theoretical frame used by the paper is too theories of text-setting. ad hoc to be After and Halle for applied systematically to the study of text-setting in different languages. analysing the setting of specific English texts into (in press) come up with a chosen tune, Dell series of mismatches which they label "illicit" English. They state their observations A a as composite is ill-formed if it contains follows: a stress mismatch which meets one of the following conditions: • the two syllables are not • the two syllables are separated by a word boundary or separated by a word boundary and s precedes S. Again, Dell and Halle's level of accuracy in the linguistic side of their investigation is not as high as would be desirable. As stated by the authors, (a) applies to simple syllable falls words whose stressed rebate as examples sequences for the - on a weak beat they give dinner, balloon, (b) applies to but does not account for compounds, while of a grammatical word plus a lexical one opposite order - see me. They do not give covered - the boy - but does not account - any further account of the cases not by the above statement. At the end of this section, Dell and Halle turn their attention to the melodic be explained by element of tunes, and thus leave aside the idea that mismatches relying solely the metrical side of alignments. They explore which elements on important in order to define 'melodic contour conservation' be tunes to contour" with the perceived a same tune, are follows: "for two they must have the same melodic requirement of a purely musical nature, which has to do harmonic, melodic and rhythmic skeletal structure of a piece of tonal music. a tune is the sequence of the pitch changes in that tune, abstracting a kind of parallelism that has to do with the contour of a tune. On the other hand, there is another kind of parallelism which Dell and Halle call "positional parallelism" (in press): "two alignments along the grid are are positionally parallel if the distributions of their syllables identical". The implication of this concept for text-setting is that composites which do not have the same number of syllables cannot be positionally parallel. Nevertheless, this statement needs like as from the number of notes attached and their timing. So, there is musical two variants of the (in press). This is The contour of away as can Spanish, where syllable compression devices 194 are to be qualified. In a language often applied to both speech - and verse, two lines with different numbers of syllables which correspond to the they what same number of beats. Dell and Halle lack by 'syllable': mean are a can be made to clear explanation of they speaking about syllables in the prosodic hierarchy or in the poetic line? The second main section in the paper explains what a mismatch is for French, according to the prosodic characteristics of this language. Dell and Halle state that "in traditional French songs, where certain melodies require a the line Halle alignments require are rigidly constrained at the end of lines", form which is phonetically feminine, while others a form which is phonetically masculine. Again, as a structural and salient component (in press) use word'. This has to do a normally corresponding to pronounced for metrical the importance of of the grouping hierarchy. Dell and 'form' meaning 'the pronunciation of with the status of schwa in French, sometimes be we can see reasons, while on an e-muet, which can other occasions it is not pronounced at all. They define the concept of'local maximum' (in press) "the grid position associated with the last accented syllable in than the as a the requirement that line must be stronger positions associated with adjacent syllables belonging to the same line", parallel to the concept of 'stress maximum' stated by Halle and Keyser (1971). As explained here, local maximum only match stress and beat is concerns line ends, although commonly found in traditional French a tendency to songs. The crucial point about French text-setting is that, while the conflicts between stress and beat at the end of lines are extremely perception, such conflicts time they French press), escape strophic a song are rare and notorious from the point of view of commonplace in other positions, where most of the the listener's notice. Dell and Halle (in press) also notice that in song all stanzas are a positionally parallel. According to them (in with positionally parallel stanzas has the following properties, among others: a) All stanzas have exactly the b) If two lines occupy the same tune. same position in the stanza, they have the number of syllables. c) The distribution of melismata is the same 195 in all stanzas. same Once again, musical term is not explained. In this a 'melisma' is case, the definition of the term missing, something which will certainly make this part of the paper unintelligible for non-musicians. Melismata very common are in traditional music, as we can observe in (179), from Kennedy (1984:653): (179) O In (179), we m who-will mar-ry my fair - jp la-dy? O observe that the syllables who and fair are both set to two different pitches. They constitute examples of melismata. Dell and Halle (in press) take the quoted three properties as consequences of combining positional parallelism with Melodic Contour Conservation. According to them, the main difference between English and French is that in the former departures from positional parallelism per are common - less syllables or more syllables line require that notes be deleted or inserted (in press). In their conclusions, Dell and Halle (in press) emphasise the importance of positional parallelism and arrive at a universal definition for this concept: "as far as possible, identical melodies give rise to positionally parallel settings". It is only at the end of their paper that Dell and Halle (in press) raise the question of why prominence matching is enforced in a much more rigorous They state that, rather than being merely difference has than in a they emphasise constraints their own, on a matter phonological basis: stress has French, something which once any acute manner more as of poetic convention, that perceptual salience in English observer could conclude. In this respect, again the importance of acknowledging the fact that the only the texts are those deriving from the linguistic without considering their potential for being set to "any intrinsic organisation in feet, lines, stanzas, form in English than in French. it is or any grammars. a tune, Texts on do not have other of the units of poetic conventionally understood" (in press). The chunks analogous to the lines and stanzas of literary verse in which products of text-to-tune alignment. They song can are not 196 be broken down are just by¬ features of the text itself but features composite text-tune. This of the seems to be in contradiction with the theoretical assumption made by the authors that the line exists structural unit previous to as a being set to music. This paper, idea that a like Hayes's builds up a theory of text-setting around the perfect match between linguistic stress and musical metre is the epitome of text-setting well-formedness. However, this all the theories with which the there have papers, seems to be precisely the weakness of analysed papers deal. As Boone (1999:212) suggests, always been two major pitfalls in theories about the relationship between music and text: "the first pitfall is overemphasis on tonic accent as the prime agent of rhythmic activity in linguistic rhythm; the second is overemphasis alignment between tonic accent and musical accents language, poetry and music and, varied traditions of folk song, or Halle's much about what is going importantly, yields a as (in press) and starts by defining it independent objects, each of them must be a text and judged as acceptable within the "composite made up an objects. On the grammars are two one hand, applicable to the acceptable interaction between music while necessary, is not sufficient to determine "both the text and the tune must be judged as a tune" (forthcoming). There a respective cognitive domains. On the other, acceptable as a well-formed text-setting: independent objects in order composite to be judged acceptable" (forthcoming). However, this statement for the to written songs song. its point of departure the concept of 'song' put observations to be made about the interaction of these two seems in simplistic account of the in the intentionally imbricated realms of art takes paper forward in Dell and Halle and text, on (forthcoming) "Text, tune and metrical form" (forthcoming) of two more us as a stress game of interacting between text and music, be it in the enormously rich and infinite ways 4.2.7. Halle really tell simple the hallmark of 'good' as prosody". The view of the relationship between music and language is distorted because it does not on a - be too strong by the in a if we think about same person - such is the songs case simultaneous sort of way. In this without the music, could in which words and music have been of Thomas Campion's case, hardly be considered it is often the verse or, or case Cole Porter's that the lyrics, indeed, good verse. Consider, for instance, the fragment in (180) by Porter, in which the lyrics would not have been considered well-formed poetry if they had been composed with 197 no tune in mind maybe they would, just because of the fact that the lines show quite a - predictable rhyme scheme: (180) It was just one of those things, just one crazy one of those bells that just one It a one one one Halle's paper now and then rings, of those nights, of those fabulous flights, trip to the just flings, of those things. just was just theoretical of those moon on gossamer wings, of those things. deviates from Dell and Halle in that it does not constitute a comparative study of two distinct text-setting traditions but rather an attempt to devise a generative model of text-setting for English. In this sense, as already pointed out, it relies theoretical and on Lerdahl and Jackendoff (1983) hypothesis on as on the well-formedness of text-settings relies concept of 'similarity metric', defined as "a mechanism of variants judgements well certain methodological OT tools. Halle's main class as which can be well-formedness on construed are as on the which formally describes the acceptable" (forthcoming). The made independently for music and text with reference to the similarity metric. Halle (forthcoming) states that two types of constraints work in pages -, three are at 'strophic song' - a term which will be defined in the next namely interactive constraints and independent constraints. First, there types of interactive constraints, constituency constraints and constraints on namely stress matching constraints, melismatic structure. In order to explain matching constraints, Halle resorts to Morgan and Janda (1989), stress one studies which demonstrate that mismatches between stress and metre along a that in gradient range Spanish there is mismatches. Halle of many are judged from natural to unacceptable. Thus, Halle (forthcoming) states English the melody gets distorted while in are no so that the stresses do not get mismatched, alteration of the tune in spite of the encountered (forthcoming) repeats the conclusion stated in Dell and Halle that 198 positional uniformity - the relation of musical in certain vocal traditions events to the metrical grid found in the original tune maintained in the face of what other text are defined as stress mismatches. In traditions, significant deviations from positional uniformity required in order to achieve the metrical Halle are be can - a proper matching of syllabic stress and grid. points out the importance of observing constraints which deal with bigger than the syllable. One of these constraints is constituency matching. The units author (forthcoming) suggests text substitution as a test to find evidence that an 'in phase' relationship between musical and linguistic constituency tends to result in more natural sounding arrangements concludes that prosodic hierarchies. Halle (forthcoming) - constituency mismatches appear to constitute a more or less exact analogue to enjambment within metrical poetry. The third set of interactive constraints is that of constraints He defines melismata structure. as "syllables assigned to more on than melismatic one note" (forthcoming) and quotes Stoquerus (1988[ 1570]) when he states that "whereas the number of notes in a passage can [...] exceed the number of syllables, the number of syllables cannot exceed the number of notes" (Halle forthcoming). This statement is not completely accurate since, in phonological syllables those syllables can a language like Spanish, the number of exceed the number of notes - a different issue is how realised phonetically, that is, whether there are are syllable compression devices that operate in order to adjust the number of syllables to the metrical positions available. Halle quotes Stoquerus's Rule 4, namely "no two adjacent events assigned to the this would not be a same melisma, but syllable a may be assigned to the same pitch" - repetition (Halle forthcoming). Last, Halle (forthcoming) explains the adaptation of Stoquerus' rules by Gauldin (1995) and Veltman (2001), which results in the rule that "the continuation of not appear in a metrically stronger position than its onset". The use a melisma may of Stoquerus's rules, which date back to the sixteenth century, seems to be anachronistic here since it has been proved that Stoquerus's rules only apply to the repertoire which he discusses. Also, with reference to Stoquerus's rule 4, it is important to note that in 199 early seventeenth Italian monody it of melisma on a note, a was indeed possible to realise specific subclass a figure called 'trillo'. Second, another set of constraints is that of independent constraints on strophic which song, strophic song as a can musical which distinct texts - be linguistic genre or musical. Halle (forthcoming) defines characterised by "the reiteration of a melody onto variously referred to as stanzas, verses or superimposed". Given the nature of strophic verse to forces verse pitch deletions whereas some as acceptable others are not - they do not compromise the integrity of the tune (forthcoming). We could say, pitch deletion -, going back to the concept or insertion is acceptable as that is, as song, as long the Melodic Contour Conservation principle. Halle (forthcoming) introduces the idea that there is regard to the metrical pattern of the first stanza of positions allow for either require the disparity of syllables from it does not disrupt the melodic contour of that specific it respects are - significant deviations in the musical form of each; also, some in Dell and Halle, that any of 'contour' long are songs, strophes occupancy appear to by events for the setting to be acceptable and finally, others requirements of similarity metric on certain metrical song, by events; other positions vacancy or occupancy required to remain vacant. Section 3.0. of the substitutions specific a gradient where, with a or paper are explores the descriptive metrical parallelism by means of testing text well-known tunes. Section 3.1. states the formal characteristics of with their representation of similarity metric. There are representations, namely, mandatory where the position must be occupied in the original and in all acceptable variants where the - represented with original is vacant, and it '+' and mandatory may vacancy, acceptable variants "the - - optional may occupancy, vacancy, be deleted in acceptable where the position in the be occupied in acceptable variants - represented with where the position must be vacant in the original and in represented with '0'. According to Halle (forthcoming), similarity metric associates each metrical position with Section 3.2. discusses the derivation of similarity from the idea that cases represented with ' 1'; optional position is occupied in the original, but it circumstances all occupancy, four main "knowing a tune means one metric. Halle (forthcoming) departs in addition to having 200 of these categories". access to the original . of the distribution of optional and mandatory positions which form, an awareness define acceptable variants Halle as these are (forthcoming) suggests specified in the similarity metric". a series of similarity metric assignment rules (SMAR), which resemble the kinds of rules developed by Lerdahl and Jackendoff (1983) under the name are mandatory, on the part of the song, assign category 'well-formedness rules'. These rules opposed to 'preference rules', which have to do with composer/listener. The SMARs listed are as a choice as follows (2005:20-29): (a) SMAR 1 (metre): given the original form of a strophic 1 to all strong metrical positions occupied in the original. A strong position is assimilated to a "tactus" level position, as defined by London (Grove Music Online). (b) SMAR 2: if X and Y are adjacent events assigned to different pitches, X precedes Y, and X is stronger than Y, assign category 1 to Y. (c) SMAR 3 (group edges): within each group, designate as mandatory (i) the metrical position corresponding to the initial event in the original and (ii) all metrical positions to the right of the final event of a group. (d) SMAR 4 (default assignment): designate all non-assigned positions optional categories - or 3 depending Section 3.5. introduces this section Halle a on as their occupancy status in the original. novelty with reference to the previous articles. In (forthcoming) talks about pitch assignment and makes two important observations, namely that optional locations inherit pitch from the nearest assigned location to the left, and that the pitch assigned to anacrusis will be determined on a case-by-case basis. Section 3.6. deals with the concept the similarity metric, requires that reiterations of attacks can repeated pitch events as expressed within mandatory status initially assigned to any may of 'fusion'. Fusion, be redesignated as optional so that these be deleted in the variant: (e) SMAR 5 (fusion): given two adjacent identically pitched events XI and X2 within the same subgroup of an original tune where XI precedes X2, redesignate the location containing X2 (forthcoming). 201 as optionally vacant category 2 Finally, section 3.7. gives some restrictions on grouping structure. As a result, SMAR 3 must be revised: (f) SMAR 3 (grouping-revised, see (c)): within each group, designate as mandatory (i) the metrical position corresponding to initial event in the original, and (ii) all metrical positions to the right of the final event of a group maximally long (forthcoming). when this event is attempt to build up a generative theory of text-setting, Halle In his by stating, concludes again, the importance of similarity metric in text-setting and metrical once form, particularly when it comes to judging the acceptability of certain cases located in a somewhat 'grey area' of marginally acceptable There is doubt that the no or unacceptable settings. approach chosen by Halle represents perspective from those explored in the previous papers. a different Nevertheless, it does not represent a novelty in that it clearly inherits its theoretical framework and methodological tools from Lerdahl and Jackendoffs (1983) generativist views applied to music, a theory which, although groundbreaking, has ultimately proved conception of music limited in its as a series of innate mechanisms manifested in metric and melodic terms. Kiparsky (2006) "A modular metrics for folk verse" 4.2.8. Kiparsky's (2006) paper constitutes a critical comment, as well as a theoretical and methodological follow-up, to Hayes and MacEachern (1998). Hayes (in press) had already pointed out that the prosodic structure of language (stress) and is verse (metre) governed by separate constraint systems which must be jointly satisfied by well- formed verse. As Kiparsky (2006:7) observes, this statement entails approach to metrics. Kiparsky (2006:7) takes the modularity idea arguing that "the composer tiers of rhythmic structure: These three tiers hierarchies of are and performer of a "a text has has an modular further by match between three linguistic prominence, poetic metre, and musical rhythm". organised according to the alternating prominence which grids. The crucial point is that, autonomous. song constructs a a step a as are kind of principles, as visually represented by trees or same Dell and Halle (in press) observed, the tiers are Kiparsky (2006:7) explains the concept of tier autonomy by saying that an intrinsic prosodic form independently of how it is versified, intrinsic metrical form independently of how it is set to music, and 202 a stanza a tune has an intrinsic musical rhythm independently of the words that other words, "the stress pattern be sung to it". In (or other linguistic prominence relation) which linguistic rhythm of determines the intrinsic may song's text is assigned by the a language's prosodic system", while "the metre of its stanzas and the rhythm of its tune are normally drawn from How the tiers and a traditional repertoire of rhythmic patterns" (2006:7). correspond to each other, and in what ways they can be mismatched mutually accommodated, is regulated by conventions that evolve historically, though within limits grounded in the faculty of language. Kiparsky (2006:8) disagrees with Hayes's idea that the metrical form of verse can be equated to the way a in which its text is aligned with the musical beats in performance. He gives several arguments against such identification and in support of the traditional division between metre and music. agrees that OT is a Although Kiparsky (2006:8) good frame to model the groundedness of metrical preferences competition within variation is better treated by partial constraint ranking than by stochastic OT, which is the model used in The paper to the is centred on Hayes and MacEachern's (1998) results with regard The conflict between well-formed verse. quatrains and ill-formed but attested quatrains encountered by the former is resolved by Kiparsky quatrain types unmetrical, except where the existing as as follows. On the plausibly be considered accidental which are so the other infrequent as a case theoretical and re-analyses the a in the more folk song corpus can only with refrain quatrains, than are once as a full sample. On metrical. Also, he "too diverse to be entirely quatrains" (2006:9). Hayes and MacEachern's conclusions and integrate his of folk examined in Hayes and MacEachern (1998) and songs collection of hymns The first part MacEachern's gap methodological additions to the study of text-setting, Kiparsky corpus the model in hand, he treats all unattested whole that the data is unlikely to be constraint system as same In order to review tests this is the one rhymes aside, arguing that their metres covered in the own - hand, he treats quatrain types attested sets nursery argues Hayes (in press). inventory of quatrains in English folk but unattested a metrical system, he also that and constraints and their of Kiparsky's (1998) findings, as by Watts. paper well as 203 is devoted to summarising Hayes and adding some new ideas to enhance the theory. As already observed, Hayes and MacEachern (1998) classify folk lines into four types on the basis of their rhythmic cadence - 3, 3f, 4 grid placement of the final two syllables. In with the they recognise other types. Kiparsky (2006:12) grouped under the which he calls 3'. The that types G, 3f and 4 metrical category, characterised by same reason is that "with respect to which has to do G latter appendix to their a argues or a paper, can be monosyllabic last foot, metrical form, type G lines are indistinguishable from type 3f lines, but with respect to musical performance they are more like type defend 4 lines" (2006:12). This is the first argument used by Kiparsky to separation between metrical and musical constraints. As he points out a (2006:12), "[t]he musical implementation of such appropriate to the tune to which it is constraint the sung a final monosyllabic foot is [...] Within the limits imposed by that singer is free to decide and variation is free to occur". The second argument has to do with the well-acknowledged fact that "the same words are commonly constraints on different musical sung to on musical stanza form" measures, (2006:14). The third argument for separating constraints performance from metrical constraints metrical constraints are applicable also to literary chanted" (2006:15). Actually, sung or yet maintain certain invariant inventions follow the same as on stanza verse that form is that "the was never meant to be Kiparsky (2006:15) observes, literary laws of stanza construction as folk poetry, namely parallelism and closure. So, song once again, the difficulty is discovering how text and tune tradition. Kiparsky's first assumption is that text and tune have "a firm conventional association" The second and (2006:15), yet they are to some lines, couplets, and entire quatrains reasons, musical as performance is Kiparsky (2006:16) argues close can float from more or a less degree independent. as Hayes and MacEachern claim. paper another" In conclusion, for the validity of a modular approach, which "makes for other, and why others do not Kiparsky's one song to Kiparsky doubts that the relation of verse form to simpler metrical inventory, and correctly predicts why with each related in assumption is that text and tune "can originate and develop separately, (2006:15). For all these a are occur at some line types alternate all". is almost entirely devoted to grouping CON, that is, SALIENCY and PARALLELISM. With regard to SALIENCY, Kiparsky (2006:16) 204 states that closure can be achieved either There two are by rhyme or by the presence generalisations to be made about the interaction between these two CON, namely that "a couplet must not have decreasing saliency by the inverse of length (shorter lines measured and that "a salient the couplet is parallel if its lines are are more a line well-formed. or as a gradient property, meaning another unit is salient in relation to another similar unit. In saliency is "the inverse of length, measured in beats: terms, salient, couplet must be either higher than the other in order for Kiparsky (2006:21) conceives of SALIENCY a salient than longer ones)", equally salient - time. In other words, one must be ranked stanzas to be that saliency is- - parallel" (2006:17). So the two main grouping CON cannot be obeyed at or same of a refrain. a a descriptive full foot (F) is non- reduced foot (f, 0) is salient" (2006:21). At the theoretical level, SALIENCY is unfaithfulness, because it causes mismatches between metrical positions and the linguistic elements that correspond to them. In this observe the interaction between way we can grouping and metrical CON: "in English saliency results from unfilled positions in the metrical grid, definable of the faithfulness constraint MAXBEAT" as a verse, violation (2006:21). The interaction between the types of CON is summarised as follows (2006:22): two • Full foot • Degenerate foot (f): [o] (mismatch, salient). • Null foot At this (F): [o o] (perfect match, non-salient). (0): [ ] (maximum mismatch, maximally salient). point Kiparsky diverges from Hayes and MacEachern (1996) by observing that SALIENCY and PARALLELISM work at higher levels of metrical structure in recursive fashion. To those main adds a third one, quatrain contains of a salient An CLOSURE, which states that the salient couplet of a salient no marked feet. This last constraint indicates that the final couplet quatrain is not only salient, but unmarked, i.e., it is of the form 43. interesting point in Kiparsky's analysis is his in-depth explanation of the peculiarities of refrain quatrains in folk have a verse. First, "[rjefrain quatrains [...] do not rhyme, probably because the closure of rhyme is served by the refrain constituent" ones. grouping CON, Kiparsky (2006:23) (2006:24). Second, they allow all metrical types, plus a few additional Third, they "do not seem to consist of two couplets, but simply of four 205 coordinated lines ([4443']), or in some cases even three lines capped by a refrain ([444][3'])" (2006:24). The main criticism made by Kiparsky to Hayes and MacEachern's account is that the latter follow stochastic OT in order to assigned "each constraint is likelihood of a place on a explain the ranking of CON, whereby scale of real numbers, which governs its outranking other constraints and being outranked by them" (Kiparsky 2006:27). In other words, the stochastic model predicts that the constraints are strictly stratified along the scale. Kiparsky (2006:27) identifies three problems with such an approach. On the fact that hand, it does not relate frequency to unmarkedness, a contradict the intuition that the most to seems one tend to be those which are the frequent metrical structures simplest. On the other hand, Kiparsky (2006:27) mentions "the harmonic bounding problem", which has to do with distinguishing two forms that differ only in strictness. Kiparsky (2006:27) exemplifies this by stanza stating that, in OT terms, "ABCB is harmonically bounded by ABAB". As for variation, stochastic OT posits that each CON has a range of fixed width, within which it freely can vary. The last problem is overgeneration, that is, Hayes and (1998) model predicts too MacEachern's quatrain candidates. According to many Kiparsky (2006:37), those three problems could easily be solved with the introduction of an extra CON, is to license any namely FAITHFULNESS: "The effect of Faithfulness candidate not excluded by higher-ranked constraints Constraints ranked above this cut-off ranked below it metrical CON consequence - are inactive" metrical. point restrict metricality, while constraints (Kiparsky 2006:33). If FAITHFULNESS outranked the grouped under MAXBEAT -, any input would be accepted, with the that the output would be prose. The solution given by Kiparsky (2006:39) has to do with what he calls "a partial ranking account", meaning that "rankings case, as can be either free or fixed". In any FAITHFULNESS "must be dominated by at least one markedness constraint" (2006:39). According to this, there are three possible grammars namely: a. SALIENCY » FAITHFULNESS b. PARALLELISM » FAITHFULNESS c. SALIENCY, PARALLELISM » FAITHFULNESS. 206 for this system, Summarising, the single additional restriction that SALIENCY is important than FAITHFULNESS has one a more double beneficial effect in the system. On the hand, it excludes the prohibited couplet types and, on the other, it generates the pattern of preferences among the remaining permissible couplet types. Among grouping CON, SALIENCY is preferred something which PARALLELISM at the couplet level, over be observed both in the corpus of folk hymns. According to Kiparsky (2006:42), the reason for this ranking "lies can musical side", that is, "it is a feature of stanzas intended to be this, PARALLELISM would dominate "in literary than singing" (2006:42). At the quatrain and in Watts' the on sung". In contrast to designed for reading rather verse level, songs outranks PARALLELISM SALIENCY, something which Kiparsky is not able to explain. As a sort of conclusion, the author (2006:44) states that "parallelism supersedes saliency dominant is almost 4.2.9. organising principle" - actually, "above the level of the stanza, parallelism completely dominant". Summary In this section I have introduced the main theoretical and on the as which OT relies for the The methodological principles study of text-setting. assumption made by OT in this particular field is that text-setting, that is, how lines of linguistic texts are arranged in time against a predetermined rhythmic pattern in sung verse, is a universal ability. Among the numerous text-setting CON explored in Hayes and Kaun (1996), Hayes and MacEachern (1996, 1998), Hayes (in press), Halle and Dell (in press), Halle (forthcoming) and Kiparsky (2006), there are three essential can be subdivided ones as - MAXBEAT, SALIENCY and PARALLELISM - which follows: a) Metrical CON: MAXBEAT a. 1) MATCH STRESS: a. 1.1) strong syllables fall a. 1.2) a syllable that falls on strong on a beats S beat must be stressed a.2) FILLSTRONG: fill the strongest positions in the line a.3) *LAPSE: avoid the strongest sequences with positions in the line b) Grouping CON: 207 no syllables between any two of b.l) SALIENCY (truncation): non-filling of metrical positions at the end of lines b.1.1) Lines are b.l.2) Couplets b.l.3) Stanzas salient are are salient salient b.2) PARALLELISM: b.2.1) the cadences ending the units of the maximal analysis of a quatrain must be identical b.2.2) the onsets of syllables coincide with musical beats. Both the interaction and the conflicts among are by the type of language that determined English, traditionally classified more we are words, if a dealing with. In a language like stress-timed, the metrical CON MAXBEAT and, as specifically, MATCHSTRESS, regulate the In other the above CON determine and process of text-setting in folk text-setting in English is to be qualified as song. well-formed, the agreement between musical beats and linguistic stresses must be as high as possible. On the other a hand, different way observe in the not enforced ranked of French, in strophic As are scarce. conclusion that the well-formedness language like Spanish, classified as syllable-timed, will work in with regard to MATCHSTRESS. As Dell and Halle (in press) case in a a song; Spanish the agreement between beats and stresses is in fact, the result of these cases where MATCHSTRESS is highly observations, importance of stress-placement in folk song we can state as a the preliminary determinant of text-setting is systematically diminished in syllable-timed languages, while it is enhanced in stress-timed languages. This fact has impact on different the behaviour and ranking of grouping CON such syllables along the grid Consequently, if lines In other Spanish PARALLELISM in languages. As explained above, positional parallelism requires that the distributions of song. as clear a are not number of can identical for each stanza of a strophic isochronous, they cannot be positionally parallel. words, only languages whose - are verse prosody requires isosyllabism - like render positionally parallel text-setting, while languages where the syllables per line is secondary to the position of stresses will hardly require positional parallelism to be judged well-formed. The analysis of the most recent papers on the topic of text-setting confirms that in vocal music there 208 are three tiers of rhythmic structure - linguistic prominence and grouping, which grouping and musical rhythm and metre prosody constraints, they determine the ranking of constraints at that level, and verse same happens with the influence of both speech prosody and constraints on link between verse prosody and musical metre is stronger than in Spanish. This PARALLELISM is enforced in English, not in Spanish, while positional Spanish and constantly overlooked in English. preliminary conclusions need to be tested against These selected from a corpus section 4.3.6. and corpus, of 444 Spanish songs and 239 English songs case - of folk studies analysed in appendices IV and V. The main criteria used to select this specific which I explore in full in section 4.3.5., could be reduced to two. On the hand, they are folk songs, one passed from generation onto generation, which entails that matching of lyrics and music responds to tune a corpus In ensuing sections I will apply the theories reviewed to specific songs. prosody verse text-setting constraints. As explained in this chapter, in English the entails that MAXBEAT is enforced in the intimately related. While the applied at the level of speech prosody cannot be directly translated into constraints the are and verse metre some intuitive understanding of text- alignment. On the other hand, in opposition to the used by the authors songs analysed here have been minimally edited by the reviewed above, the ones collectors. 4.3. Object of study: Folk This section is devoted to songs analysed in my song discussing the song, to which the opposed to 'art song'. Since the goals of linguistic and, to anthropological, it is beyond the folk culture. Section 4.3.2. by the collectors of folk of my certain extent, musicological, but not a purposes of this dissertation to cover every aspect of explores the transcription and classification criteria used song and, more specifically, of the two collections that I analysing, namely Kennedy (1984) and Manzano Alonso (2003). I shed on corpus 'song' and 'folk song', pointing to the most important characteristics of the genre, as are of folk dissertation belongs. Section 4.3.1. explores the meaning and evolution of the terms dissertation genre the controversies related to ideas of 'faithfulness' to the informants. In section 4.3.3. I carry out a originals as sung brief analysis of the possible 209 some am light by the reasons why lyrics have always been given more importance than melodies in the study of folk In connection to the previous issue, section 4.3.4. introduces the topic of the song. relationship between lyrics and melody in folk song, which is the nucleus of my dissertation, with special emphasis on the concepts of lyric and melodic 'identity' and 'variation'. Section 4.3.5. constitutes in my brief introduction to the corpus a analysed dissertation. Finally, section 4.3.6. is devoted to the analysis of the first sample made up of two folk corpus, songs, and presents some preliminary conclusions on the topic of text-setting in different languages. 4.3.1. Definition and characteristics of folk song As musical form, a 'song' can functional elements. There is is, a can an be defined as an integrated whole of inseparable indispensable melodic element to which lyrics, that verbal text endowed with meaning, is joined. This composite of tune and text be associated with accordance with the dance, that is, series of controlled body movements in a melody and the rhythm of the song. Songs accompanied by sound-producing instmments which enhance the melody establishing the rhythm. In this As a pages, song and folk a or help complex cultural manifestation. be subdivided into two broad but sharp musical genre, song can categories, namely art is sense, song often are song. As will be explained in the following the main difference between the two has to do with an exogenous factor, that is, whether the songs have been kept written down or whether they have passed orally from one person to to the life and become a habits of country - mainly in rural environments and in connection people - in the course of history, until they have part of a specific musical and, more broadly, cultural tradition. When has another talking about folk experienced a song we need to acknowledge that the term 'folk' considerable semantic evolution over the years. As Kennedy explains (1984:3ff.), at first 'folk' referred to unlettered country people or peasants. During the Romantic period, the word gradually acquired another nuance definitely left its derogatory word 'folklore' was sense attention of composers and behind. At the end of the nineteenth century, the coined in order that time that the music of the and to refer folk, both to local beliefs and customs. It songs and musicians looking for at and dances, started to attract the new literary and musical art forms, by people working in educational fields. The words 'folk song' 210 was are a literal translation of German word, Volkslied. Folk song a multiple manifestations of folk culture, evolved conscious as is, therefore, only of the one the behaviour pattern of non self- peoples. As Kennedy (1984:8) points out, include These [...] those country folk who still depend unconscious process such as songs on an of acquiring their traditional craft skills and lore, the oral transmission of music, poetry and stories. Traditional and ballads reflect the social conditions and ways of life of a community. There First of are all, several issues which make folk song a the definition of folk difficult genre to deal with. has been controversial amongst song ethnomusicologists of all times. Second, stemming from the difficulty to find a unifying definition, there is the issue of determining the characteristics that make folk song what it is. Third, when we come to all sorts of methodological problems song, collecting of the songs to we can emerge, ranging from the recording and their transcription and the grouping criteria used to jlassify directly linked to the relatively short time span to them. All these difficulties which deal with the scholarly study of folk are resort in order to hypothesise about folk song. As the Spanish ethnomusicologist Manzano Alonso (2001:31) points out, "the hypotheses about the origins of these become song songs can only go as far behind as five or six centuries, and never anything else than hypotheses". One of the first and most influential folk collectors and scholars in Britain English Folk Song: Some conclusions - was Sharp (1859-1924), whose volumes originally published in 1907 - and English folk songs, collected and arranged with pianoforte accompaniment by Cecil J. Sharp - first published in 1916 folk song - set a milestone in the academic study and, indeed, in the revival witnessed at the beginning of the twentieth century in Sharp's perceptions about the nature of folk 60 song are England.60 still taken to be basic for any phases during the twentieth century, namely before World War I and in the period between wars. The most notable folksong collector in the decade before World War I was Cecil Sharp. After World War II and the ensuing period of austerity, there grew a national desire to celebrate British culture and tradition. The BBC undertook to coordinate the fieldwork. Expertise for the project, referred to as the BBC Folk Music and Dialect Recording Scheme, was drawn from folk music experts such as Peter Kennedy and Seamus Ennis. The revival of British folk music occurred in two distinct 211 study of the topic. In fact, Sharp's (1954:1) definition of 'folk song' has become classical in the literature the on In every land beautiful quality, of although they of the same topic: do find music of we nation. This spontaneous utterance addressed people', the which colour is called folk-song. problems. Sharp plays around with certain always convincing per se. What does 'having the unlettered classes are - or who by Sharp (1954:4), who explains that "the non-educated, are the unlettered, those whose faculties have undergone educated persons to never 'the common formal training been brought into close enough contact with common origin and its spontaneity, multiple in form" nature of folk music, among as well as song is other things. never completed, the "radical importance" of lyrics (1954:19). In the above to or no Apart from this, Sharp (1954:15) emphasises the "communal, allusion they at the be influenced by them". So, according to Sharp, folk by its beauty, its in this genre were a specific limitations of folk song? The second of these questions is whatsoever, and who have defined are not quality' mean? Who are own a very be entirely shared by the educated of art-music may not highly subjective terms which time? What distinctive and often of beauty and character of its a This definition is not without beautiful a quotations, the oral quality of folk the 'unlettered people', who were song is hinted at through presumably musically as well an as linguistically unlettered. Orality is, in fact, the most important defining element of folk song in opposition to art song, which is typically written down. In order to complete Sharp's definition, Lloyd (1967:16) suggests that in its natural state folk song a mouth-to-mouth transmission inflexible survival is poetry and music perpetuated by not by print', it is founded on certain principles but subject to personal variation; its acceptance and depends on how well it accords with the tastes, views and experience of the community (emphasis mine). Sharp's initial ideas, though undoubtedly valuable at the time when they were stated, have encountered disagreement among present-day scholars. To start with, his emphasis to Russell on oral transmission as a defining element of folk (Grove Music Online), with a song has to do, according controversial "post-Darwinian theory for its 212 evolution based "led to a on concepts of continuity, selection and variation", a view which preoccupation with identifying leaving aside songs that fulfilled such criteria", thus which could otherwise have qualified songs folk as Russell songs. (Grove Music Online) also points out that Sharp's conception "led to concentration by scholars many fact which has raised (Grove Music Online) acknowledges that traditional music is indeed oral in sense individual in a problems when dealing with tonal melodies. In spite of all these observations, Russell the the modal characteristics of folk melodies", on that it "does not exist singers or the on printed page but in the performances of instrumentalists, in the contexts of family gatherings, singsongs public houses, meetings of social clubs and other social groupings". The issue of authorship has also been ethnomusicologists, who have not been able to reach songs are composed by a community, or by an topic of debate for a constant an agreement as to whether folk individual, like art songs. Manzano (2001:44) categorically states that Alonso we must process abandon the idea that popular music is the result of a collective of creation. The creator is the individual, both in popular and cultivated music. In both from which she gets the individual creates within cases the structures used, and then the group a group accepts [or not] the created work (emphasis mine). Not only does Manzano Alonso state the individual authorship of folk also emphasises the professionalism of folk influence exerted from the popular by folk song and art widely accepted view that folk song song composers, an song composers are had to have emerged from musical professional authors, and vice taken from song composers as versa. sources well songs as but he the mutual idea which gets away uncultivated: "European created by more or less Professional authors used musical elements popular music" (Manzano Alonso 2001:33). According to Manzano Alonso's statement, it is not authorship that establishes the radical difference between folk song As he (2001:34) points out, "folk and art have co-existed for many song. centuries. The former is oral and while the latter is written and has left those elements behind Summarising, what are some the formal differences between song? I have already mentioned the presence 213 uses a song and art song archaic elements, centuries ago". folk song and an art of certain archaic elements in the important difference between the two former. But the most quality of folk with the oral song. Given that folk is faced with the need to create composer people. Thus, given that composers any is not written down, the song short form, which a object of oral transmission requires brevity several times in the same Linked to the folk song, well song, brevity of forms are the formal which have to do with simplicity, 'Archaic' elements are a - a short space, it is important to note are normally repeated clear or structural characteristics of narrowness of melodic archaic character of their musical those which cannot be found in art song five centuries but have nevertheless survived Manzano Alonso's be remembered by with different lyrics. the afore-mentioned as can of popular music only develop their ideas in brevity has to do with the musical structures, which that subgenres has to do song configuration. in the last four through the centuries in folk (2001:33-34) words, "those elements are range, as song. or In (i) modal melodic systems, (ii) limited melodic range, (iii) instabilities and chromatic movements, (iv) freedom in the development of the melodies, (v) irregularities in the rhythmic formulae". The issue of modal melodies is gives its name to and thus changed stylistically like a simple the whole song by genre, never over a tricky problem, ceased to be the centuries, an art-composer so sung that a as the Volkslied, which by all classes in Germany, Volkslied of 1820 sounded of that period. The English folk archaic to its collectors around 1900 because it had been song sounded forgotten by educated people and thus had scarcely changed since the late Middle Ages. The happened with Spanish folk song. What Manzano Alonso shares with traditional that folk song has its origin in rural linked to the lives and costumes of areas: ethnomusicologists is the idea "Popular music of oral transmission is mostly, though not exclusively, rural people" (2001:29). This link to the lives of people has to do with the function of folk While cultivated music and a same enjoys a time and space which we call 'concert', a song. performer hearer, popular music has always been integrated in people's lives, at any time, and has never had specific protagonists or receivers - with the exception of certain examples of instrumental music. Nevertheless, with the loss of rural life and the coming of urbanisation, the relationship between music and function has somehow been lost (Manzano Alonso-Alonso 2001:63-65). 214 further, not uncontroversial, characteristic of folk A song opposed to art song as is the fact that most of it is unaccompanied, that is, performed by sometimes a solo singer - accompanied by simple percussion instruments. Russell (Grove Music Online) observes that, although "unaccompanied singing has been the most common form of traditional singing that has been recorded during the past century and this does not presuppose organised into musical stanzas which share the certain amount of variation a 4.3.2. it directly affects the to which the process lyrics, also organised into very important for my of text-setting in different folk traditions. repertoire has been collected from the nineteenth song century in what we call 'folk song collections' used to transcribe the folk songs the controversial aspects are melodic contour and structure - same Transcription and classification criteria good part of the folk A - is that it is strophic, that is, it is again. This characteristic is stanzas, are set over and over purposes, as half, that it had always been the case". The last essential characteristic of folk song often with a that make up in the study of this reluctant to render folk song - Spanish cancioneros. The notation these collections is just another one of Traditional ethnomusicologists genre. transcriptions in music notation, arguing that this type of notation does not have enough means to account for the innumerable melodic and rhythmic nuances faithful to the original (1984:13) observes in of folk songs a song. is to rely completely consequence, as a basic a on into conventional musical notation, they must version. As of the tune, a can only a serve idiomatic original, for whereas conventional music notations can show the basic essentials really accurate scientific notation of the performance would mean using some source singer's kind of continuous graph tracing of the sound which could follow all the subtleties and In sung the researcher has to accept that ordinary music notation an of remaining field recordings. As Parfrey certain amount of 'unfaithfulness' to the guide to the way short introduction to Kennedy's (1984) compilation, when researchers render collected songs acknowledge It is obvious that the only idiosyncrasies of original performance by the traditional singer. spite of all the alleged unfaithfulness of folk bear in mind that the ultimate song transcription, it is important to goal of transcribing popular music is to make it 215 possible for the researcher to extract conclusions about its basic elements, namely, its its intervals, rhythmic organisation and melodic structure. As melodic system, (2001:232) emphasises, Manzano Alonso Although a live performance and/or direct recording will always be the most valuable documents in order to is analyse only through the analysis of a written classified, analysed and related to a score specific type of music, it that songs can be studied, another. one In sum, traditional music is transcribed to be analysed in terms of melody and texts, and not necessarily to be performed. collections used in In the two song Manzano Alonso (2003) -, my dissertation - Kennedy (1984) and transcriptions have been made departing from recorded documents, according to the following criteria: a. Each song are b. c. chosen to be The to is written as if it was to be sung, so the register and key signature easily intelligible. key signatures used keep the melody's character. There is use Time a tendency not complex key signatures. signatures are used like in art-music. There are no open bars or suppressed bar lines. Time signatures tend to be appropriate to the basic rhythmic structure and the tempo of the piece. d. Melodic variants: when the melody changes, the transcribed melody is the most frequently used for the Most ethnomusicologists follow classify oral poetry. The most song. common a classification which has also been used to categories would be, according to Finnegan (1977:12-13), the following: a. Epic: narrative character, long. b. Ballad: sung, c. Lyric: non-narrative, c.l. Love narrative poem, shorter than epics, concentrated lyrics. c.2. Psalms/Hymns. c.3. Dancing c.4. Political/Topical c.5. War songs. verse. songs/ Initiation songs. 216 on one episode. Spirituals/Laments. c.6. c.7. Work songs. c.8. Lullabies. classify the In order to used corpus mixed criterion, a of songs analysed in my dissertation, the collectors taking into account not only the functional aspect of the which is the basis of Finnegan's classification, but also the musical aspects songs, and the ones dealing with the contents of the texts. importance of lyrics in the study of folk song 4.3.3. The Nearly all the popular literature written before the twentieth century that comprises the song rather than the texts or was the beauty of the melodies that Despite this undeniable fact, there is still organisation that studies element, linguistic element and neglected been area or of expertise, a or no literary had a musical The musical side of songs has often one. pay attention to their that is, lyrics. Among linguists, both diachronic and synchronic, the as an string of sounds has seldom a text or a essential step for the study of the history of of its prosodic characteristics at protection of the lyric side of songs a a specific specific point in time. The scientific in contrast to the neglecting of their music, together with the fact that throughout history the mechanisms for have songs discipline in present-day minimised by literary scholars in order to acknowledged language the first priority, unitary wholes which comprise songs as in-depth analysis of the musical setting of been was performance styles of the singers from whom the academic a know of lyrics. However, Russell (Grove Music Online) observes that "[f]or folksong collectors it been notated". we song transmission proved to be naturally more efficient for texts than for melodies, has resulted in the unsurprising outcome that we have kept As a consequence, song as a many more popular lyrics than melodies. scientific studies have not paid much attention to the study of three-dimensional phenomenon, with a text, a tune and the result of the interaction between those two. Ethnomusicologists have always emphasised the importance of lyrics in the genre of folk song. Kennedy (1984:10) questions the musical status of the songs, saying that the tunes "are nearly always completely subservient to the story or emotion, rather than forming the basis of any great musical achievement". We could say that the importance of the study of folk songs 217 from a linguistic perspective has to do with the fact that their musical side literary side in such was that the interaction between certain characteristics of the a way language and their musical setting respond to a series of unified rhythmic criteria, presumably cannot be violated gratuitously. The fact that the most salient which is their strong rhythmic vitality points at characteristic of folk songs connection between a series of characteristics with a) linguistic simplicity for song, the two genres regard to lyrics, namely successful communication a b) the most commonly used devices as deep a language prosody and musical rhythm and metre. spite of the apparent differences between art and folk In share neatly linked to their linguistic and are the most elementary and powerful, such repetition, antithesis, hyperbole, irony c) the use of linguistic formulae - starting, greeting - is pervading d) the style of versification has to do with the use of simple stanzas and rhyme. 4.3.4. Text-tune As relationships in folk song repeatedly mentioned in this dissertation, elements, namely text and tune. The place in different process a song we deal with the differences between art song and folk analyse the principles of text-setting in general. In art art-music, what usually happens is that the - be it a well-known text, a text text exists prior to the music and the music is created process do not know whether a text song case the has in a an same existed melody, and vice specific text which can aria, for instance. The and vice versa, a versa. of a pre¬ herself of the previous or a cases, the melodic declamation of the long before the melody was created and, we more not bi-univocal: different texts can be Thus, it is not accurate to think that each only be associated with that tune, same text as would be the often appears joined to different melodies melody is adapted to fit different texts. The possibility of interchange between texts and melodies does texts as a genres melody for composer we in popular music is completely different. Very often important, the relation between text and tune is set to any but also when and other vocal created by the specifically written in order to be set to music. In speak. The song song composer creates a text text, so to composite of two main which is worth taking into account not only when take existing text a of connection between the two elements can ways, is has been created for a not mean, however, that none of the specific melody. We 218 can presuppose that each text was originally associated with 'its tune'. Nevertheless, it given the oral character of folk song, as dissociated and re-associated with connection tunes and texts, that lead onto the creation of new frequent source that, time passed texts and melodies started to be respectively, until the original completely lost. Sharp (1954:25-26) explains this was variation stages A new seems correct to assume process in terms of songs: of variation arises when the singer, having partially forgotten his words, has substituted corrupt and unmetrical lines. The attempt to adapt the tune to these irregularities will often lead to the inventions, unconscious of Another] source tune to new course, of interesting melodic changes [... of variation springs from the attempt to adapt words of slightly different metre [...] If, happen, the metre of the word is not exactly the as must same as an old frequently that of the tune, they will alter and adapt the latter to meet the This possibility for interchanging texts and melodies in popular music is essential factor to be taken into account when gives rise to a very The text to any an analysing text-setting. This factor adjustments that popular music solves in a very manners. question arises of why it is relatively folk tune. The reason easy to set virtually any given folk for this is metrical: The usual stanza of poetry of conditions. singular relationship between text and tune, often comprising series of frictions and lack of idiosyncratic new contains four measured lines, not necessarily equal length, though forming a just balance; normally, of four phrases, the points of division being marked off by means of cadences (Sharp 1954:72). (2001:66) explains, following Sharp's statement, "the As Manzano Alonso this is that the poetic analysis of the case measure to a musical cycle explains, stanzaic English traditional or tune song is also based on same syllabic and the stanza and a sung that is repeated. As Russell (Grove Music Online) songs usually have for of most texts is the same". As will be observed in the studies (4.3.6.), most Spanish texts have the stanzaic measurement. reason four-line isometric structure with (usually ABAB, AABB, or a simple rhyme scheme ABCB), each line organised into alternate 219 lines of four and three feet (ballad or common feet) is also prevalent. Some lines of four others have a chorus of the later songs formal chorus section, more metre). Long metre (four contain refrain lines; notably the monumental parlour ballads. Such items demonstrate a more complex and varied structure. only exceptions to the interchangeability of texts and melodies The which peculiar poetic forms and they individualise each illustration to this, in the Alonso 2001:67). As in dissertation, 70 an seguidillas - can can be no corpus from the rest (Manzano of Spanish songs a very be divided into eight plus eight syllables analysed -, - or 20 per widely-used stanza in Spanish poetry and music, composed of four lines alternating and 10 per cent are song more cent of the texts are octosyllabic quartets per hexadecasyllabic lines which cent are refrains, usually less flexible than stanzas because they normally have are my are seven (or six) and five syllables, rhyming -a-a - really varied poetic measurements and stanzaic formulae. There doubt that the overwhelming presence of the octosyllabic quartet (cuarteta octosilabica) and the seguidilla in the lyrics of the conditions the rhythmic songs patterns and the melodic developments of the songs. interchange phenomenon, which is nowadays exclusively associated This with folk song, was the also accepted eighteenth century for were music whole operas was new as common practice in art song. It was common in words to be written to existing operatic arias; there (called pasticci, 'jumbles', Grove Music Online) in which old used, sometimes with its original words, and with new words on other occasions. 4.3.5. The choice of The a specific folk song corpus following two sections contain of the two folk song a brief summary and explanation of the contents collections which compose the corpus analysed in this dissertation. 4.3.5.1. Manzano Alonso's Cancionero de This volume is the fifth in a transcribes hundreds of songs these six volumes is devoted to Burgos collection of six volumes where Manzano Alonso collected in the Spanish region of Burgos. Each of a specific topic related to the function of the The fifth one, entitled Canciones del ciclo anual y 220 songs. vital ('Songs of the annual cycle life'), comprises all those of are liturgical not linked to or songs which refer to and complement traditions which religious rites. These all profane are songs related to the cycle of field works and/or fiestas. In this dissertation, I have analysed 444 terms songs in of the OT metrical constraint MATCHSTRESS. 4.3.5.2. Kennedy's Folk songs of Britain and Ireland This is unique volume which contains 360 songs collected by Kennedy throughout Britain and Ireland. It is divided into sixteen parts, classified following two different a criteria. On the the II one hand, there region where the songs were are six parts grouped according to the language and collected. Thus, we find I Songs in Scottish Gaelic, Songs in Irish Gaelic, III Songs in Welsh, IV Songs in Manx Gaelic, V Songs in Cornish and VI Songs of the Channel Islands. On the other hand, there is collection of songs of the in British English, which corresponding songs. In the group of are a vast subdivided according to the function songs in English find the following we subgroups: VII Songs of Courtship, VIII Songs of False love and true, IX Songs of Seduction, X Songs of Uneasy Wedlock, XI Songs of occupations, XII Songs of country life, XIII Songs of good company, XIV Songs of diversion, XV Songs of newsworthy sensation, XVI Songs of the travelling people. In this dissertation, I have analysed the 239 songs in English. As with the Spanish metrical CON MATCHSTRESS to this corpus, statistical chart which would work 4.3.6. I have applied the in order to be able to draw a indicator of tendencies. Analysis of the folk song corpus: overview and general observations Given the high number of songs contained in the chosen the purposes of this dissertation I have opted for neatly shows the stated in the of as an corpus, ways a corpus - a total of 683 -, for clear-cut analytical method, which in which music and lyrics interact in English and Spanish. As introduction, one of my objectives is to define the nature and properties linguistic stress in each of these two languages, which implies observing how syllables are interaction aligned with musical beats in between MATCHSTRESS to song. The best way to observe the syllables and beats is applying the metrical constraint each of the 683 songs in my corpus.61 After applying conclude, first of all, that both the MATCHSTRESS to both song collections, Spanish and the English show lexical and grammatical mismatches. However, songs we can 61 The detailed analysis of the mismatches in the 683 songs appears at the end of this dissertation, Appendix IV (Spanish songs) and Appendix V (English songs). 221 as the number of stress mismatches in lexical words in the nouns, verbs, adjectives, adverbs - lexical stress is mismatched a encounter a group so of disyllabic that the musical beat falls mismatches. The mismatched both of which syllable, end with a <y> correspond to the phoneme III. This is (/<—>/), which might be The nouns which on the second beat of of a musical cases where a - a are we nevertheless set fact that gives rise to lexical or its spelling variant <ie>, peculiar phoneme, known as unique characteristics, namely (i) it is neither a why it can be stressed when it is set to music. undergo this type of mismatch are pen-nie, where -nie falls and therefore, 6/8 bar, which is also the last beat in the song, on the second beat of a 6/8 bar, which is line; coun-te-rie, -rie falls corresponds to the end of a a reason phrase; 1 a-dy, where -dy falls also the last bar in the beat of that is, long and (ii) it is tense. In other words, this phoneme cannot be reduced to nor schwa with initial stress, which the second nouns - grouped into three categories. First, be nouns "neutralised /t/", which shows two short corpus is comparatively low. Moreover, the can on English a on the first beat of a 6/8 bar, which line; North Coun-te-rie, where -rie falls 6/8 bar. In all these cases, the second on mismatches coincide with the end of the corresponding musical phrase. This type of mismatch is idiosyncratic, to the extent that it word a can be said to be a style or genre mor-ning, where -ning falls minim and is, once on phrases Second, there the first beat of a 3/4 bar, which corresponds to again, the last figure in the idiosyncratic lexical mismatch which the end of musical can are two occurs the second beat of occur on makes it a In sum, this first type of in disyllabic words with initial stress at instances of mismatches occurring in compound nouns, beat, and therefore gives rise to on song. be called "Country mismatch". namely shoe-ma-kers, where -kers falls falls marker. Something similar happens in the a very on the last beat of a 4/4, which is a secondary gentle mismatch; game-kee-per, where -per 6/8. In these cases, the fact that the mismatch does not the main beat of the corresponding musical bar minimises its impact and acceptable. Last, there is only one instance of mismatch which be-came, where be- falls on the third beat of acceptable mismatch. 222 a occurs on a verb, namely 3/2, which makes it a relatively Summarising, lexical mismatches them are rare in the English corpus. Most of belong in the class that I have called "Country mismatch", where a tense vowel that cannot be reduced is made to fall on a strong beat. As mentioned above, this is a style marker. Moving on to grammatical mismatches, these English mismatches in the corpus, are more numerous than lexical and respond to what Giegerich (1978) calls "Hierarchy of stressability of function words". According to this, certain function words under can be stressed without further consequences, practically any reproduced below -, while others cannot be stressed circumstances. According to Giegerich's scale of stressability prepositions and conjunctions are never stressable, followed by The number of auxiliaries, pronouns, modals, wh- words and demonstratives. mismatched words in the corpus appears under the corresponding category: less stressable more stressable ► others < Count: PREP, CONJ < last AUX < PRO < MODALS 49 As 7(5) 12 we 23 (2) < 2 WH- < DEM 2 3 observe, the number of mismatched function words in each can category coincides, to a certain extent, Giegerich's scale. Prepositions are frequently mismatched auxiliaries ever (49 instances), followed by pronouns (25), conjunctions (12) and (12). On the other hand, demonstratives, wh- words and modals mismatched in the corpus. are hardly This could also respond to Giegerich's (1978) second hypothesis for the stressing of function words, the so-called "Rhythmic Stress Change", whereby a monosyllabic function word can be assigned stress (Table 30). [+DEM] [+WH-] [+MOD] [-stress] -> [+stress] / [-stress] # [+PRO] <AUX>a [+PREP] [CONJ] Table 30: Rhythmic Stress Change 223 # [-stress] <-AUX>b In the English folk song corpus, most mismatched function words are monosyllabic, with the exception of the prepositions unto, into, and the As same the adverb until auxiliary verb could with the negative particle attached, couldn't, each of which appears in the over, a mismatched only once. English folk Number of songs: a as one statistical If the song corpus. it counts song, below is summary, same summary of all the mismatches observed mismatch happens more than once in the mismatch only. Songs with mismatches: 62 239 = 25.94% Total number of mismatches: 118 1. On lexical words: 8 = 6.78% b) Full verbs: 1 a) Nouns: 7 2. On grammatical words: 110 = 93.22% a) Prepositions: 49 b) Conjunctions: 12 c) Pronouns (rel.): 23 (2) d) Determiners (art./dem.): 13 (2/2) e) Auxiliaries (mod.): 7 (2) f) Particles: 6 Spanish The corpus is much syllable/stress/beat mismatches. There richer are nearly grammatical mismatches, although the latter several times in the are the most same song. and complex as many are more in terms lexical mismatches frequent and often Prepositions, determiners, conjunctions and widely mismatched grammatical words, while to nouns as appear pronouns and verbs are, by far, the most commonly mismatched lexical words. The general tendencies are the English same as in the Spanish corpus. Unlike in the accounted for as corpus, English but the number of mismatches is much higher in the corpus, where lexical mismatches idiosyncratic, in the Spanish corpus can be grouped and they do not respond to any single stylistic criterion. Indeed, the act of mismatching lexical words constitutes the style. By observing the behaviour of mismatches in this one could easily conclude that Spanish folk mismatches. This folk song to brings on an which most song corpus (see Appendix IV), allows for all kinds of lexical important issue, which has to do with the subgenre of songs Appendix IV, the majority of the in this songs corpus belong. As can contained in this specific 224 be observed in corpus are dance songs, that is, they were traditionally performed in order for people to dance. This implies that both the tempo and the metre of the definitely more so singer. As a than in result, we can presume languages allow for mismatches are bound to be rigid that words will be mismatched mismatches in dance more Spanish The mismatches found in the on even more of song. However, this does not necessarily songs; corpus are mean that in English, lexical generally disallowed regardless of what function the Number of songs: - ballad, which is performed ad libitum, often by a solo a occasions than in another type all song are song fulfils. summarised below: 444 Songs with mismatches: 292 = 65.77% Total number of mismatches: 945 On primary counts: 920 On secondary counts: 25 1. Lexical words: 449 = 47.51% a) Nouns: 264 b) Full verbs: 128 c) Adjectives: 36 d) Adverbs: 11 e) Numerals: 7 f) Tonic 2. Grammatical words: 494 = pronouns: 4 52.28% a) Prepositions: 156 b) Conjunctions: 87 c) Pronouns: 69 d) Determiners: 154 e) Auxiliary verbs: 12 f) Relative pronouns: 15 g) Particles: 2 In the next two one extracted from the These analyses constraints are sections, I develop an in-depth analysis of two case studies, English intended as corpus, and one extracted from the Spanish detailed applications of the metrical and grouping explored in the literature review, which will lead to the nature of stress and the corpus. syllable in the two languages. 225 an informed view of 4.3.7. Case studies One of the difficulties in underlying such choice. Both criteria with songs, songs from vary a without are any songs song for the case studies had to do with the collections comprise dance and non-dance slow tempo and songs with a fast tempo, and the time signatures The chosen songs have a number of common characteristics, song to song. namely (i) they sung choosing two strophic (ii) they are not dance tunes and they are meant to be songs, accompaniment (ii) they are written in binary time signatures (4/4 and 2/4, respectively), (iv) they have roughly the same number of bars (16 and 15). In both cases, * the constraints applied for the analysis of text-setting are MATCHSTRESS, LAPSE, FILLSTRONG, PARALLELISM and SALIENCY. 4.3.7.1. Case study I: 'The farmer's boy" "The farmer's boy" is and Ireland, Britain a strophic song collected in Kennedy's (1984) Folksongs of within the subsection entitled "Songs of country life". According to Kennedy's (1984:547) introduction, "The farmer's boy" popular throughout England, although the version collected by Kennedy song incorporates a tune originally from Galloway, in Scotland. The divided into four four-line stanzas, in Table was a very each of which is followed by song a 31) with two variable lines at the beginning which work the narrative of the is a narrative refrain (in italics as a link between previous stanza and that of the refrain itself, and two invariable lines at the end. As can number of be observed in the scansion of the syllables per line goes lyrics given in Table 31 below, the from six to nine. There synalepha (underlined in Table 31), which in English is a Keyser 1971:71) with non-phonological status. In this two syllables in little case of little, c 62 is a poetic device with no instances of correspondence rule (Halle and as one are two case, the realisation of the phonological status.62 The pronounced /'litl/, corresponds to Fabb's (2002:10) Non-projection mle (quoted in section 3.3.3. and repeated here for convenience): non-phonological status of Non-projection rule b is arguable since the degree of sonority of phonological segments is taken into account in several prosodic processes. Non-projection rule c corresponds to the invisibility of certain syllables with a syllabic consonant as their nucleus in the stress assignment process. As an illustration to this, observe how in a word like badminton, antepenultimate stress can only be expressed by taking the last syllable, with a syllabic [n] as nucleus, as invisible or non-projected. In other words, the syllable /on/ does not count as such for the purposes of stress assignment - if it did, the stress would fall on the penult. The 226 Non-projection rule its nucleus or one which has c: of the as Optionally, do not project a syllable which has following sonorant consonants: [1], [r], [m] its nucleus the weak vowel schwa followed by or as [n], one of these sounds. hand, the synalepha between the article the and the first syllable of the On the other adjective eldest corresponds to Fabb's (2002:9) non-projection rule b (quoted in section 3.3.3. and repeated here for convenience): Non-projection rule b: Optionally, do not project on a a syllable which ends vowel, when that syllable precedes a syllable which begins on a vowel. Non-projection rule b is equivalent to synalepha proper in Spanish, which uses the linking of vowels in order to bring the actual number of syllables in the line to the right number in metrical terms. In this case, given that the third line of the first stanza that the third and the third line of the second stanza have six lines, line of the third stanza will have six lines, too, something which is only achievable we can presume by the realisation of the synalepha between the and eld- in eldest, whereby the schwa in the is elided and the sound /5/ becomes the onset of the first realised syllable in the eldest, /'deldist/. as The lines in the poem are structured into alternating iambic tetrameters and trimeters, that is, odd lines consist of four feet with the structure weak-strong per line, with three minor variations, while some feet with the same internal even lines are arranged into structures of organisation weak-strong. This structure corresponds to the so-called 'common metre' in Church hymns. A minor variation in the structure of iambic tetrameters is the substitution of an anapaest (xx/) for an iamb (x/). In spite of this, all the odd lines in the poem tetrameter, which in some cases entails the promotion of certain unstressed syllables - to conform to the structure of iambic (4th line), I and you (5 th line), will and me (6th line), if and me (9th line), this (12th line), her (18th line), what (19th line), for (28th line) certain stressed syllables is a syllables - ask - as well as the demotion of (5th line), give (6th line). Promotion and demotion of thoroughly standard rhythmic device in English poetry and can be compared to the phenomenon of stress shift in English prosodic phonology. As for rhyme, even lines (trimeters) rhyme, while odd 227 ones (tetrameters) do not rhyme. Syllables Line The went sun Across yon down, beyond yon hills and lame, a 6 boy there came Feet (accents) x/x/x/x/ 8 dreary moor When weary Rhyme a 9 — x/x/x/0 x/xx/x/x/ x/x/x/0 Up to the farmer's door. 6 May I ask you, if any there be 9 — x/x/x/xx/ 6 b x/x/x/0 That will give me To plough And to be employ and sow, to a 8 me employ At break 8 mother's left 8 With her five children small worse c - b 6 plough and sow... And what is - 6 employ my c 9 of day I'll trudge away My father's dead, - 6 wintry blast? Elsewhere to seek b 8 shelter me, till break of day From this cold - 7 thing I have to ask Will you To and mow, farmer's boy? And if that thou won't One reap a 8 e - 8 — If thou wilt me employ 6 b course The of time, And his 8 grew a man 6 boy the house now has 8 daughter for his bride boy that was, the farm 7 now has 8 He thinks and smiles with joy Of the lucky day he For to be a e — good old farmer died And left the The he came that 6 way farmer's boy. — f — f - b 9 — 6 b Table 31: Scansion of "The farmer's 228 xx/x/x/x/ x/x/x/0 x/x/x/x/ x/x/x/0 x/x/x/x/ Though little I be. I fear not work In x/x/x/0 - 6 and sow... x/x/x/x/ x/x/x/x/ I'm the eldest of them all To plough xx/x/x/0 - 6 for mother still x/x/x/x/ x/x/x/0 x/x/x/x/ x/x/x/0 x/x/x/x/ x/x/x/0 x/x/x/x/ x/x/x/x/ x/x/x/0 x/x/x/x/ xx/x/x/0 x/x/x/x/ x/x/x/0 xx/x/x/x/ xx/x/x/0 boy" lyrics. The fact that the number of not syllables changes from line to line in English is unimportant since it has musical implications. If the premise for text-setting is that, as a general rule, one syllable corresponds to higher number of syllables will render means be that the melodic and an or example of this, observe that scanned version of the lyrics eliminated in the musical assigned to a in most songs, a bar, which - the relatively high a delete syllables and, accordingly, note one of the two synalephas realised in the between the and eld- in the one 20th line - is setting of the text, where each of the syllables the and eld- different note value. Little, syllable assigned to per a rhythmic contour of the first stanza will not necessarily preserved throughout the whole composition. There is, therefore, values. As at musical note value, then higher number of values a degree of freedom for the singer to insert is one single note value, on as seen the other hand, is kept in Figure 1 below.63 the only factor that must remain constant is the as a In this occurrence single song, as of accents equal intervals, something which is automatically achieved in vocal music since there is always when accented a strong accent on the first beat of each bar. The difficulty arises syllables must be made to correspond to primary and secondary counts, that is, salient beats, as is the case in challenged at that level. In this sense, English song. the musical setting of the text is perfectly aligned with the scansion given in Table 31 above, in such promotion and demotion of syllables beats respectively, while no The singer's skill is a way that all the cases of made to correspond to strong and weak are other syllables are mismatched. We can thus say that the agreement between verse prosody and music rhythm and metre is virtually total in this song 63 and, presumably, in English folk In order to of the four song. complement the written analysis of the songs, I have included a CD with the recordings studies two folk songs and two art songs - examined in my dissertation. case - 229 The And sun went if that My fa ]n - down^ titer's course of boy there came Up day From break of 12 m- m mo - ther still Pm the house now has And his be - thou_won't dead, my time,, he yond yon me em- hills. A - cross yon drear - y moor ploy One thing I have to ask mo-theds lefl_With her five chil-dren small grew a man. The good old far - mer.-4ied the far-meds door May I ask you, this_ cold win-try blast? At breakof day eld_ dest of them all Thoughlittle 1 be, to daugh-ter for ploy? To plough and sow, ploy To plough and sow, ploy To plough and sow, joy Oftheluck - y day his bride and mow And to and mow And to and mow And to he came that way For to to to to reap reap Figure 1: Musical As be be be be reap a a a a Will And far-meds far-mods far-meds far-meds explored in 4.2., in text-setting there what_ And, left is worse the boy farm now has He thinks and me for the em- smiles with boy. boy. boy. boy. of "The farmer's boy' score y and lame, a youshel-ter me, till wear- if an-y there be That will give Til trudgea-way Elsewhere to I fear not work If thou wilt boy that was,the The When 64 are two groups of constraints at work, namely metrical and grouping constraints. The first metrical constraint, MATCHSTRESS, predicts that the rises and falls of stress within the line matched to the rises and falls of the metrical pattern Nevertheless, it is violated instances of mismatch - - First stanza: Second stanza: line), this There - There are several as seen in (181)), me (1st line, see (181)), you (3rd (2nd line), what (3rd line), the (4th line). the (7th line), he (7th line), to (8th line). Fourth stanza: - song. (4th line), elsewhere (7th line). - - 64 //(1st line, Third stanza: her - several occasions in the (4th line), be (8th line). - are (Hayes and Kaun 1996:10). primary counts: on to on are also mismatches First stanza: you on (5th line), be (5th line), me (6th line) Second stanza: thou Third stanza: secondary counts: (1st line), of (4th line), be Fourth stanza: for me (3rd line) (5th line), me (6th line) (4th line), that (7th line) collection, Kennedy underlays only the first stanza and the first refrain of the song, while the are written out beneath the score. As a consequence of this, there are cases where the alignment of melody and text can be ambiguous. I have aligned all the stanzas following native intuitions of text-setting, which does not mean that my setting is the only possible setting for these In his other stanzas specific stanzas. 230 (181) Mismatches in the first line of the second stanza The sun went And if that Curiously, mismatches down, are not yond me yon em- hills. A ploy One case of what I call 'duration mismatches', where values, not only the position of the beat within the bar, signal a mismatch. Let us observe on a - always signalled in terms of musical accent, but also in terms of note value. This is the note be thou_won't (182), where the function word to, realised main beat quaver plus a content word assigned to - stress mismatch crotchet -, - but is made to in speech, not only falls as correspond to something which renders it a more a long note value - a salient syllable than the boy in the bar before, which, in spite of containing a quaver. In this sense, we could say a diphthong, is that the mismatch of the grammatical word to is doubly salient. (182) Duration mismatch ft J J' J —i— --J^ —• boy there came Up_ to of day From this as the J O the far-mer's door cold win- try blast? break Duration mismatches such K 1 L—J one above contradict Hayes and Kaun's (1996:16) Syllable Duration Rule (quoted in section 4.2.2. and repeated here for convenience): Syllable Duration Rule: reflect the natural phonetic durations of syllables in the number of metrical beats they receive. The constraint, FILLSTRONG, predicts that the four second metrical strongest positions in the line are filled with syllables (Hayes and MacEachern this constraint is violated only at the end of each couplet 1998:492). In this song, since all the strong positions position of every even are line - trimeters, not tetrameters. Let filled with as us a syllable with the exception of the fourth explained above, have a even lines are structured as look at the grid notation for the first two 231 lines (183), where we can observe that the fourth position of the second line is left empty: (183) Non-filling of strong positions X X X X X X X X X X X X The sun went down be- yond yon hills X X X This X X X X X X X X X A- cross yon drea- ry moor phenomenon is linked to truncation, which in this case, even verse X X lines and musical - X means X 0 that the last foot in the line - is left unfilled. In terms of the correspondence between grouping, this phenomenon gives rise to systematic non- correspondences between prosodic constituents (lines) and musical constituents (bars and phrases). In musical terms, the violation of the second metrical constraint entails not only that the unfilled strong position is forced to be linked to the previous one(s) - in this song, the end of each musical phrase is marked with also that the very in that line - the very from the next "theft" of last beat might be left without last beat of each even any a long note value - but corresponding prosodic position bar in this song is filled with a syllable poetic line. This is what Hayes and MacEachern (1996:16-17) call positions (184), which, in English, happens only to the left: "it is only in the leftward direction that the positions are there to be stolen" (see section 4.2.3.). (184) Theft of positions (where { } signal x a line, and [ ] signal a bar) x x x x x XXXXXXXX X {A- When] [cross yon drea- ry 232 moor} 0 FILLSTRONG is closely related to grouping constraints. This is the reason why it only violated when the grouping constraint that regulates the saliency of can constituents - related to truncation metrical constraint that - is active. In any other case, a violation of the requires the filling of strong positions would render the text- setting instance unacceptable. The third metrical constraint, *LAPSE, states that sequences syllables are placed in the interval between the line should be avoided. In this song, any two in which no of the four strongest positions in the constraint is violated whenever the previous constraint is since weak positions between the third and fourth positions in even lines when the are not filled. A lapse comprises lapse of a single syllable is regarded more as normal; it is only than three syllables that the text-setting instance is unacceptable. (185) Presence of lapses x x X X X X xxxxxxxx A- cross yon The second group drea- ry of constraints is that of grouping constraints, subdivided into SALIENCY and PARALLELISM. SALIENCY is of what 0 moor directly related to the action Hayes and MacEachern (1998:476) call "rhythmic cadences", which characteristic MacEachern grid placements of the final syllable or two are the of the line. Hayes and classify cadences into four main types, namely G, 4, 3 and 3f (for complete explanation of these terms, see a Hayes and MacEachern 1998 and section 4.2.2.). As explained above, SALIENCY works at the level of the couplet, which is systematically made salient by the insertion of a 3 cadence, that is, the second line in each The have couplet has three strong positions filled, while the fourth one rhythmic cadence is reinforced by rhyme, which is at work in a look at how this works in the first is left unfilled. even couplet of the first stanza (43): 233 lines. Let us (186) SALIENCY X X X X X X X X X X X X X X The sun went down be- yond yon hills X X X X X X X X X X X X A- cross yon drea- ry moor jij. J' jjJ The went sun (4) down, be - PARALLELISM works at two iJ yond r yon 73 J hills. A X - X 0 iJ j j j cross yon drear - y (3) ij. moor r When levels, the quatrain level and the stanza level. At the quatrain level, it states that the cadences that end units of the maximal analysis this case, must true the couplet are in always identical (Hayes and MacEachern 1998:17) and rhyme with each other (Hayes and MacEachern 1998:17), something which is in question, where the structure of couplets is always 43 and the two for the song 3s in each stanza in - - rhyme. At the strophic level, positional parallelism (Halle and Dell press) establishes that settings to the onsets with respect to same tune the grid. This constraint is violated second, third and fourth stanzas, where there is the deletion of a pitch. (187) Positional PARALLELISM jiilr May At I ask break of have the if day Til you, r 3 there be That trudge a - way Else an-y 234 a on same distribution of the third line of the deletion of a syllable, which implies The went sun be- down X X X X X X X A- cross yon drea- ry moor we - X although at the prosodic level there is a (3) 0 song anticipates the need of where melismata happen violations of PARALLELISM cannot be in X observe that the melodic contour of the syllable insertion at certain points see (4) hills X X X general, yon X X In yond in such a way that strictly observed at the musical level, clear insertion of extra syllables, as we can (188): (188) Melismata ^ ^ J boy there came Up break of day From mo - ther still I'm the house now has And his In conclusion, violate the ^ to this eld J J U. the far-mefs door cold win - try blast? dest of them all daugh-ter for his bride alteration of the musical structure is made in order not to any matching of stresses and beats or, at least, to minimise the number of violations of this metrical constraint, to which all other constraints are This is clearly seen in the violation of positional PARALLELISM, which takes place in order to avoid mismatches between stressed sense, metrical constraints seem to mismatches bear the consequences 4.3.7.2. Case de a regulate text-setting in English, while grouping of the former. strophic popular de Burgos, aguinaldo syllables and strong beats. In this study il: "Alegria, caballeros" "Alegria, caballeros" is Cancionero subsidiary. en more song collected in Manzano Alonso's (2003) specifically, in the subsection entitled "Cantos la fiesta de los Reyes". In this type of festive 235 song, the singers were entitled by traditional custom to change the lyrics and adapt them to the of audience to whom they circumstances and the type singing. In terms of the were general relation between lyrics and tune in this specific it is worth noticing song, that, in spite of the singers' compositional freedom, the lyrics always remain by the tune, that is, neither the melodic contour constrained structure of the song can song in the notated to be changed in order to fit the lyrics, in such has four note values which word is intended exactly the score or per needed to fall on that bar. We with no added or la yes ya son ma - fia - na, fioes-ta-baenBe- ldn, se de-do pi - na - do? pri - Re-yes. merfies-tadel han de - ja-do las o /,Me quie - res lle-var a Figure 2: Musical score The fact that the melodic and prosodic the melodic and for the moment consequences - that, if a no matter observe this mechanism can song are set eliminated note values. - res,que su - - to - fio que se ce-le-braenEs - pa - fla. jas yhan a - pre-ta - doa co-rrer. jui-cio?Per-dd - na-me los pe-ca-dos. a - ni- dos, los Re pie- ron queel niquie-res de-cir, ni - no, con e- Los Re-yes ya son ve Los pas iQue me S has several a way bar, they will have to be kept untouched gri-aca-ba- Lie-ros,no-blefies-tadelos - the rhythmic (Figure 2), where the two stanzas that configure the same tune, A-le nor - ve A-le - of "Alegria caballeros"65 rhythmic contours have to remain untouched we could also say that the prosody determines rhythmic contour, but we will leave the matter of directionality aside - one of which is the observe that each note value of the melismata on preservation of isosyllabism. In Figure 2, corresponds to the seventh a prosodic syllable syllable of each line -, - we with the exception in such a way that the preservation of the number of values in each musical phrase naturally leads to the preservation of the number of syllables in each line. It is 'beat' means clap.66 In an the same as 'tactus', defined instance of 2/4 metre, the crotchet level. On the other which the most 65 66 as hand, common ones are the level at which as a make clear that listener would in "Alegria, caballeros", the tactus is located at a two 2/4 bar can crotchets comprise various note values, of combined, a series of four quavers, comprises two songs. I analyse the first one, which lasts 1:09. nowadays used as synonyms, the latter is better applied to music while the former is applied to music organised into bars. Track number 2 in the attached CD Although 'beat' and 'tactus' written before the bar era, necessary to are 236 eight semiquavers, or any combination of these. In this combination is four quavers or syllables fall itself a - on the beat - three quavers -gri- in alegria the beat in 2/4 is the crotchet -, bigger value, either realised on the the most plus two semi-quavers but have a or per common bar. Some smaller values than the beat whereas others, while falling same pitches (melisma) - -lie- in caballeros - song, on the beat, take musical pitch - -yes in Reyes - or on two -ha- in manana. Line Syllables Rhyme Rhythmic accents Alegria, caballeros, 8 - 3,7 noble fiesta de los Reyes. 8 - venidos, 8 - Reyes Los los la Reyes ya son ya son manana, primer fiesta del que se celebra en 8 ano 8 Espana. 8 a - a Alegria, caballeros, 8 - noble fiesta de los Reyes. 8 - Los pastores, que supieron 8 - 7+1 b 8 - aue el nino estaba han dejado las ovejas en Belen. 7+1 b Alegria, caballeros, 8 - noble fiesta de los 8 - 8 - v han apretado Que me con ese Me a correr. Reyes. quieres decir, nino, dedo pinado? quieres llevar a juicio? Perdoname los pecados. 8 c 8 - 8 c Table 32: Scansion of "Alegria 1,3,7 2,5,7 2,5,7 3,4,7 4,7 3,7 1,3,7 3,7 1,4,7 3,7 1,4,7 3,7 1,3,7 1,3,6,7 2, 4,7 2,5,7 2,7 caballeros". 67 In the English case studies, the third column corrresponds to the analysis of the foot structure of lines, which implies analysing their rhythmic accents. Spanish poetry is not organsied into feet, so for the Spanish case studies, I have only specified the rhythmic accents in each line. 68 Translation: 'Joy, gentlemen,/noble party of the Kings./The Kings have come,/the Kings come tomorrow,/the firt party in the year/that is celebrated in Spain./Joy, gentlemen.../The shepherd, who kew/that the boy was in Bethlehem,/have left the sheep/ and have hurried up./Joy, gentlemen.../What 237 As can be observed in Table version of the song 32, the eighteen lines that configure the recorded have eight syllables each. The main compression device used in is synalepha, both between two different vowels, in line 6 this song and between different kept in the vowels, in line 10 This song. song shows an - que by eliminating its last syllable, instance of as celebra en -, el. All the instances of synalepha are apocope, a compression device. As explained in section 3.2.3., word - in the case apocope much less common is the truncation of a of the metrically-conditioned disyllabic word primer in line 5, which would otherwise be a trisyllabic word, primera. It is the poem interesting to note that, in the scanned version, the accented syllables in fall roughly on the same syllables strictly following the rules of Spanish set to with certain minor variations -, thus prosody. However, when the lyrics are music, the placement of prosodic accents is disregarded in favour of the musical - verse - accentuation, which, in 2/4 metre, naturally falls on the first beat of each bar primary count crotchet or - and the second beat of the bar, corresponding to the second the third quaver disagreement between winning on over verse - secondary count. This prosody and music rhythm in the the former. In Table 33 we effect of total causes an song, with the latter observe the musically accented syllables (in bold), which often do not coincide with the prosodically accented syllables (in italics). you trying to for my sins.1 are tell me, boy,/with that straight finger?/Do 238 you want to take me to court?/Forgive me Line Rhythmic accents Alegria, caba/Ieros, 3,7 noble fiesta de los Reyes. 1,3,7 ven/dos, 2,5,7 los Reyes ya son manana, 2,5,7 Los Re yes ya son la primerfiesta del due se ce/ebra en Los pastures, que 4,7 Espana. 3,7 su/u'eron ciue el nino estaba han de/'ado las ovejas v 3,4,7 ano en 1,4,7 Be/en, 3,7 (1), 4, 7 han apretado a correr. iQue me quieres dear, «/no, con ese Me de do 1,3,6,7 pinado? 2,4,7 quie res llevar a juieiol Perc/oname los 2,5,7 2,7 pecados. Table 33: Musical accent Rhyme is systematically kept versus on even prosodic accent. lines, while odd lines rhyme. The resulting poetic form, the copla, is a very are left without widely used stanza in popular poetry and folk song. In order to make the above metrical and would be this possible to translate them into constraints. As chapter, the first group of constraints caballeros", MATCHSTRESS musical beats are in caballeros - occurs at seen in the introduction to the metrical level. In "Alegria, the constraint that requires that lexical stresses and aligned - is often violated. In the grid notation of the first two lines of the song, we can - grouping observations systematic, it -, observe that sometimes prominence falls while stressed musical arrangement ones are relegated to of the sentence - in noble,. 239 a on unstressed syllables secondary level by the (189) X x x X X x x X X A- le- gri- a X X x X x X X X X X X ca- ba- lle- X ble- x ros X X X x X x X X X X x X X X X X ta de los Re- fies- Second, FILLSTRONG is with X X x no- X X x never violated, as X yes all the strong positions are filled syllables. The violation of this constraint would render the setting of the lyrics unacceptable for alternative setting for the left empty. As a Example (190) gives an lyrics, where the first and fourth strong positions are native listener of Spanish folk a same song. result, the setting of the line becomes unacceptable. (190) X X X X X X X X X gri- a X X X X X X X X XXX X between the third and fourth strong of lapses as there many other X X ros are empty positions positions. Nevertheless, violating this constraint composition ill-formed. Indeed, it is the is violated in X lie¬ ba- ca- Third, *LAPSE is violated within stanzas, does not render the 's. X songs of the 240 same case tradition. that the prohibition (191) X X X X X > X rT 1 X X X gri- X X X X no- ble- X X X X fies- qualitatively different X X XXX ta de way violations of this constraint which the X 0 ros X X X X X X los X X X X X 0 Re- happen in a yes quantitatively and in English and Spanish. While in the former all the are 'translations' of the verse scansion into the musical implies that those mismatches do not really count as such - in Spanish text-setting process separates itself from the scansion of the lyrics in isolation and re-creates the rhythm of the anarchic manner, so on X X X It is essential to observe that mismatches score X X X X X lle- X X X X ba- ca- a X X XXX X X X X X poem that neither linguistic stresses main beats. In this respect, much weaker in by assigning syllables to beats in the link between apparently poetic accents necessarily fall nor verse an prosody and musical setting is Spanish than in English. The second level to which constraints mentioned in the introduction to this apply is the grouping level. As chapter, SALIENCY is directly related to the concept of rhythmic cadence, which in turn concerns how the last or two last are fulfilled. In "Alegria, caballeros", the refrain and first and third stanzas share their rhythmic cadence structure, with four 'Green O' lines. This positions in the line means, as explained in section 4.2.4., that the four strong positions in each line, including the last one, are filled with syllables, but there between the third and fourth strong couplets nor stanzas are positions - weak positions filled in (191). In this case, neither salient in cadential terms. There is another structural principle that signals constituency levels in this rhyme scheme (from as seen are no -a-a to -b-b and so song, on) signals 241 a namely rhyme. The change in change in stanza. However, the second stanza presents a different line structure. Given that the first and third lines of this stanza have seven type 3, with three strong positions filled and the fourth syllables each, the resulting cadence is the fourth position is not empty, but filled with a a In musical terms, one empty. melismatic realisation of the third position, Belen. The structure of the second stanza, is therefore, 4343, where the couplet is salient. This is reinforced by the rhyme scheme, which does not differ from that in the other stanzas and refrain. (192) x x XX X X X XX X X XXXX XX XX X X X Los pas- to- res su- que x XXXX pie- X (4) ron x x x xx x xx xxxx XXXX X X X queel ni- hoes- ta XX Be- baen PARALLELISM works both at the at the X X XX XXXX X 0 len quatrain level and at the song (3) level. First, quatrain level, I have already mentioned that the cadences ending the units of the maximal analysis of the quatrains the second stanza, maximal which is structured are as identical (4444), with the sole exception of 4343 (see (192) above). The cadences of the analysis rhyme with each other (in this case, the cadence marking the couplet is realised by means of rhyme - see Table 32). Second, we observe positional parallelism at the strophic level, which has to do with the fact that settings to the same tune have This is the same distribution of onsets with respect to the grid (see (189)). directly related to the observation that the number of beats per bar cannot change in Spanish, which results in the preservation of isosyllabism throughout the song (Table 32). Synalepha happens second one in order to achieve once in the first stanza and five times in the parallelism, which again responds to isosyllabism. 242 Summarising, in Spanish, musical structure dominates structure, in such a way in that text-setting constraints are applied much over prosodic more leniently Spanish than in English. The only underlying principle in Spanish text-setting is parallel alignment of syllables and beats, which must remain constant throughout the the whole song. PARALLELISM is the most important constraint in Spanish. 4.3.7.3. Conclusions The analyses carried out in this section point to a correspondence between the timing typologies of language and rhythmic typologies of music. As I have shown, both English and Spanish show inconsistencies on the work one hand, and differently in a verse or mismatches between speech prosody, and music rhythm, on the other. These inconsistencies syllable-timed language like Spanish than in a stress-timed language like English. While in the first type of language there is counterpoint or a natural dialogue between speech prosody and musical rhythm, in the second type this counterpoint is considered non-rhythmic and, therefore, unacceptable. Spanish word setting would be impossible in English. The small mismatches in English could be regarded clumsiness that signal the folk stress mismatches observed in text because at all as nearly insignificant instances of compositional song style. On the other hand, the continuous radical Spanish folk song would make nonsense of an English English word stress is vital to the rhythmic arrangement of the language levels and, therefore, to kind in relation to the of languages. meaning. In other words, there exists a difference in dialogue between prosody and music for each of the two types In English, the level of agreement between the two rhythmic patterns is really high while in Spanish the counterpoint between the two is actually used expressive device. What I have described is, therefore, not only a difference in the ranking of metrical and grouping constraints for English and Spanish, but disparity in the degree of violation each constraint is allowed to incur. 243 as an a radical 4.4. Stretching out the theory: Art In section 4.3. I analysed two folk song songs, one in English and one in Spanish, and explored the similarities and differences between the rules of text-setting in these two languages. As mentioned in the introduction, folk of analysis, because one can presume number of constraints which potential link between speech prosody, essential, and which elements comes to realising are less central specific utterance a specific tradition. In this prosody and text-setting verse phonological system of which elements of the is the best locus for this kind that its composition is governed by native to that are song specific language a or even or verse are a definite sense, can reveal structurally completely marginal when it instance. There is, nevertheless, potential problem with assuming this unidirectional relation between speech, and music if the it could analysed corpus is reduced to a the single idiom within a verse Indeed, a genre. easily be argued that the observed behaviour of metrical and grouping is completely idiomatic and cannot be generalised constraints in folk song as a phonological fact. That is why the theories explored and applied in the previous sections of this chapter need to focus subgenre of song which is built on least a sample of art corpus similar principles to those of folk song song, a but differs important aspects. Section 4.4.1. is devoted to explaining from the latter in several the basic characteristics of art song, former and folk song on at in such a way that the differences between the become apparent. Sections 4.4.2. and 4.4.3. introduce and analysed in section 4.4.4. contextualise the two songs 4.4.1. Definition and characteristics of art song Art song differs from folk down in order to be song in that it has a performed in front of known author and it is always written an audience, normally with a piano accompaniment. The relationship between text and music in art found in folk song. always a As a to is very similar to that major characteristic shared by both subgenres, there is tension between the prosody of the difference with folk song song has to do with the avoid stress mismatches as much as verse and the rhythm of the music. The awareness of composers, who often try possible, tending towards a high level of agreement between speech prosody and music. In spite of this conscious effort, that original conflict is never completely solved. As Kramer (1984:160) points out, 244 The primary fact about is what might be called song distortion of utterance under the a a topological rhythmic and harmonic stress of music: pulling, stretching, and twisting that deforms the current of speech without negating its basic linguistic shape. The art exploitation of this expressive topology experience and as a song as a genre its shaping both - as a is the primary reflection of the contest between musical and poetic meanings. Summarising, any song, be it a folk song or an art song, the formation of the lyrics, and the tension between the versification laws that govern musical laws that with the is characterised by a regulate the flowing of the tune. Moreover, those two prosodic rules of the language itself, more can conflict specifically with stress assignment rules. 4.4.2. Art song Outstanding in Britain composers of art song in Britain, namely Ralph Vaughan Williams, Percy Grainger and George Butterworth, in folksong collecting, which became an among were also actively involved resource for their compositions others, invaluable (see Barlow 1997; Stevens 1960). Generally, the folk composers were vocal scores It is worth transcribed and published in the with same songs format collected by these as art song, that is, in piano accompaniment, arranged to be performed by trained singers. noting that their lyrics normally underwent a process of 'stylisation' whereby the often erotic lyrics of the original were made subtler or toned down. Butterworth (b. 1885, d. 1916) used two main elements as the bases of his compositional technique, namely folk music and the poetry of A.E. Housman. Butterworth was one of the before the First World War, The composer key figures in the folk music revival which took place after having met Vaughan Williams and Cecil Sharp. joined the Folk Song Society in 1906, eight (see Barlow 1992:72ff.) and started collecting folk collected around 300 folk songs, a fellow composer became a songs all after its foundation over England. He figure only comparable to the 810 collected by his and famous folk founder-member of the years song collector Vaughan Williams. In 1911, he English Folk Dance Society. Butterworth was a great believer in the power of folk music to endow art-music with a distinctive flavour, to the extent that "[fjolksong influence 245 can readily be discerned in the accompaniments of [Butterworth's] songs" (Barlow 1997:95). As melodic lines and noted, "if ever the opportunity occurs for a truly national production of ballet he once or opera, the mastered the success of the undertaking will rest in the hands of those who have technique and absorbed the spirit of our English dances and songs" (Banfield, Grove Music Online). Butterworth's most famous vocal eleven poems and pieces were a series of settings of six of the that form A.E. Housman's A Shropshire lad (1911), arranged for voice piano. Housman (b. 1859, d. 1936) wrote A Shropshire lad in 1896. This collection attracted an enormous following among composers such as Sir Arthur Somervell, Vaughan Williams, John Ireland, Ivor Gumey and C.W. Orr, among others (see Stevens 1960:151-173). According to 1997:55), the popularity of Housman's poetry a composers of the last was Englishness, their pastoral atmosphere, their rhythm, and their simple spontaneity of feelings drawn English Quinlan (1959:137-138) observes, "the brevity of the lines, coincidence. As their essential among researchers (see Barlow and the first decades of the twentieth century decade of the nineteenth century not some by the folk were song the fact that his poetry ballad and classical contributing factors". In other words, simplicity of Housman's language and metre, together with is itself a combination of the styles of the traditional English verse. In Butterworth's case, settings of the have been said to outshine other and directness" composers were his settings of Housman's poems same materials for their "simplicity (Barlow 1997:55). Butterworth's musical setting of A Shropshire lad stands out in that it is and art song gets one of the few collections where the distinction between folk blurred, and thus the native language of best observed. In Barlow's song composition can be (1997:56) words, "the influence of folksong undoubtedly helped Butterworth to illustrate the essential spirit of Housman's words, thereby creating a distinctively lyrical form of English song". 4.4.3. Art song in Spain Following German influences, the genre of art song extended to the whole of Europe, including Mediterranean countries like Spain. Many Spanish nineteenth century and early twentieth century, such Mompou, cultivated this traditional genre, which was as composers of the late Falla, Granados and ideal for the recreation of the various styles of Spanish singing and dancing. Federico Mompou (b. 1893, d. 246 1987) essentially was piano miniaturist and songwriter. As happened with a Butterworth, the most salient characteristics of Mompou's simplicity and their reliance folklore of the different distinctive symbols on is more uses their surface drawn from the modes and figures often his melodic writing rhythmically and structurally suggestive of Catalan folksong, to the extent that, occasionally, authentic Mellers 1987:81). In quasi-authentic Catalan melodies or He is composer a of his race an are indeed pervading. As Mellers (1987:81-82) puts it, of refinement, who finds in the folk that The collection Becquerianas was possible for him in Catalonia, The titles of these poems are and heavens smile air), "Yo (1971) sets to music six soy on "Hoy la tierra y los cielos swallows will poems collected in the me sonrien" (Today the earth ardiente, yo soy morena" (I burning, I am brown), "Yo am oscuras orthodoxy of Becquer's as a Spanish reason or celebrate the Becquer's cual return), "Olas gigantes" (Gigantic waves). Although the brevity and poems make them particularly apt for being set to practical difficulty for writing composers worked with texts one se golondrinas" (Dark music, their highly romantic character has always been regarded by poems. it me), "Los invisibles atomos del aire" (The invisible atoms of objeto" (I know what the object), "Volveran las themselves as by the Andalusian poet Gustavo Adolfo Becquer (b. 1836, d. 1870). work Rimas, metrical and dances Vaughan Williams in industrial Britain. wasn't for Hoist and el songs aspect of his own identity. He 'absorbs' (...) the country that made him, in a way the also used (see are in Mompou's vocal works, the structural and modal sum, idiosyncrasies of folk music es or gestures Spanish regions. For instance, he typical of Andalusian and other regional idioms, but songs are a composers musically worthy setting of those like Albeniz, Falla, Turina, among many others, have all by Becquer, but most of those vocal works have been forgotten for another. Mompou's collection was written as a commission to poet's death centenary, and became the best cycle of settings of poems ever written. 4.4.4. Case studies In this section, I explore a second sample corpus in the light of the theoretical apparatus applied to the study of folk song in section 4.3. This second corpus comprises two art songs. In section 4.4.4.1. I analyse Butterworth's setting of 247 Housman's poem "Loveliest of trees". In section 4.4.4.2. I study Mompou's text- setting of Becquer's confirming either a "Olas gigantes". The goal of these song idiosyncratic principles of Spanish and English folk as 4.4.4.1. Case genre will be either song, or in Spanish and English. first, and probably the most famous one, out of the six that form the collection Six in 1911. The other five are no more, A Shropshire lad', songs from entitled "When I was a cycle completed one-and-twenty", "Look not in lad", "The lads in their hundreds" and "Is ploughing?". "Loveliest of trees" became particularly well-known after it into orchestral an The poem rhyme works at the couplet level - are predominantly iambic means that there are - trochee are four stressed (stressed-unstressed) one - a syllables and five, where per cc, and so we can phenomenon known is four feet, and those as observe the second foot has lines, those lines, in principle at least, unmetrical a a substitution 'inversion' (in bold) - established for the weak position missing, something which renders other as poem line. This pattern gets and in lines 4, 6 and 10, which do not follow the metrical pattern - in the metrical analysis of the song we explore the different musical solutions given to these lines. Since all the lines in the poem of bb, that is, they follow the pattern unstressed-stressed, challenged in the first foot of lines will aa, the intelligibility of the structure of the as jeopardised by the rhyme scheme. Within each line, there a turned while punctuation and syntactic closure both work at the quatrain level. This which of was by punctuation and syntax rather than by rhyme. As disagreement is not significant, feet my team "Loveliest of trees" is organised into three stanzas of four lines be observed in Table 34, on -, not my rhapsody by Butterworth in 1912. each. Each stanza is delimited can re-defined study III: "Loveliest of trees" "Loveliest of trees" is the eyes", "Think studies will be counter-acting the conclusions drawn in section 4.3.6.3., in such general principles of text-setting for any vocal songs case that the principles of text-setting established for folk way confirmed as or poem have four feet, it is logical to think that they will all have the syllables. This is indeed the case, as number each line has eight syllables, with the exception of lines 4, 6 and 10, which have only 248 same seven syllables each. Line (accents)/Type of line S. R. Feet 8 a Inverted iambic tetrameter 8 a Iambic tetrameter And stands about the woodland ride 8 b Iambic tetrameter Wearing white for Eastertide. 0 7 b Inversion + Now, of my three-score years and ten, 8 c Inverted iambic tetrameter Twenty will not again, 0 7 c Inversion + springs 8 d Iambic tetrameter 8 d Iambic tetrameter 8 e Iambic tetrameter 0 7 e Inversion + About the woodlands I will go 8 f Iambic tetrameter 8 f Iambic tetrameter Loveliest of trees, the Is cherry now hung with bloom along the bough, come And take from seventy It only leaves me fifty more. And since to look at Fifty springs To see the are a score. things in bloom little room, cherry hung with snow. missing weak syll. missing weak syll. missing weak syll. Table 34: Scansion of "Loveliest of trees". Bearing in mind the high degree of metricality of the composer poem, we can presume did not want to upset its pristine structure. Figure 3 gives of Butterworth's setting of the poem into music. 249 us a that the general idea Molto moderato, sempre rubato e con espressione Love rit. poco long the 15 ■ bough, p a tempo - And stands a -bout the wood-land m rideZ Wear-ring / East for white P Now, tide._ er - Is hung with bloom a - the cher-ry now liest of trees, - of my three-score It - 24 h years 30. „ Twen-ty wilLnot and ten, poco rit. ==— fif - ty J i dim. flargamente sense, score see the cher blending between art goes song - ry hung with song, a song m lit-tie room, snow. _ (193) is introduced of senza misura bars points where bar lines fall are - - is that it constitutes fact which any other can perfect a be observed in the composition. The first stanza describes melodic contours as a » Fif - ty springs are hand in hand with the stylistic analysis. In this and folk metrical and melodic structure of the only work ly leaves me of "Loveliest of trees" (melody) the first observation to be made about this sequence *' r*11 it is important to note that in "Loveliest of trees", like in the metrical analysis melodic idea on analysing the metrical arrangements of the musical setting of Housman's poem, art song, To I will go Figure 3: Musical a score, J I irr lr., A-bout the wood-lands Before a-gain, And take from seven-ty springs a tempo And since to look at things in bloom more 37 come p first on bars 1-6 typical of art the piano and then -, on song. the voice The first - with a that is, the shape of the bars and the relatively meaningless in metrical terms in that they loose guide for the performers. For this notated metre is different from the 'real' metre, 250 reason, we could say that the which is loose, nearly ad libitum. (193) In the first musical points throughout the At several other in (193) overall is - and sense phrase, the delicacy of the piano accompaniment is remarkable. very song, the piano is not present at all - bars 4-5 often its role is reduced to articulating the odd note to keep the of form. On the other hand, on the few occasions where the piano's line relatively fulsome and melodic, it never overshadows the vocal part. The looseness of the first musical listener's metrical intuition tends towards a phrase (194a) is evident in that the 3/2 time signature (194b). (194a) poco 7 J Is ij r r hung with bloom f a - rit. z==— iJ- j1 ^ L^r long the bough, (194b) -m Is =±= rJ- hung with snow a (195) - + long the After the first 9-17 in v - — m bough, phrase, a W L^l 7 _ series of crotchets establishes a sense of metre - bars which is in agreement with the notated time signature and bar divisions. 251 (195) East er - tide. The clear-cut musical into turns triplets on the song - rhythm until bar 17, where the accompaniment goes on triplets (196), thus re-establishing the initial metrical freedom. After the the piano, there is of the first melodic idea a repeat - the main theme of in crotchets and minims. The combination of the triplets with the crotchets and minims gives a grand character to this part. (196) * ytg— Sa The * piano part in (196) also works as a transition into the second stanza, achieved by means of a modulation which is, nevertheless, stanza - bars 22-31 - is in C sharp minor, while the accompaniment remains in the original key of E Major. At this point at the ambiguous. The melody in the second rhythmic level, where the song we can observe an imitation of folk becomes totally metrical. 252 song both (197) 22 Now, of my three-score years and ten, Tvven-ly will not come w a-gain, T7 P^ & The third stanza shows 32-37 in marked first a on - bars the syllable wood— bar 38. This rise is by the crescendo and the ascending intervals in the melody second beat of each bar in or steady rise to the melodic and lyric climax which takes place (198) i - observe the (198): B, C sharp, D sharp, E. (198) 30 doco rit. fir'37 ty cresc. more flargamente A-bout the wood After the marked by - lands And since to look at things in bloom I will go climax, there is an art song position of cher- The melodic ends on a C ending - see the cher - ry hung with lit-tie room, snow. the rhapsodic rhythm of the beginning, bars 38-42 in (198). Again, bar lines means are not - To a return to character meaningful metrically, which unstressed Fif - ty springs are dim. that the accent on are not the syllable / and the significant in terms of mismatches. bars 41-42 in (198) - is inconclusive, as the melody sharp, while the piece is in E Major. This is another trait of the folk song style of minor modes, where C sharp indeed might be the tonic. Nevertheless, the piano completes this ambiguous structure by returning to E Major (199) -, - bars 43-48 in thus cancelling the modal character of this part and re-establishing the initial tonic. 253 (199) In conclusion, the song could be divided into two main parts. One comprises the first and third stanzas, which are set in a actual metrical notation is not prosodically significant, so rhapsodic, art rigid, that is, the bar lines the few stresses which fall constitute real mismatches. The second part composed in a folk song on style. In this part, the are not musically weak syllables do not style. It follows the prosody of the words and, at the perfect blend of folk song are and art combined in such an realised as the surface structure of the melody shows a can parallelisms which bring it close to the basic principles of folk we can but song, be series of subtly composition. If we divide the melodic line into stretches comprising two each, that song. impact at the microstructural level. The two main stylistic parts subdivided, further same a way Grouping does not only work regarding the macrostructure of the also has or comprises the second stanza, which is time, achieves perfect metricality. These two parts the song represents a song verse song lines make the following observations. First, there is an instance of parallelism between the first two lines (200a) and the last two lines (200b). In both cases this parallelism leaves out three syllables - the last three syllables of line one {-long the bough) and the first three syllables of line eleven {a-bout the). (200a) p sA poco Love rit. long the - liest of trees, bough, 254 the cher-rynow Is hung with bloom a- (200b) 37 (A dim. flargamente pp A-bout the wood - lands seethecher To I will go - ry hung with snow.. Second, line 3, which is almost bare harmonically, is linked to line 4 to form the second musical phrase, which is loosely parallel to the musical phrase formed by lines 9 and 10. In (201) we can observe that lines 3 and 9 are practically identical melodically. On the other hand, the parallelism between lines 4 and 10 is less clear. Line 10 repeats its musical the note-values and intervals of line 9, and is less conclusive, within phrase, than line 4. Nevertheless, it has the they both establish a sense same function as line 4 in that of closure in their respective phrases. (201) poco long the bough, 15 ■ And stands a -bout the wood-land Wear-ring ridel white / East 30 p a tempo rit. poco fiF - er tide.. rit. - p ty a tempo And since to more Third, lines 5 and 6 look at things in bloom Fif - ty springs are lit- tie room, melodically parallel. are (202) 24 P Now, of my three-score Last, lines 7 and 8 first and third bars the fourth one, are are years and ten, Twen-ty wilLnot also loosely parallel. As can come ip* a-gain, be observed in (203), the in terms of the note-values, and so is the second bar. As for given that it closes the musical phrase formed by these four bars, it 255 for introduces new note-values, which nevertheless do not disrupt the feeling of parallelism achieved by the previous three bars. (203) 30 come a-gain, And take from seven-tysprings illustrated in Table importantly, the musical and sense of closure of the ly leaves me fif'- ty more.. Lines <- song, as way while the middle lines develop the that the listeners perceive a sense of composition. 1, 2 X Line 7 •*Lines on - 35, where the extreme lines mark the beginning and, more lyric elements in such direction in the Line 5 It conclusion, the setting of the poem describes a chiastic structure, In Line a score, poco rit. -> 3, 4 Line 6 Line 8 9, 10 Lines 11, 12 Table 35: Musical structure. At the metrical between the level, the song displays a series of interesting mismatches prosody of the words and the note values, related to the art song a phenomenon which is character of the first and third parts. Although Butterworth consciously simplified the piano accompaniment in order not to upset the flowing of the poem realised in the melody, at certain points the musical setting intentionally 'disagree' with the verse of durations. The first line poem's very seems to prosody, mostly in terms of the arrangements - bars 4-6 - is a good example of this rearrangement of durations (see (200a) for the musical notation). 256 (204) x x X XXX X XXX X XXX X XXXXXX XX XXXXXX liest Love- In of the trees (204), the word loveliest is set to music in a way which defies both prosodic and difficult high E which vocal music expectations. Its first syllable, love-, is set to then forms melisma with D and is subdivided into two, as the musical the a X a setting forces singer to prolong the vowel sound in the third beat of the bar, while the second syllable, -liest, is practically squeezed into the last beat of the bar together with the preposition 'of. deception of the listeners' expectations in terms of durations This the insertion of certain note values in the middle of effect. Such is the case musically longer than in the first any of the a line creates setting of trees in line 1, other in the same a a recurs as sort of caesura- word which is made line, like white (set to a high E, like love- line) and East- in line 4 (205), and wood- in line 11. (205) X X X X X X X X X ring Wea- X X X X X X X X X X X X X X XXX for X X X X East- X X white X X X X X X X X X X X X X X X X X X tide er 257 X X X X X X X XXX syllables that All the diphthong as their nucleus, so 'Durational mismatches' in the setting seems to get as long note-values have are set to close to they are not mismatch as or a examples of what I referred to as analysis of English folk a long vowel a song (section 4.3.). This it is possible in English, but without actually mismatching anything. Summarising, there is a series of beautiful effects achieved by mismatching certain lexical words in terms of duration and pitch. The overall effect is that of mismatching in the normal enunciation of the phrases, which ways that the listener nevertheless understands. These unmetricality of lines 4, 6 and 10, which, Table 34, start with inverted foot, while the second foot has a Butterworth's solution for this metrical • In line twen-, the • 6, Twenty will not and the same mismatches. applied to the setting of Housman's The relative durational freedom solves the apparent contradicted in are are not proper problem branches as a as we poem observed in missing weak position. follows: again, the durationally stretched syllable is come happens in line 10, Fifty springs are little room, where elongated note corresponds to the syllable fif-. For line inserting 4, Wearing white for Eastertide, the music makes it metrical by a long note in white, thus compensating for the missing weak syllable. The two musical realisations of the defective lines are summarised in Table 36: (X) 0 / X / (x) / X / Twen- ty will not come a- gain Fif- ty springs are lit- tle room Wear- ing white for East- Table 36: Musical settings of defective lines. 258 There them happen few instances of stress-beat mismatches in this song. are a primary counts: on First stanza: a. A few of a bout (3rd line; bar 10), a mismatch that follows the verse prosody, not the phonological prosody. b. (3rd line; bar 39), which has Third stanza: / to do with the crossing between real and notated metre. The vast majority of those mismatches occur on secondary counts, that is, on the third beat of the bar: First stanza: loveliest c. 2nd line; bar 22), will (3rd line; bar 25) Third stanza: since (1st line; bar 32), about (line 3rd line; bar 37) d. Second stanza: e. As (1st line; bar 4) observed, most of the mismatches found in the text-setting of be can of(line correspond to grammatical words, and most of them fall Housman's poem secondary counts. They are minor mismatches, which setting and, therefore, cannot be judged In sum, characteristics of a as are acceptable in English text- the grouping and metrical ill-formed. "Loveliest of trees" prototypical song presents in English, that is, a rigid alignment between beats and syllables at the metrical level, whereby stress mismatches ruled out, and a looseness of art song analysis of this the pre-established structure. In this a common 4.4.4.2. Case "Olas the case, bar division and the exactness of folk song principles applied to folk constraints are practically relative flexibility at the grouping level, where the different lines do necessarily follow not on the song song blends the phrasing. Overall, points to the confirmation of the hypothesis that the song are not to verse and song idiosyncratic, but respond to a general set of composition. study IV: "Olas gigantes" gigantes" is one of the six poems by Becquer set to music by Mompou under generic title Becquerianas, in 1971. Like most of Becquer's poetic works, "Olas gigantes" is a highly regular composition in terms of metre and rhyme. The poem is subdivided into four stanzas of four lines each, of which the fourth is much shorter than the other three (heptasyllabic versus hendecasyllabic lines). Synalepha (underlined in Table 37) is realised whenever possible realisation of synalepha, see 3.2. - so - for the rules that the number of syllables 259 per on the line remains stable. Using Hayes and MacEachern's (1998) taxonomy of lines, how the fourth line in each which a type throughout the whole poem, where Syllables Olas eieantes las que os romneis bramando playas desiertas con vosotras! Rafagas de huracan del alto 7 arrebatais que jllevadme con Nubes de 11 el ciego torbellino en 11 vosotras! 7 tempestad que rompe el rayo 11 fuego ornais las desprendidas orlas, arrebatado entre la niebla jllevadme con Llevadme por con 11 bosque las marchitas hojas, arrastrado v en 11 esnumas. la 7 piedad donde el vertigo 11 la memoria. 11 razon me arranque mi dolor a solas! 11 7 Table 37: Scansion of "Olas Stress-placement structures of structures which gets line of each 69 4, are positions 1,4, 8,11 3,6, 10 a 2, 6, 10 — 2,6 a 1,6,11 — 2, 4, 8, 10 a 3,6, 10 — 2,6 a 1,6, 8, 10 — 2, 8, 10 a 4,8, 10 11 oscura. jPor piedad! j Ten go miedo de quedarme con 11 vosotras! a Stressed Rhy. 11 11 y remotas, envuelto entre la sabana de jllevadme are type lines rhyme, while odd lines do not. even Line en 3 line, while the other three observe quatrain is salient. The rhyme scheme remains untouched that the means quatrain is we can 2,6 a 2, 6, 10 — 4, 6, 10 a 3,4, 6, 10 - 4,6 a gigantes". also consistent throughout the repeated. Of these, the most salient poem. one There is a series is that in the fourth quatrain, which realises only two stresses (on the 2nd or 4th and 6th 'Gigantic waves that break crying/in the desert, remote beaches,/tuck into the foamy with you!/Hurricane blasts that steal/from the high wood the faded leaves/carried away in the blind whirlwind/take me away with you!/Rainclouds that the thunder breaks/and in a fire you adorn the unpinned trims/carried off among the dark fog,/take me away with you!/Take me away, I beg you, where vertigo/with reason tears off my memory,/I beg you! I am afraid of being left/alone with my pain! Translation: sheet./take me away 260 positions), while the rest show three combinations of stressed are or even positions in the four stresses. Table 38 shows all the poem (numbers signal the positions which stressed). 2 6 10 (x3) 2 6 10 (x2) 6 10 (x2) (x2) 3 2 1 4 8 10 4 8 10 8 10 6 1 3 4 1 4 6 10 6 10 6 10 8 4 4 6 Table 38: Table 38 illustrates a tendency towards the Rhythmic accents. certain degree of parallelism in stressed positions. There is occurrence of a stress in the middle of the line, in such that the overall structure of most lines is five a way (6th) plus last-but-one (10th) syllables plus stressed syllable syllables. Apart from this, there is a fixed stress syllable of each line, something which is a rule in Spanish poetry, five a on the as explained in section 3.2. If we stresses on analyse the position of stresses line by line, the first position happen in the first line of tendency towards second (with one or exception, stress a in that it has song, an will conclude that all the a stanza, while there is third position stress in second lines. The on the fourth position) while fourth lines follow the pattern As we 2, 6 and, in the can very same a tendency be observed in third lines, last line, 4, 6. Mompou's "Olas gigantes" is different from "Loveliest of trees" operatic, rather than pastoral, character, meaning that its declamatory style imitates the rhetoric of operas (see Figure 4). 261 /- m O- las_ gi U^r t~■ ^ ifcr JJJ11' cr gsm- tcs_ - qucosrom- pdis En las bra-man-do umplitud 10 Utt "i* ur pla-yas de-sicr-las ^ y re - so mo- Lri'^CJi'Ll En-vuel-lo las, /-yv A-jr-N - Lie - vad - pu- mas. me N quea-rrc-ba - (aisJDcl al-lobos-que las mar-chi-las ho-jas. Ra-fa-gasdehu-ra c6n lias en-trelas sa-ba -nas decs piii espr. * s—v .convo - v J2. piii espr. ~ 30 A - rtas - LW^cJ ^ cie-go tor-be tra-do - lli- Lle-vad no, — me convo-so tras. - 40 n.-t/i Nu-bes de teui-pes tad ,1-3 querom-peel ra mm Yenfiie-goor-niiis las des-pnai-di-das yo, - $f 1- las, A-rrc-ba 1 in - ta-docn- tre lame-blaos - v£r - ti 1>J rF 0■] Con la ra t 7 611 lim mea Lie- vad cu- ra, - MM rran-que por me a me - - mc convo ITIO- tit. & mie-do de que - dar - Figure 4: Musical The setting of the are not normally corresponds to metrical in themselves - pie-dad Por mi score do- lor_ genre a a don-dcel MMM LH Por pie- ia, dad ten-go tempo r\ so - las._ of "Olas gigantes" (melody) does not fully conform to that of a lyrical poem explores the limits between this which tras. far-iln Con me_ so- - pie dad a 77 - m [J Lie- vad *5 lento p piii 50' or and opera. song, but In spite of this operatic character, high degree of metricality, some of the proportions observe, in (206), the long note half way through the 262 first line, the syllable -tes, then on on -peis, but not at the very end of the line, on the syllable -do. (206) i - O-las_ In most cases, style visible rhetorical to T|'tXr "r' r i ■ gi-gnn-tcs qucosrom- p&s 'hr ^JP bra-man-do the metricality of the words is used to render the song's - in (206), the first beat of bar 6, we can observe the setting of the synalepha corresponding to a que os high pitch (G flat), which makes this syllable doubly salient. There of the song are many numerous examples of this interaction between the character and the metrical setting of the words. Bars 14-18 (207) towards the climax, following from are a chromatic rise in the melody and a a build-up dense piano part in the accompaniment. From bars 16 to 21 (207), the rhetorical style of the song is evident, with high notes and lengthening of notes altogether. The syllable -vad- in llevadme is the dramatic climax of the phrase. (207) ampliiud h==i ■g '"' LT' En-vucl-lo en-lrc las The line llevadme folk song. yy=i sa-ba-nas decs con whereas the quatrains as - previous three climactic line is not that of a the three a role is imitating how specific line. In - me con hFtrl fry r 'O so lias. analogous to that of the refrain in poem ■ a finish off with this line, which previously mentioned, this last line is heptasyllabic, are invariably hendecasyllabic. The setting of this typical refrain, but follows the rhetorical character of the line in question is set to composer speaks has In each of the three instances where it the song. -, a I.lc- vad pu- IIias. Three of the four stanzas in the demarcates the 208) vosotras i.r-rL J', tr~\ fHr bf —ft T an actor any case, a appears - bar 18, bar 34, bar 55 (see different musical phrase. We could would not deliver the whatever the syllables in vosotras. 263 same composer say that the the second time s/he does, he always matches (208) Bars 18-22: amplitud Llc-vad- me ton vo Iras so - Bars 34-37: Lie-"tad me con vo - so - tras Bars 55-57: p piu lento ! i'Lrnii Lie- vad - me con vo In terms of gigantes" seems to follows the latter's can - so- I tras. grouping constraints, although the internal structure of "Olas be more complex than that of "Loveliest of trees", the former tendency towards macrostructural parallelism. In this respect, we observe that the first and last stanzas parallel, and the same bars 3-22 and 63-84 (209) - happens with the second and third stanzas what in musical terms is called 40-57. Bars 63-84 are exposed 3-22. in bars - This a - are bars 23-37 and 'reprise' of the theme clear instance of parallelism stretches approximately 20 bars. (209) Bars 1-22: O- las_ gi-gan-tcs con vo - so musically queosrom- Iras 264 p6s bra-man-do En las along Bars 63-84: a mie- do de que a - Con me mi do- lor a setting of "Olas gigantes" describes, The trees", dar - tempo so as - las happened with "Loveliest of chiastic structure: Stanza 1 Stanza 2 ^ ^ Stanza 3 Stanza 4 Table 39: Music structure. The enforcement of these bars, parallelism partially explains the mismatches found in the composer is using identical metrical structures and laying the as syllables under the corresponding notes, almost without rearranging the consequent mismatches. This is made can be observed in bar 63, parallel to bar 3, where the half-line Olas gigantes is set (210). The fact that the former has one more syllable than the latter forces the extra note-value in bar 63. but where the half-line llevadme por piedad a new one occurs composer to The mismatches found in bar 3 do not appear in the latter. (210) O- las gi- gan- Lle- vad- me por b»l,^« ±TOf" ■ O- las L- nff-tT gi - gan- tes_ b l> ■ i T,le- vad - me » por introduce m pie- dad_ 265 tes pie- dad an in line 63, preposition The repeated por is mismatched several times, something which does not happen in the - as poem. the line por piedad is This indicates that the highly consciously done. mismatch is The poem setting of the correspond to shows twelve instances of synalepha (see Table 37). The musical poem a keeps synalepha in eleven single note value. Such is the relative pronoun que and the reflexive pronoun cases, case os making two syllables of the synalepha between the in the first line. (211) O-las_ gi-gan-tcs In bar 52 (212), the synalephas so composer bra-man-do qucosroni- pcis is forced to divide either of the two prosodic that the line ends in two crotchets. The setting of the two contiguous vowels in arrebatado entre to two different note values breaks the turns it into an instance of hiatus - for a synalepha and section 2.2.3. definition of this term, go to (212) 4=f=\ ^ zS r-fvu-eJ ' las, A-rre-ba - 0 ta-doen- trelanie-blaos Mompou's setting of Becquer's on primary counts - poem -cu-ra, shows first beat of each bar - a series of stress-beat mismatches, both and on secondary counts given that the time signature is 2/4, the second beat of each bar is Let us have a look at the mismatches First stanza: que os on en tre case, secondary count. primary counts first: (1st line; bar 24), las (2nd line; bar 28), (3rd line; bar 30) Third stanza: 48), in this (1st line: bar 6) Second stanza: de huarrastrado a - rempestad (1st line; bar 41), c/esprendidas (2nd line; bar (3rd line; bars 52-53) Fourth stanza: pie dad line; bar 61), por (1st line; bar 64), a (1st line; bar 67), de el (1st (3rd line; bar 75), mi (4th line; bar 80) 266 There mismatches are numerous on (1st line; bar 3), First stanza: olas 12), entre (3rd secondary counts: line; bar 15) would be the 1st person - en (2nd line; bar 9), y (2nd line; bar entre with the stress on the last syllable singular of the past tense of entrar {to in); this is what Janda and Morgan (1988:160, see come 4.2.1.) refer to as levelling of stress-marked distinctions, where the context the disambiguates the meaning bar 20), vosotras sabanas (3rd line; bar 16), con (4th line; (4th line; bar 21) Second stanza: arrebatais forbellino - (1st line; bar 25), hojas (2nd line; bar 29), (3rd line; bar 32), con (4th line; bar 35) Third stanza: v en (2nd line; bar 46), a/rebatado (3rd line; bar 51), con (4th line; bar 56) Fourth stanza: donde line; bar 69), la (1st line; bar 67), vertigo (1st line; bar 68), la (2nd (2nd line, bar 72), de (3rd line; bar 77) The metrical mismatches in this song syllables, which often fall However, mismatches syllables mismatched more salient. An on are can have to do with the position of the stressed weak beats, while unstressed emphasised by pitch, that is, example of this phenomenon on happens in bar 25 on strong on many can beats. occasions the even be found in bar 24 (213), where a high note, while the heavy the first beat of the bar makes it explicitly - fall be set to high pitches, which makes the mismatches the metrical mismatch in de huracan is set to accompaniment ones a mismatch. The same a/rebatais -, which is parallel to the former. (213) fn,1 J-yi u - Ra-fa-gas del C ?o 1 Wi \}S> \rn 4^ On other occasions, mismatches duration, in such a way are realised exclusively by pitch and that these two contradict what the metrical arrangement 267 establishes. Such is the case of the last syllable in arras trar/o - bar 30 (214) which is set to the weakest beat in the bar, and to a melisma at the same time. In terms of syllable -do is not mismatched, but the metre, the presence of the melisma on this syllable makes it salient in terms of pitch and duration.70 (214) piii espr. 30 r <,,, A - rras - m m cic-go tor-be - Hi- tru-do This observation contradicts no, Morgan and Janda's (1989:277) statement that pitch, duration and loudness do not work syllables which coincide with the musical downbeat unstressed being stressed as long accent which is at least "Olas for stress in music: "ordinarily as cues gigantes", a as as the normally stressed syllable has syllable that falls musical downbeat is indeed perceived on a secondary stress reinforce the effect of the downbeat or contradict it. Spanish art there song, are as many mismatches, something which does not happen in the English Given that the setting has a time signature of 2/4, particularly those with long words involved, one of the reasons explain the that Janda and Morgan recurrence see 70 4.2.1. - are many by these authors the fact that stress is - for lexical song as cues in text- grammatical analysed in 4.4.1. of these mismatches, virtually unavoidable. This is indeed (1988:167-168, in 4.2.1.) give in order to of lexical mismatches in Spanish frequency of polysyllabic words forces the As observed as while any other extra musical accents are perceived as as In this instance of as musical/rhythmic hand, pitch and duration act marked. On the other can a perceived pronounced". As observed in the analysis of mismatches in stressed in the unmarked case, setting, and are not appearance a song. In their words, "the of stress-shift". full review of their papers on the topic, predictable, and thus recoverable, makes stress-shift score places -do- on the second quaver of the bar, as shown in (214). However, in the version, the performer realises a melisma on -tra-, the stressed syllable, which occupies three The vocal recorded quavers, while -do- is made to correspond to the last quaver in Whether this is a circumstantial decision of the performer or a the original score is beyond my knowledge. 268 the bar and not realised as a melisma. widely-accepted 'correction' made to a possible device, much even in a than in more common language like Spanish, where stress minimal pairs are English. of text-setting in Spanish differs from that in English in that, in The process the prosody of the words the latter, be slightly altered in order to fit can pattern, but the musical setting must preserve that pattern without prosody of the words to conform to a any specific further. In Spanish, the prosody of the words verse altering the can be altered design, but that design is not enforced when the text is Actually, that intermediate stage set to music. a verse from speech to - practically neglected, and the prosody of the words can verse - can be be directly rearranged to fit a specific tune. In Janda and Morgan's (1989:284) words, "[in Spanish, t]ext/tune matching begins with erasure of all lexical stresses and proceeds mechanically by left-to-right, one-to-one association of syllables with notes of the tune". In conclusion, the analysis of "Olas gigantes" confirms the hypothesis that, in Spanish, music rhythm and metre do not agree with verse prosody but work why metrical constraints constraints we speech prosody, independently of both. This is facilitated by the fact that the preservation of word stress is not crucial for the if or - acceptability of a musical setting. That is the are not enforced in Spanish text-setting, while grouping chiefly, PARALLELISM take into account that the - are always observed. This is not surprising primary building principle in Spanish section 3.2.2. Spanish text-setting, both in folk isosyllabism - song, proves that the syllable is the most important unit of rhythm, in such the go to parallel arrangement of syllables constitutes the in this reason essence song verse is and art a way that of rhythmic perception language. Stress signals the division between the parallel rhythmic units, but where that stress falls is not important for determining the acceptability of a setting. 4.4.4.3. Conclusions In the introduction I argued that folk of the relation between speech, that that folk we could presume verse song song represented the best locus for the analysis and music at the prosodic level. The composition and performance is governed by definite number of constraints native to concluded that this worked was reason was a a specific tradition. In section 4.3. I indeed the case, and that metrical and grouping constraints differently in Spanish and English due to the nature and prosodic functions of stress in both languages. However, this conclusion might have been slightly 269 ambitious, as the musical analysis was restricted to a specific subgenre of song. The observed behaviour of metrical and described idiomatic, in which as phonological fact. This is the another song subgenre of song, in this and art order to be do not reason song grouping constraints in folk case song might be it could not have been generalised as a why the initial hypothesis needed to be tested on case, art song. is that the latter has a The essential difference between folk known author or performed in front of an audience. We could authors and is composed in say that art song composers necessarily apply 'native' linguistic constraints when they set They often try to avoid stress mismatches much as as a text to music. possible, tending towards a high level of agreement between text and music. In the first song case study analysed in 4.4.1., in English responds to metrical level, and lines do not a we observed that a prototypical art rigid alignment between beats and syllables at the a relative flexibility at the grouping level, where the different necessarily follow of trees" confirmed the a pre-established structure. The analysis of "Loveliest hypothesis that the principles applied to folk idiosyncratic but respond to a general set of constraints song are not and song Spanish prototypical art song common to verse composition. In the second confirmed the the most not study, the analysis of case hypothesis proved for Spanish folk important unit of rhythm, in such essential for the a way a song, namely that the syllable is that the position of lexical stress in perception of that rhythm, while the parallel arrangement of the syllables in different lines and stanzas is. The characteristic that the two songs dimension of every have in is the aesthetic single device applied to the interplay between text and melody, something which is not so clear in folk song, given the 'intuitive' character of text- setting in the latter. Although the constraints applied to art at common work in folk song, art song song are the same to those consciously plays with the available constraints, stretching out the possibilities available to the language and thus clearly marking the borderline between what is Summarising, art song acceptable and what is not for each specific language. brings to the surface text-setting 'mechanisms' that unconsciously applied to folk between speech, verse and song, which makes the initial hypothesis of song even more powerful. 270 a are link 5. Summary, conclusions and further implications The last decade has witnessed a a steadily growing interest in the study of music from linguistic perspective. A significant phoneticians and syntacticians practically neglected during the first half of the only to be marginally re-incorporated into linguistic investigation in the fifties, and even more so in the seventies and eighties. It is relationship between language and music that the was Enlightenment from Ancient Greece to the a overspecialisation and was thus The on but between once a lesser position among a as an relevant mathematical fields even - the modern university system and disciplines the part of researchers, such relegated to the consequent relationship became unimportant linguists and music theorists. interdisciplinary nature of the research enterprise presented here is capricious neither separation radical -, well-known fact a always regarded topic of investigation in the philosophical, literary and established mainly phonologists, - have started to turn their attention towards the - musical field, which had been twentieth century of linguists group nor accidental. My dissertation intended to follow the tradition of the research done in the fields of phonology the twentieth century, taking such work and text-setting during the last decade of a step further by means of adding both linguistic and musical academic depth to it. As stated in the introduction, the main aim of this dissertation of speech, verse to was to explain the potential connections between the rhythm and vocal music, be made about the as well as to define the limits of the generalisations matching of speech rhythm with poetic metre and with musical rhythm. In other words, the study of vocal music, both at the lyric and melodic level, was intended topic that has as a linguistic project in the broad proven to sense, which could shed light on a be central to phonological investigation. My project encompassed the application of specific linguistic theories, such as OT, to non-linguistic or partially linguistic objects analysis of linguistic prosody from theoretical framework for the become a are songs - as well musical point of view. The choice of OT study of text-setting was not random, as as as the the OT has major theoretical framework in the study of the relationship between linguistic prosody, topic a - verse prosody and musical rhythm. However, most studies preoccupied exclusively with English verse their theoretical claims biased. In order to solve this 271 and songs, a on the fact which makes problem - at least partially -, the present work incorporated the analysis of materials in a second language to check the I - Spanish - validity of the theoretical assumptions made by OT. of my dissertation is that the scope am aware broad, very only text-setting in English and Spanish, but also speech and languages. This fact forced me to devise a verse as it covers not prosody in both simple structure for it, where I would present each of those rhythmic manifestations separately, in order to be able to reach a comprehensive view of the topic. I started off introducing the nuclear concept temporal manifestation of speech, shown nature rhythm - which pervades the and music. In order to illustrate the temporal by these three rhythmic manifestations, Liberman's (1979:313) observation that quoted verse - "[a]ll temporally ordered behaviour is metrically organised" was which would stimulate the discussion from the very as a strong statement beginning. I also explored the deep structural parallelism between language and music, related to the fact that they share a hierarchically organised metrical structure. depends the nature of the language and the The articulation of such a structure in question, a fact that forces the researcher to music genre languages and musical the two genres to languages I know best where the on be explored. For this - Spanish and English narrow reason, - down the set of I decided to deal with and the vocal music genre relationship between melody and text is strongest - song. Chapter 1 introduced prosody, the discipline that studies the mechanisms with which languages articulate the flow of sound in segments are grouped around a salient one Depending the on rhythmical way, where weaker what in poetry is called 'metre' proper. the mechanism/s used to assign prominence to certain segments others, languages stated in this - a are classified into stress, pitch-accent or tone over languages. As chapter, English and Spanish, like most languages in Europe, are stress languages, since they make use of stress in order to mark syllable prominence. Chapter 2 narrowed the inventory of languages to be analysed, presenting the two main subgroups into which stress languages have been divided, namely the subgroup of stress-timed languages and the subgroup of syllable-timed languages. As explained in this chapter, in syllable-timed languages syllables are perceived as being rhythmically equal, while in stress-timed languages, stresses fall at perceptually equal intervals. As a consequence, the syllable is considered the major unit of rhythm 272 in the former group, differences were while in the latter this role is played by the foot. Once these established, the languages, namely English timed of study scope was narrowed again to two which has been traditionally classified - language par excellence - and Spanish - as the stress- which has been classified as one of prototypically syllable-timed languages. In order to understand the current status the of stress and the syllable in those two languages, the second half of chapter 2 presented their evolution, starting from Latin and Old English, respectively, and then overview of gave an some of the current theories of stress assignment in both languages. PdE stress assignment was explored following Halle and Vergnaud (1987), Giegerich (1992) and Gasiorowski (1998). The most important points about PdE the purposes stress for sensitive language - of this dissertation are, first, that English is a quantity- that is, stress is assigned according to syllable weight, which is, and second, English in turn, closely related to vowel length and vowel tenseness stress is assigned in accordance with two series of rules, namely phonological rules morphological rules. and - The interaction between stress, length and weight is, therefore, essential in the organisation of English speech rhythm. Regarding stress assignment in Present-day Spanish, two major theories, Harris (1983) and D'Introno et al. (1995), elaborates the a specific analytical apparatus were on explored and contrasted. The former the observation that the characteristics of Spanish stress system follow from the morphological structures of words, where etymology plays an essential role, and from the contrast marked vs. unmarked stress placement. The most important observation of the system is that nonverbal stress falls its on one of the last three syllables of the word. Beyond the word-level, in spite of predominantly syllable-timed rhythm, Spanish requires that stresses fall at specific points in the utterance, which work points where unstressed syllables the speed of discourse. As must come and not into as rhythmic 'leaning-points'. This entails that, at are very numerous, a consequence there will be of this, certain processes a need to increase of 'compression' play. The two compression devices applied in Spanish, synaeresis synalepha, take place at the syllable level, and affect the length of vowels, but their quality or their tenseness. 273 Chapter 3 introduced the analysis of systems of versification for Spanish and English. The point of departure for such observation that an analysis was Abercrombie's (1967:98) "[t]he rhythm of everyday speech is the foundation of verse, in most languages". According to this, the speech prosody of a language correlates to its prosody. verse English is defined verse as accentual-syllabic, stating that it is built principle that there has to be basic which gives rise to and English verse feet are a specific number of stresses per upon the line, each of its respective foot. As happened in speech, in governs perceptually isochronous. The devices which are used in English in order to adjust the duration of feet do not respond to phonological rules. Synalepha, for instance, is highlighted syllables line. In per standard metrical device which has to do representation of the with the abstract metrical of as a summary, English poem, not verse with the actual realisation prosody is found to respond to the building principles of English speech prosody, that is, reduction of vowel sounds in order to get a much specific number of isochronous feet per line. Spanish verse is defined Spanish verse has isosyllabic, Spanish a verse as syllabic, which can be proved by the fact that perfectly regular syllable count. Apart from being also takes metrical accents into account. The most important observation in this regard is that, in order to regularise syllable counts, Spanish isosyllabic poetry uses a series of metrical devices, which coincide with the compression devices used in speech devices used in speech are - in fact, we observed that the compression phonological rules that then get transferred into Chapter 4 constituted the nucleus of this dissertation, relation between speech, verse and music rhythm chapter started off by presenting two objects each with its object - a as as it examined the triple realised in folk and art The definition of song - "a composite which combines own structure, a linguistic object - text - and a musical by text-setting, which has to do with how lines of linguistic texts arranged in time against a predetermined rhythmic pattern in Given that both text and tune take where strong place in the time line and thus have and weak units alternate 274 are sung verse. rhythm, they share the characteristic of being arranged hierarchically into - song. tune" (Dell and Halle in press). The relationship between those two objects is mediated level verse. - and a grouping level - a a metrical where small units are arranged into bigger constituents. In OT terms, at the metrical render a different well-formed metrical are applied composite. This interaction will be different in languages, and will have well-formed series of CON level, thus regulating how text metre and tune metre can interact to direct impact a grouping level, which regulate how text a a the ranking of CON at the on and tune groups groups are arranged to yield composite. In turn, the ranking of CON at the grouping level will have the ranking of metrical CON. Among the direct consequences on numerous text- setting CON explored in Hayes and Kaun (1996), Hayes and MacEachern (1996, 1998), Hayes (in press), Halle and Dell (in press), Halle (forthcoming) and Kiparsky (2006), there are three PARALLELISM, which essential can ones be subdivided as MAXBEAT, - SALIENCY and follows: a) Metrical CON: a.l) MATCHSTRESS: a. 1.1) strong syllables fall a. 1.2) a syllable that falls on strong on an beats S beat must be stressed a.2) FILLSTRONG: fill the strongest positions in the line a.3) *LAPSE: avoid the strongest sequences with syllables between no any two of positions in the line b) Grouping CON: b.l) SALIENCY (truncation): non-filling of metrical positions at the end of lines: b.1.1) lines b. 1.2) are couplets b.l.3) stanzas salient are are salient salient b.2) PARALLELISM: b.2.1) the cadences ending the units of the maximal analysis of a quatrain must be identical b.2.2) the onsets of syllables coincide with musical beats. Both the interaction and the conflicts among and are determined by the type of language that we The conclusions reached in corpus chapter 4 of 444 Spanish and 239 English folk 275 were the above constraints determine are dealing with. based songs, and on on the metrical analysis of a the metrical and grouping analysis of four case studies, of which two correspond to folk Spanish, and two correspond to art song in English and Spanish. The main conclusion derived from these stress-placement as a in English and song analyses is that the importance of determining factor in text-setting well-formedness is systematically diminished in syllable-timed languages, while it is enhanced in stresstimed languages. In English, the link between verse prosody and musical metre is has stronger than in Spanish. In OT terms, the metrical CON MATCHSTRESS enormous impact text-setting of on an the process English between musical beats and other hand, a or of text-setting in English. According to this, if the is to be qualified as linguistic stresses must be well-formed, the agreement as high possible. On the as syllable-timed language like Spanish will ignore MATCHSTRESS in order to concentrate stressed song an on unstressed - the one-to-one correspondence between syllables and musical beats - - be it weak. In other words, be it strong or Spanish observes the grouping CON PARALLELISM, which follows from the fact that Spanish syllables verse per is predominantly isosyllabic. In English, where the number of line is secondary to the position of stresses, the observation of PARALLELISM is not The required for lines to be judged well-formed. analyses carried out in this dissertation point to between the verse and music differently in like a rhythm, on on one hand, syllable-timed language like Spanish than in a stress-timed language natural counterpoint a speech prosody and musical rhythm, in the second type this counterpoint is exceptionally rare. The small mismatches in the English corpus could be regarded nearly insignificant instances of compositional clumsiness that signal folk song in the the other. However, these inconsistencies work English. While in the first type of languages there is between correspondence timing typologies of language and rhythmic typologies of music. Both English and Spanish show mismatches between speech prosody, and a or as imitate the style. On the other hand, the continuous radical stress mismatches observed Spanish folk and art song would make nonsense of an English text because English word stress is vital to the rhythmic arrangement of the language at all levels. In other words, there exists a difference in kind in relation to the dialogue between prosody and music for each of the two types of languages. In English, the level of agreement between the two rhythmic patterns is really high, while in Spanish the 276 counterpoint between the two is actually used only described is, therefore, not constraints for expressive device. What I have as an difference in the ranking of metrical and grouping a English and Spanish, but a radical disparity in the degree of violation into which each constraint is allowed to incur. This entails that the rhythmic nature of English and Spanish is essentially different. Whether believe in the old or not we dichotomy between stress- and syllable-timed languages, a structural difference in the way we must admit that there is that these two languages articulate rhythm. The initial aim of this work of research providing was new insights into the study of linguistic prosody and text-setting. I firmly believe that this aim has been achieved. Moreover, new research level, there is still much work needed in order to build At the theoretical consistent paths have been opened for the future. theory of text-setting. Let methods of mention the potential application of the analysis developed for these data to which could be chosen from traditions sub-Saharan tradition. This would 'mora-timed' order to us languages, such as a new diverse as folk the song and art song corpora, Celtic, the Japanese, or the imply dealing with what linguists have called Japanese, and reaching beyond stress languages in as explore tone languages. As a previous step, given that this study has focused of the data, it is still necessary to be able to qualitative analysis apply quantitative methods of analysis in order to re-analyse the data collected from which will add to the on a a statistical point of view, something strength of my conclusions. Last, but not least, certain analytical flaws in the current state of the text- setting theory must be addressed. These flaws have to do with the lack of analysis of two elements which essential in the are analysis of text-setting in any musical tradition, namely the role of pitch and the placement and realisation of melismata. Up until now, the focus of text-tune syllables and beats from element has been On the one prominence, metrical point of view, in such a way that the melodic systematically neglected in favour of the rhythmic/metric element. hand, it is essential to determine the role of pitch in the perception of as well as the interaction between stress and pitch in vocal music taken from different traditions will a alignment studies has been the relationship between - analysing the setting of a tone language such certainly bring really significant insights into 277 a a Mandarin theory which has exclusively languages. If we accept that the three major dealt with stress pitch, loudness and duration, are which focuses exclusively on we cannot accept an cues for linguistic stress integral theory of text-setting the interaction between the strength (loudness) of beats syllables, leaving pitch and duration aside. On the other hand, in practically all and text-setting traditions, melisma is interesting interacts in ways a widely used device. The placement of melismata with the assignment of pitch and duration to specific notes/syllables. I believe that, to a certain extent, melisma linguistic and poetic phenomenon of synalepha, regulate the that occurrence so can be made parallel to the that explaining the constraints of melismata could help us understand the nature of synalepha in language. In develop short, new some of the potential specific steps which could be taken in order to research relate to a) the definition of the role of pitch in the realisation and perception of prominence in the b) a corpus analysed in this dissertation statistical analysis of rhythmic and pitch mismatches in English and Spanish c) the selection of a new corpus: Ugandan folk music, Japanese traditional music, Scottish traditional music the analysis of the d) on using OT implemented constraints, focusing new corpus beat-syllable mismatches, pitch mismatches, melismatic mismatches, closure e) an f) a in-depth comparison with the results in the old statistical classification of the g) the discussion of the new songs in the new corpus corpus results h) the exploration of linguistic and musical implications. In sum, this dissertation opens up new possibilities of research within the field of phonology, thus contributing to the development of traditional disciplines such prosody as or metrics. There are also potential applications for it in disciplines stemming from the findings of music theory and theoretical linguistics, but also applied linguistics, sociolinguistics, ethnomusicology and most promising currents of practical research to which involves the well as the use of songs even my research could contribute in the teaching and acquisition of positive impact of rhythm and melody 278 on medicine. One of the a second language, as pupils with learning disabilities. References Abercrombie, David (ed.) (1973[1964]) In honour of Daniel Jones. London: Longman. Abercrombie, David (1967) Elements of general phonetics. Edinburgh: Edinburgh University Press. Abercrombie, Lascelles (1923) Principles of English prosody, Part 1. London: M. Seeker. Adams, Stephen (1997) Poetic design: tables an introduction to meters, verse form, and ofspeech. Ontario: Broadview. Allen, George D. (1975) "Speech rhythm: its relation to performance universals and articulatory timing". Journal ofPhonetics 3:75-86. Allen, William S. (1973[ 1964]) "Quantity and quantitative verse". Eds. David Abercrombie Fry. In honour of Daniel Jones. London: and Dennis Butler Longman, 3-15. Allen, William S. (1973) Accent and rhythm. Cambridge: Cambridge University Press. Aske, Jon (1990) "Disembodied rules of Spanish stress versus patterns (testing the psychological reality rules)". Berkeley Linguistics Society 16:30-45. Attridge, Derek (1982) The rhythms ofEnglish poetry. London: Longman. Baehr, Rudolf (1970) Manual de versificacion espahola. Madrid: Gredos. Spanish translation by K. Wagner and F. Lopez Estrada. Balaguer, Joaquin (1954) Apuntes castellana. Madrid: para una historia prosodica de la metrica Consejo Superior de Investigaciones Cientificas. Balbin, Rafael de (1962) Sistema de rltmica castellana. Madrid: Gredos. Baltzer, Rebecca, Thomas Cable and James Wimsatt (eds.) (1991) The union ofwords and music in medieval poetry. Austin, Texas: University of Texas Press. Banfield, Stephen, "Butterworth, George". Ed. L. Macy. Grove Music Online (Accessed 15 May 2006), http://www.grovemusic.com Barkanyi, Zsuzsanna (2002) "A fresh look at quantity sensitivity in Spanish". Linguistics 40 (2):375-394. Barlow, Michael (1997) Whom the Gods love: the life and music of George Butterworth. London: Toccata Press. 279 Barnwell, Thomas P. (1971) "Relation of syllable boundaries to vowel durations". Journal ofthe Acoustical Society ofAmerica 50:117. Barry, Barbara R. (1990) Musical time: the sense of order. New York: PendragonPress. Beare, William (1957) Latin verse and European song. London: Methuen & Co. Beckman, Mary E. (1986) Stress and non-stress accent. Dordrecht: Foris. Becquer, Gustavo A. (1963 [ 1871 ]) Rimas. Madrid: Espasa-Calpe. Bell, Alexander Melville (1876) Visible speech. London: Simpkin, Marshall & Co. Bernstein, Leonard (1976) The unanswered question. Cambridge, Mass: Harvard University Press. Bertinetto, Pier Marco and Michele Loporcaro (eds.) (1988) Certamen phonologicum: papers from the 1987 Cortona phonology meeting. Torino: Rosenberg & Seller. Bertinetto, Pier Marco (1989) "Reflections on the dichotomy 'stress- vs. syllable- timing'". Revue de Phonetique Appliquee 91/93:99-130. Blacking, John (1973) How musical is man? Seattle: University of Washington Press. Bloomfield, Leonard (1933) Language. New York: H. Holt & Company. Bolinger, Dwight (1958) Interrogative structures of American English: the direct question. Alabama: University of Alabama Press. Bolinger, Dwight (1965) Forms of English: accent, morpheme, order. Tokyo: Hokuou. Bolinger, Dwight (1981) Two kinds of vowels, two kinds of rhythm. Bloomington, Ind.: Indiana University Linguistics Club. Bond, Zinny S. and Joann Fokes (1985) "Non-native patterns of English syllable timing". Journal of Phonetics 13:407-420. Boone, Graeme M. (1999) Patterns in play. Lincoln, Neb.: University of Nebraska Press. Burling, Robbins (1966) "The metrics of children's American verse: a cross-linguistic study". Anthropologist 68:1418-1441. Burzio, Luigi (1994) Principles of English stress. Cambridge: Cambridge University Press. Butterworth, George (1911) "Loveliest of trees". Six London: Augener Ltd. 280 songs from 'A Shropshire lad'. Byshee, Edward (1968[1702]) The art ofEnglish poetry. Menston: Scholar Press. Cable, Thomas (1974) The meter and melody of Beowulf, Urbana: University of Illinois Press. Cable, Thomas (1991) "The meter and musical implications of Old English poetry". Eds. Rebecca Baltzer, Thomas music in medieval poetry. Cable and James Wimsatt. The union of words and Austin, Texas: University of Texas Press, 49-71. Cable, Thomas (1996) "Clashing stress in the metres of Old, Middle, and Renaissance English". Eds. Christopher McCully and John Julian Anderson. English historical metrics. Cambridge: Cambridge University Press, 7-29. Cabre, Teresa and Pilar Prieto (2003) "Vowel contact resolutions across word boundaries in Catalan". Eds. Maria Romero. Josep Sole, Daniel Recasens and Joaquim Proceedings of the 15th international conference of phonetic sciences, Volume II, Barcelona: Universidad Autonoma de Barcelona, 1687-1690. Canellada, Josefa and John Kuhlmann Madsen (1987) Pronunciacion del espahol. Madrid: Castalia. Catford, John C. (1977) Fundamental problems in phonetics. Edinburgh: Edinburgh University Press. Cemuda, Luis (1996) Antologia poetica. Ed. Philip Silver. Madrid: Alianza Editorial. Chomsky, Noam and Morris Halle (1968) The sound pattern of English. New York: Harper and Row. Chomsky, Noam (1971) Selected readings. Eds. J. P. B. Allen and Paul Van Buren. London: Oxford University Press. Classe, Andre (1939) The rhythm ofEnglish prose. Oxford: Oxford University Press. Coleman, C. L. (1974) A study of acoustical and perceptual attributes of isochrony in spoken English. Ph.D. Dissertation. Washington D.C.: University of Washington. Cooper, Grosvenor and Leonard B. Meyer (1963) The rhythmic structure of music. Chicago: University of Chicago Press. Couper-Kuhlen, Elizabeth (1986) An introduction to English prosody. London: Edward Arnold. Couper-Kuhlen, Elizabeth (1993) English speech rhythm: form and function in everyday verbal interaction. Amsterdam: Benjamins. 281 Crystal, David (1969) Prosodic systems and intonation in English. Cambridge: Cambridge University Press. Cummins, Fred and Robert Port (1998) "Rhythmic constraints on stress timing in English". Journal of Phonetics 26:145-171. Cureton, Richard D. (1992) Rhythmic phrasing in English verse. London: Longman. Cureton, Richard D. (1993) "The auditory imagination and the music of poetry". Ed. Peter Verdonk. Twentieth century poetry: from text to context. London: Routledge. Dauer, Rebecca (1983) "Stress-timing and syllable-timing reanalyzed". Journal of Phonetics 11:51-62. Delattre, Pierre (1951) Principes de phonetique frangaise a I'usage des etudiants anglo-americains. Middlebury, Vt.: Ecole Franfaise D'Ete, Middlebury College. Delattre, Pierre (1965) Comparing the phonetic features of English, French, German and Spanish: an interim report. Heidelberg: Groot. Delattre, Pierre (1966) Studies in French and comparative phonetics: selected papers in French and English. The Hague: Mouton. Dell, Francois, and John Halle (in press) "Comparing musical text-setting in French and in English songs". Ed. Jean-Louis Aroui and Andy Arleo. Towards a typology ofpoeticforms. Amsterdam: Elsevier. D'Introno, Francesco, Enrique del Teso and Rosemary Weston (1995) Fonetica y fonologia actual del espahol. Madrid: Catedra. Dogil, Grzegorz (1979) Autosegmental account ofphonological emphasis. Edmonton: Linguistic Research. Dowling, W. Jay and Dane L. Harwood (1986) Music cognition. Orlando: Academic Press. Downing, A. Thomas (1995) Music and the origins of language. Cambridge: Cambridge University Press. Duffell, Martin (1996) "Chaucer, Gower, and the history of the hendecasyllable". Eds. Christopher McCully and John Julian Anderson. English historical metrics. Cambridge: Cambridge University Press, 210-218. Duffell, Martin (1999) "Modern metrical theory and the 'verso de arte mayor'". Papers of the Medieval Hispanic Research Seminar 10. London: Queen Mary and Westfield College. 282 Duffell, Martin (2000) "The Santillana factor: the development of double audition in Castilian". Ed. Alan Seminar Deyermond. Papers of the Medieval Hispanic Research (PMHRS) 28:113-28. London: Queen Mary and Westfield College. Duffell, Martin (2002) "Mio Cid studies: 'some problems of diplomatic' fifty years on". Papers of the Medieval Hispanic Research Seminar (PMHRS) 42. London: Queen Mary and Westfield College. Eddington, David (2000) "Spanish stress assignment within the analogical modelling of language". Language 76 (1):92-109. Eliot, Thomas Stearns (1942) Music ofpoetry. Glasgow: Jackson. Fabb, Nigel (1997) Linguistics and literature. Oxford: Blackwell. Fabb, Nigel (2002) Language and literary structure. Cambridge: Cambridge University Press. Fant, Lars (1984) Estructura informativa en espahol. Estudio sintactico e informativo. Upsaliensia 34. Stockholm: Almqvist & Wiksell International. Studia Romanica Finnegan, Ruth (1977) Oral poetry: its nature, significance and social context. Bloomington and Indianapolis: Indiana University Press. Fitzgerald, Colleen M. (1998) "The metre of Tohono O'Odham songs". International Journal ofAmerican Linguistics 64(1): 1-36. Fraisse, Paul (1964) The psychology of time. London: Eyre and Spottiswoode. Garcia-Bellido, Paloma (1993) "La cantidad fonologica Ralph Penny. Adas del primer congreso en el espanol actual". Ed. anglo-hispano, Volume I. Madrid: Castalia, 151-168. Garding, Eva (1977) The Scandinavian word accents. Lund: CWK Gleerup. Gasiorowski, Piotr (1998) Optimal stress for English. Poznan: Motivex. Gasparov, Mikhail Leonovich (1996) A history of European versification. Oxford: Clarendon Press. Gauldin, English translation by G.S. Smith and Marina Tarlinskaja. Robert (1995) A practical approach to sixteenth/eighteenth-century counterpoint. Prospect Heights, 111.: Waveland Press. Giegerich, Heinz J. (1978) "On the rhythmic stressing of function words: a modest proposal". Work in progress. Edinburgh: Department of Linguistics, University of Edinburgh, 42-51. 283 Giegerich, Heinz J. (1985) Metrical phonology and phonological structure: German and English. Cambridge: Cambridge University Press. Giegerich, Heinz J. (1992) English phonology: an introduction. Cambridge: Cambridge University Press. Gili Gaya, Samuel (1950) Elementos defonetica general. Madrid: Gredos. Glass, Leon and Michael C. Mackey (1988) From clocks to chaos: the rhythms of life. Princeton: Princeton University Press. Grammont, Maurice (1933) Traite dephonetique: avec 179 tables dans le texte. Paris: Delagrave. Gray, Catherine T (1998) "Temporal heterogeneities in Kiganda music". Eds. Costin Miereanu and Xavier Hascher. Les universaux congres sur en linguistique: actes du quatrieme la signification musicale. Paris: Sorbonne, 617-627. Guthriw, Steven (1991) "Meter and performance in Machaut and Chaucer". Eds. Rebecca Baltzer, Thomas Cable and James Wimsatt. The union of words and music in medieval poetry. Austin, Texas: University of Texas Press, 72-100. Halle, John, and Fred Lerdahl (1993) "A generative text-setting model". Current musicology 55:3-23. Halle, John (forthcoming) "Text, tune and metrical form". Halle, Morris and Roman Jakobson (1964) Fundamentals of language. The Hague: Mouton. Halle, Morris and Samuel J. Keyser (1971) English stress: its form, its growth, and its role in verse. New York: Harper & Row. Halle, Morris and Jean-Roger Vergnaud (1987) An The Massachusetts Institute of Technology Halliday, Michael (1967) Intonation and essay on stress. Cambridge, Mass.: Press. grammar in British English. The Hague: Mouton. Halliday, Michael (1973) Explorations in the functions of language. London: Edward Arnold. Halliday, Michael (1985) Spoken and written language. Victoria: Deakin University. Halliday, Michael (1994[ 1985]) An introduction to functional grammar. Arnold. Hamer, Enid (1966[ 1930]) The metres of English poetry. London: Methuen. 284 London: Hanson, Kristin and Paul Kiparsky (1996) "A parametric theory of poetic meter". Language 2/2:287-335. Harris, James W. (1983) Syllable structure and stress in Spanish. Cambridge, Mass.: The Massachusetts Institute of Technology Press. Hayes, Bruce (1995) Metrical stress theory. Chicago: University of Chicago Press. Hayes, Bruce (in press) "Text-setting Andy Arleo. Towards a as constraint conflict". Ed. Jean-Louis Aroui and typology ofpoeticforms. Amsterdam: Elsevier. Hayes, Bruce and Abigail Kaun (1996) "The role of phonological phrasing in sung and chanted verse". The Linguistic Review 13:243-303. Hayes, Bruce and Margaret MacEachern (1996) "Are there lines in folk poetry?". UCLA Working Papers in Phonology 1:125-142. Hayes, Bruce and Margaret MacEachern (1998) "Quatrain form in English folk verse". Language 74(3):473-507. Heaney, Seamus (1999) Beowulf. London: Faber and Faber. Heffner, Rose-Merrill (1949) General phonetics. Madison, Wis.: University of Wisconsin Press. Hogg, Richard and Christopher B. McCully (1987) Metrical phonology: a coursebook. Cambridge: Cambridge University Press. Housman, A.E. (1896) A Shropshire Lad. London: K. Paul, Trench, Treubner. Hualde, Jose Ignacio and Monica Prieto (2002) "On the diphthong/hiatus contrast in Spanish: some experimental results". Linguistics 40 (2):217-234. Huggins, A.W.F. (1972a) "Just noticeable differences for segment duration in natural speech". Journal ofthe Acoustical Society ofAmerica 51:1270-1274. Huggins, A.W.F. (1972b) "On isochrony and syntax". Eds. Gunnar Fart and Marcel Andre Tatham. Auditory analysis and perception of speech. London: Academic Press, 455-464. Hyman, Larry M. (1975) Phonology, theory and analysis. New York: Holt, Rinehart, and Wilson. Hyman, Larry M. (ed.) (1977) Studies in stress and accent. Los Angeles, Ca.: University of Southern California. Hyman, Larry (1985) A theory ofphonological weight. Dordrecht: Foris Publications. 285 Jackendoff, Ray and Fred Lerdahl (1983) A generative theory of tonal music. Cambridge, Mass.: The Massachusetts Institute of Technology Press. Jackendoff, Ray (1989) "A comparison of rhythmic structures in music and language"-. Eds. Paul Kiparsky and Gilbert Youmans. Phonetics and phonology, Volume 1, rhythm and meter. San Diego, Ca.: Academic Press, 15-44. Jakobson, Roman (1960) "Closing statement: linguistics and poetics". Ed. Thomas A. Sebeok. Style in language. Cambridge, Mass.: The Massachusetts Institute of Technology Press, 350-377. Jakobson, Roman (1966) Poetics. The Hague; Paris: Mouton Warszawa, Polish Scientific Publishers. Janda, Richard D. and Terrell A. Morgan (1988) "El acento dislocado - castellano: on other Romance - pues cantado explaining stress-shift in song-texts from Spanish (and certain languages". Eds. David Birdsong and Jean-Pierre Montreuil. Advances in Romance linguistics. Dordrecht: Foris, 151-170. Johnson, Samuel (1818) A dictionary ofthe English language. London: Longman. Jones, Daniel (1950[ 1909]) The pronunciation of English. Cambridge: Cambridge University Press. Jones, Daniel (1956[ 1917]) English pronouncing dictionary. New York: E.P. Dutton. Kant, Immanuel (1884) Critique of translation pure reason. London: George Bell. English by J.M.D. Meiklejohn. Kennedy, Peter (1984[ 1975]) Folk songs of Britain and Ireland. London: Oak Publications. Kiparsky, Paul (2006) "A modular metrics of folk verse". Eds. Elan B. Dresher and Nila Friedberg. Formal approaches to poetry. Berlin, New York: Mouton de Gruyter, 7-52. Kramer, Lawrence (1984) Music and poetry: the nineteenth century and after. Berkeley: University of California Press. Ladd D. Robert and W. N. from Campbell (1991) "Theories of prosodic structure: evidence syllable duration". Actes du Xlleme congres international des sciences phonetiques, Volume 2. Aix-en-Provence: Service des Publications, Universite de Provence, 290-293 286 Ladefoged, Peter (1967) Three areas of experimental phonetics: stress and respiratory activity, the nature of vowel quality, units in the perception and production of speech. London: Oxford University Press. Ladefoged, Peter (1975) A course in phonetics. New York: Harcourt Brace Jovanovich. Langer, Suzanne (1953) Feeling andform: in a new a theory of art developedfrom philosophy key. London: Routledge. Lass, Roger (1994) Old English. Cambridge: Cambridge University Press. Laver, John (1994) Principles ofphonetics. Cambridge: Cambridge University Press. Lehiste, Ilse (1970) Suprasegmentals. Cambridge, Mass.: The Massachusetts Institute of Technology Press. Lehiste, Ilse (1977) "Isochrony reconsidered". Journal ofPhonetics 52:53-263. Lerdahl, Fred, and Ray Jackendoff (1983) A generative theory of tonal music. Cambridge, Mass.: Massachusetts Institute of Technology Press. Liberman Mark and Alan Prince (1977) "On stress and linguistic rhythm". Linguistic Inquiry 8:249-336. Liberman, Mark (1979) The intonational system of English. New York: Garland Publishing. Lipsky, John M. (1997) "Spanish word stress: the interaction of moras and minimality". Eds. Fernando Martinez-Gil and Alfonso Morales-Front. Issues in the phonology and morphology of the major Iberian languages. Washington DC: Georgetown University Press, 559-594. Lloyd, A.L. (1967) Folk song in England. London: Laurence and Wishart. Lloyd, Paul M. (1993) Del latin al espahol (I. Fonologlay morfologla historicas de la lengua espahola). Madrid: Gredos. Spanish translation by Adelino Alvarez Rodriguez. London, Justin, "Rhythm". Ed. L. Macy. Grove Music Online (Accessed 1 March 2006), http://www.grovemusic.com Luce, Gay Gaer (1971) Biological rhythms in human and animal physiology. New York: Dover Publications. MacPherson, Ian (1975) Spanish phonology: descriptive and historical. New York: Barnes and Noble. 287 Major, Roy C. (1981) "Stress-timing in Brazilian Portuguese". Journal of Phonetics 9:343-351. Malof, Joseph (1970) A manual of English meters. Bloomington, Ind.: Indiana. University Press. Manzano Alonso, Miguel (ed.) (2001) Cancionero popular de Burgos: I Rondas y canciones. Manzano Burgos: Diputacion Provincial de Burgos. Alonso, Miguel (ed.) (2003) Cancionero popular de Burgos: V Canciones del ciclo anual y vital. Burgos: Diputacion Provincial de Burgos. McCully, Christopher B. (1996) "Domain-end phenomena and metrical templates in OE verse". Eds. metrics. Christopher McCully and John Julian Anderson. English historical Cambridge: Cambridge University Press, 42-58. Mellers, Wilfrid (1987) Le jardin retrouve. York: The Fairfax Press. Miller, M. (1984) "On the perception of rhythm". Journal ofPhonetics 12:75-83. Minkova, Donka (1997) "Constraint ranking in Middle English stress-shifting". English language and linguistics 1:135-175. Mompou, Federico (1980) "Olas gigantes". Canciones becquerianas. Barcelona: Tenora edicions musicals. Morgan, Terrell A., and Richard D. Janda (1989) "Musically-conditioned stress shift in Spanish revisited: empirical verification and nonlinear analysis". Eds. Carl Kirschner and Jannet Ann Decesaris. Studies in Romance proceedings from the XVII linguistic symposium on linguistics, selected Romance languages. New Jersey: Rutgers University, 273-288. Napoli, Donna Jo (1975) "A note on synalepha and stress maxima". Poetics 4:401- 410. Navarro Tomas, T. (1956) Metrica espahola. New York: Syracuse University Press. Navarro Tomas, T. (1965[1918j) Manual depronunciacion espahola. Madrid: Gredos. Nespor, Marina and Irene Vogel (1986) Prosodic phonology. Dordrecht: Foris Publishers. Nunez-Cerdeno, Rafael A. and Alfonso Morales-Front (1999) Fonologla generativa contemporanea de la lengua espahola. Washington D.C.: Georgetown University Press. 288 Olsen, C. L. (1972) "Rhythmical patterns and syllabic features of the Spanish sense- group". Proceedings of the ninth international congress ofphonetic sciences. The Hague: Mouton, 990-995. Penny, Ralph (1991) A history of the Spanish language. Cambridge: Cambridge University Press. Pensom, Roger (1998) Accent and metre in French. Bern: Peter Lang. Piera, Carlos (1980) Spanish verse and the theory of meter. PhD Dissertation. Los Angeles, Ca.: University of California Los Angeles. Pike, Kenneth L. (1945) The intonation of American English. Ann Arbor, Mich.: University of Michigan Press. Pike, Kenneth L. (1948) Tone languages: A technique for determining the number and type of pitch contrasts in a language, with studies in tonemic substitution and fusion. University of Michigan Publications in Linguistics 4. Ann Arbor: University of Michigan Plato Press. (1998) Republic. Oxford: Oxford University Press. English translation by Robin Waterfield. Pointon, Graham E. (1980) "Is Spanish really syllable-timed?". Journal of Phonetics 8:293-304. Pound, Ezra (1917) "Vers libre and Arnold Dolmetsch". Egoist July:90-91. Preminger, Alex (ed.) (1974) Princeton encyclopedia of poetry and poetics. London: Macmillan. Prince, Alan and Paul Smolensky (1993) Optimality Theory: constraint interaction in Generative Grammar. New Brunswick, NJ: Rutgers University. Quilis, Antonio (1984) Metrica espahola. Madrid: Ariel. Quilis, Antonio (1993) Tratado de fonologlayfonetica espaholas. Madrid: Gredos. Quinlan, John (1959) "A.E. Housman and British composers". The musical times. Volume c, 1393:137-138. Raven, David Sebastian (1965) Latin metre, an introduction. London: Faber and Faber. Roach, Peter (1982) "On the distinction between 'stress-timed' and 'syllable-timed' languages". Ed. David Crystal. Linguistic controversies. London: Arnold, 73-79. 289 Roca, Iggy (1986) "Secondary stress and metrical rhythm". Phonology Yearbook 3:341-370. Roca, Iggy (1988) "Theoretical implications of Spanish word stress". Linguistic- Inquiry 19:393-423. Roca, Iggy (1997) "On the role of accent in stress systems: Spanish evidence". Fernando Martinez-Gil and Alfonso Morales-Front. Issues in the Eds. phonology and morphology of the major Iberian languages. Washington DC: Georgetown University Press, 619-664. Roca, Iggy and Wyn Johnson (1999) A course in phonology. London: Blackwell. Ross, John R. (1972) "A reanalysis of English word stress". Ed. M. Brame. Contributions to generative phonology. Austin, Texas: University of Texas Press. Russell, Ian, "England (i) II: Traditional music". Ed. L. Macy. Grove Music Online (Accessed 16 May 2006), http://www.grovemusic.com Schlegel, August Wilhem von (1883) Lectures London: G. Bell and Sons. Translated on dramatic art and literature. by John Black. Schliiter, Julia (2005) Rhythmic grammar. Berlin, New York: Mouton de Gruyter. Schwitters, Kurt (1993[ 1926]) Poems, performance pieces, proses, plays, poetics. Philadelphia: Temple University Press. English translation and edition by Jerome Rothenberg and Pierre Joris. Scruton, Roger (1997) The aesthetics of music. Oxford: Clarendon Press. Selkirk, Elisabeth O. (1984) Phonology and syntax: the relation between sound and structure. Cambridge, Mass.: The Massachusetts Institute of Technology Press. Sharp, Cecil (1954[ 1907]) English folksong. London: Methuen and Co. Sonnenschein, Edward Adolf (1925) What is rhythm? Oxford: Blackwell. Sosa, Juan Manuel (1999) La entonacion del espahol. Su estructura fonica, variabilidady dialectologla. Madrid: Catedra. Stevens, Denis (1960) A history of song. New York: The Norton Library. Stoquerus, Gaspar (1988[ 1570]) De musica verbali libri duo: two books music. Music on verbal Edition, translation and introduction Albert C. Rotola. Greek and Latin Theory 5. Lincoln: University of Nebraska Press. Sweet, Henry (1878) Handbook ofphonetics. Oxford: Clarendon Press. 290 Todorov, Tzvetan (1988) Literature and its theorists: a personal view of twentieth- century criticism. London: Routledge and Kegan Paul. English translation by Catherine Porter. Tovey, Donald Francis (1979[1949]) "Words and music: stream some obiter dicta". The main of music and other essays. New York: AMS Press, 202-219. Trager, George L. and Henry Lee Smith (eds.) (1951) Outline of English structure. Norman, Okla.: Battenburg Press. Trubetzkoy, Nikolai S. (1969) Principles of phonology. Berkeley: University of California Press. van Leeuwen, Theo (1999) Speech, music, sound. London: Macmillan. Veltman, Joshua (2001) Textual accent and mensural rhythm in sixteenth-century setting of the 'Pater Noster'. Master's thesis: The Ohio State University. Weil, G. (1960) "Arud". The encyclopedia of Islam, London: H.A.R. Gibb et al. new edition, Volume I. Leiden, (ed.), 675ff. Weinreich, Uriel and Andre Martinet (eds.) (1954) Linguistics today. New York: Linguistic Circle. Wenk, Brian J. and Franfois Wioland (1982) "Is French really syllable-timed?". Journal ofPhonetics 10:177-193. Wightman, Colin, Stephanie Shattuck-Hufnagel, Mari Ostendorf and Patti J. Price (1992) "Segmental durations in the vicinity of prosodic phrase boundaries". Journal of the acoustical society ofAmerica 91:1707-1717. Woodrow, Herbert (1951) "Time perception". Ed. Stanley Smith Stevens. Handbook ofexperimental psychology. New York: Wiley, 1224-1236. Yavas, Mehmet and Brian Goldstein (1998) "Phonological assessment and treatment of bilingual speakers". American journal ofspeech-language patholology 7:49-60. Youmans, Gilbert (1996) "Reconsidering Chaucer's prosody". Eds. Christopher McCully and John Julian Anderson. English historical metrics. Cambridge: Cambridge University Press, 185-209. 291 Appendices Chapters 1 -4: Appendix I: Glossary of terms and symbols a) Terms: suprasegmental mechanism displayed by languages in order to signal Accent: prominence. Accent be subdivided into three main classes, can the other two being tone and pitch. Allophone: phonetic variant of a phoneme in a Ambisyllabicity: phenomenon by the onset of unstressed an means syllable can a one of which is stress, particular language. of which a consonantal segment working also function as as the coda of the preceding syllable if this is stressed. Amphibrachic: a metrical foot consisting of short/unstressed syllable either side of it. a long/stressed syllable with Anapaest: foot formed by two unstressed positions and a stressed one, one in this order. Aphaeresis: suppression of the first syllable of a word. Apocope: suppression of the last syllable of a word. Bar line: the vertical line Beat: regular pulsation; placed a on the staff to divide the music into bars. basic unit of length in musical time. Clausal intonation: rise and fall of the voice Coda: component pitch. of the syllable rhyme that includes the segments that follow the syllabic nucleus. Compression: the phrase, clause the process or verse and result of reducing the number of syllables in line by joining the nuclei of two or more a word, of those syllables (i.e. vowels) together. by raising the blade of the tongue. Coronal: consonant Couplet: a stanza Dactyl: metrical foot consisting of one accented syllable followed by two unaccented or a of one consisting of two successive lines of verse; usually rhymed. long syllable followed by two short. Decompression: the Degenerate foot: Diacritic: articulated a a process and result of undoing compression. foot that lacks one or more graphic mark added to distinguish words that are a positions. letter to indicate otherwise graphically identical. 292 a special phonetic value or separation Dieresis: the or resolution of one syllable into two; the opposite of synaeresis. Directionality: the serial order in which linguistic phenomena take place. shifting of the position of stress within Diastole: forward to the next a word, whereby the stress goes syllable in a word. Dipod: the basic foot of dipodic syllable, a verse, lightly accented syllable, consisting (when complete) of an unaccented an unaccented syllable, and a heavy accented syllable, in that succession. Diphthong: union of two vowel sounds pronounced in syllable. tendency to elide the unstressed vowel whenever there Elision: next to one are two equal vowels each other. Epenthesis: addition of a syllable in the middle of a word. Euphony: agreeable sound, especially in the phonetic quality of words. Extra metrical: verse a syllable which does not count as a metrical position when scanning a line. Foot: the basic unit used in the scansion Fixed stress or measurement of verse. (languages): languages which place lexical stress always on the same syllables of words. Fricative: a consonant produced by the forcing of breath through a constricted passage. Also called spirant. Gerund: the term verbal present generally is used to refer to the present participle in English and the participle of Spanish. These English and -ando or progressive form the are the verb forms that end in '-ing' in -(i)endo in Spanish. In both languages, the gerund is used to or continuous tenses. The English gerund is frequently used as a noun. Glide: a vowel-like sound that serves as a Hiatus: two vowels which appear consonant; semivowel. in succession and are pronounced in separate syllables. Hemistich: of the line a by half line of verse, especially when separated rhythmically from the rest a caesura. metrical foot Iambic: a accented syllable. consisting of one 293 unaccented syllable followed by one Ictus: the accent that falls on a stressed syllable in Isochrony: regular occurrence in time of an event Leader-timed a or line of scanned verse. events. (languages): languages that create right-headed stress groups where the. joins with the preceding unstressed syllables to create the stress group. tonic Lexeme: a minimal unit (as Lexicon: the lexemes of a a word or stem) in the lexicon of a language. language considered as a group; a language user's knowledge of words. Margin: the coda of a syllable, which comprises the consonant sounds that follow the syllabic nucleus, which is usually Middle English (ME): name a vowel. given to the diverse forms of the English language spoken between the Norman invasion of 1066 and the mid-to-late 15th century. Medial: in the middle of a word. Melisma: a Metre: any case passage of several notes sung to one syllable of text. kind of patterning, either consciously used and delimited by rules, of verse, or unconsciously used and loosely regulated, speech; the basic pattern contained in a as in the case as in the of normal particular piece of verse; the template plus a specified amount of tension. Mode: any scalewise arrangement of pitches; more generally, the term refers to the patterns upon which medieval music was structured, which preceded the development of major and minor scales and tonality. Mora: the minimal unit of metrical time in quantitative verse. Morpheme: the smallest meaningful linguistic unit, which cannot be divided into smaller meaningful parts. Nominative: the category Nucleus: the of nouns serving as the grammatical subject of a verb. point of the syllable when the airstream is least obstructed - the sonorant peak, normally a vowel. Old English (OE): what is now an early form of the English language that was spoken in parts of England and southern Scotland between the mid-fifth century and the mid-twelfth century. Oxytone: a word that has a heavy stress accent on its last syllable. Paragoge: addition of a syllable, typically constituted by the vowel Id, at the end of a word. 294 Paroxytone: a Pitch accent syllable word with stress on the penultimate syllable. (languages): those languages that assign prominence to or mora Quatrain: serving to distinguish phonemes a stanza or poem Rhythm: that property of a the observer the groups accented of a word using variations in pitch (frequency). Present-day English (PdE): the English language Phonemic: an or as is spoken nowadays. distinctive features. of four lines. sequence of events in time which produces on the mind of impression of proportion between the duration of the several events of events of which the Scansion: the process sequence of measuring or is composed. verse, that is, of marking accented and unaccented syllables, dividing the lines into feet, identifying the metrical pattern, and noting significant variations from that pattern. Schwa: a mid-central neutral vowel (/o/), typically occurring in unstressed syllables. Sliding: the formation of a glide. Sonorant: usually voiced speech sound characterized by relatively free air flow a through the vocal tract and capable of being syllabic, vowel, liquid, as a or nasal. Sonority: the degree of resistance of an articulation to free airflow through the vocal higher resistance equals lower sonority. tract, where Stochastic OT: view of OT where constraints have one can Stress: read off the actual a phonetic peak Stress-timed: a or a range of ranking positions and frequency of different variants. climax occurring typically in every word. type of rhythm where stressed syllables tend to occur (at least perceptually) at equal intervals in time. Stop: a consonant the airflow. produced with Stops may affricates, where there is a complete closure in the vocal tract which interrupts be subdivided into oral stops, essentially plosives and no nasal airflow, and nasal stops, commonly just called nasals, where the soft palate is lowered and the airstream escapes via the nasal cavity. Suppression: all the phenomena that result in the elimination of a syllable within a word. Syllable: a phonological structure composed of speech sounds. The syllable is the domain of association for such phenomena as accent, stress 295 and lexical tone. Syllables are generally considered to be composed of nucleus and coda. Words Syllable-timed: is an a type are a number of constituents: onset, rhyme, of syllables. made up of rhythm where syllables tend to have equal duration. French. example of a language said to be preponderantly syllable-timed. Syllable weight: the weight of syllable depends a the make-up of its rhyme on constituent. If the rhyme consists of a long vowel (or diphthong) followed by zero or consonants (a so-called strong cluster), the syllable is heavy. If the rhyme more consists of then the a short vowel followed by maximum of one consonant (a weak cluster), a syllable is light. Synalepha: that takes place when two process different words that come into contact in a or more rhythm vowels which belong to group are combined to form a single syllable. Synaeresis: process that takes place when two word but different same syllables are vowels which belong to the or more combined to form a single syllable. Syncope: suppression of a middle syllable. Syncopate: to shift the accent in music by putting it Systole: shifting of the position of stress within to Tactus: term for and 16th a beat, i.e. centuries. In motion of the a times per hand) stood for goes back hand movement, used in the a a downward and an upward semibreve/whole-note in normal tempo, and a man was breathing normally (between 60 and 70 minute). compositions by a Tenseness: vowel Timbre: the sound a theory, each tactus (representing Tempo: the speed at which music is descriptive language: a on written metronomic direction to the performer. quality. quality of production or ought to be played, often indicated or or a musical note or sound that distinguishes different types of musical instruments. language, as Mandarin distinguishes the meanings of words that Tonic word, whereby the stress unit of time measured by approximately equal to the pulse of Tone normally unaccented beat. previous syllable. the 15th a on a or Somali, in which pitch are syllable: stressed syllable. 296 otherwise the same or the pitch contour phonologically. Trochee: metrical a unaccented consisting of foot accented syllable followed by one one syllable. Truncation: the shortening of a metrical verse line by omitting a syllable or syllables (usually unstressed) from the full complement expected in the regular metrical pattern. Typology: the study traits in or Trillo: systematic classification of languages that have characteristics common. Trailer-timed tonic joins or (languages): languages that create left-headed stress with the following unstressed syllables to create the stress Unmetrical: a line or poem phoneme: vowel. Voiced: uttered with vibration of the vocal cords. length: the perceived duration of a vowel sound, b) Symbols: * incorrect, unacceptable : or non-existing form. syllable division. -: " : concept or term; English translation. /: strong \: weak position. position. > : 'leading to'; 'more < : 'coming from'; 'less open than'. open than'. (): translation into English. [ ] : phonetic transcription. 'plus'. + : »: / /: 'ranked higher than'. phonemic transcription, position x: metrical 0: non-projected syllable. Bold: mismatched Italics: group that does not conform to the established foot patterns. Vowel where the musical trill. a Vocalic groups syllables. non-English terms. Underlined: synalephas. 297 or syllable Chapter 2: Speech prosody Appendix II: Instances of synalepha and synaeresis in a corpus of 444 Spanish folk songs (Manzano Alonso 2003) Synalepha: 1. 1.1. Between unstressed vowels: 1.1.1. Mismatched u-na (falling es-tre-lla. ha-bi-ta en, que el. v a, que on strong en. que ha-bi-ta. so a. la en-ho-ra-bue-na. gra-cia rran-do a, ce-le-bra pa' en. a-pa-gar. 1.2. Between an pa-ra Se-vi-lla a, es-ta. a. ya-dri-no que es. v. li-bre hier-ba en. v al. tri-so v. to-do el. he en-tra-do. ten- v. que a-si. que a-yer, pi-san- si al-gu-no. mi a-bue-la. bo- que el, a-le-gri-a di-sa-le us-ted. cor-te-si-a ha. ban-de-ra v a, por-que v u-na, 1.2.2. Not mismatched: 2. en, v. ha. ca-ba-lle-ro hon-ra-do. unstressed and 1.2.1. Mismatched: a ex-tien-das. duer-ma us-ted, el. ce-le-bra en. v. ha-bi-ta pi-dien-do-[e no-che en-tra-ra. a-e-no-ia a-e-no-ia. la han. pa-ra a-, si es. la en-ho- ra-bue-na, pa-ra em-, ie-na. me la has. a 1.1.2. Not mismatched: en. he. en-fri-a. la han. su-be se do es-pi-nas. te es-ta. mar-cho un. ca-sa es. can-to a. es-ta-mos. co-mo es. ca-sa el pa-ra que ca-sa no-che en-tra-ra. li-cen-cia he. beats): man-da-do de e-lla. a a al. fres-qui-ta v. do-ce a-pos-to-lo-rum. v el. v un stressed vowel: de a-ho, o-tro a-ho, puer-ta he-mos. si y u-na. que es-ta, pa-sa-ra es-ta Synaeresis: Des-a-ho-gue-mos 298 es us-ted. San-ta A-gue-da. ce-ba' a- Chapter 4: The relationship between speech, verse and music Appendix III: Index of OT text-setting terms and constraints with examples Notation: TERM: pages CLOSURE: 202, 203, 240, 247, 248, 270 English example: X X X X X X X X X X X X X X The sun went down be- yond yon hills X (4) X X X X X X X X X X X X A- cross yon drea- ry moor X (3) Spanish example: X X X X X X X X los pas- to- X X X X X X X X X que res X X X X X X X queel ni- noes- X X X ta- X X X X X baen X X Be- CON: 157, 158, 165, 176, 179, 180, 181, 182, 183, 206,219, 267, 268 CON X X X X X ron X X X X pie- su- X ranking: 181-182, 187, 204, 205,267 Constrained: 193,229 299 X x X x X x len oo 186, 49, 70, 81, 95, 103, 110, 111, 133, 139, 142, 154, 155, 156, Constraint/-s: II, 4, 27, 169, 176, 179, 182, 184, 186, 194, 196, 197, 200, 201, 202, 203, 204, 206, 219, 223, 225, 226, 227, 228, 232, 233, 234, 235, 236, 251, 256, 261, 262, 267, 269, 270, 274; 277, 280, 281 *FILL EXTRA WEAK: 186, 187 English example: Not attested in case studies (see Hayes in press) Spanish example: x x x x X XX X X XX XX XX X XX queel ni- XX FILLSTRONG: baen ta- noes- XX X X XXXXX len Be- 180, 205, 225, 226, 267 English example: X X X X X X xxxxxxxx The sun down went be- yond yon hills Spanish example: X X X XXX X X XXX A- le- X X X XXXXXXXXXXXXX gri- *LAPSE: 181, X X a ba- ca- 183, 184, 186, 187, 205, 226, 233, 267 English example: X X X X X X X X X X X A- cross yon drea- ry moor 300 lie- ros Spanish example: X X X X X X X X X queel ni- noes- X X X X X X X baen ta- X X X X X X X X X X X 0 len Be- X 179, 180, 182, 183 LONG-LAST: English example: Not attested in studies (see Hayes and MacEachern case 1998) Spanish example: Llevadme por con la piedad a donde el vertigo razon me arranque la memoria. 2, 6, 10 3 4, 6, 10 3 3,4, 6, 10 4 (2), 4, 6 2(3) jPor piedad! jTengo miedo de quedarme con mi dolor a solas! MATCHSTRESS/MAXBEAT: 165, 166, 167, 181, 203, 204, 205, 206, 207, 219, 223, 267, 268 English example: x x X x x X X x X X Up to Jf fc— s (ftr « =^=.N \-0-i X x X X x x X the far- mer's door s— 9 1—9 boy there came Up_ to_ day From this break of x \i •—J 9 9 ■ d the far-mer's door cold win-try blast? Spanish example: O- las g'- gan- Lle- vad- me por O- las_ gi - gan- tes_ tes dad pie- lie- vad 301 - me por pie- dad X PARALLELISM: 180, 182, 192, 194, 198, 202, 203, 204, 205, 206, 207, 226, 227, 228, 234, 235, 246, 247, 248, 253, 256, 257, 261, 267, 268 English example: X X X X X X X X X X X X X X The sun went down be- yond yon hills X X X X X X X X X X X X X A- cross yon drea- ry moor X 0 Spanish example: X X X X X X X X A- le- gri- X X X X XX a X X X X X X X * no- ble- X X X X ros X X X X X X lle- ba- ca- X X X X X X fies- X X X XX ta X X X X de X X X los X X X X X Re- X X yes RUN-ON: 187 English example: Not attested in case studies (see Hayes in press) Spanish example: Not attested in case studies SALIENCY: 164, 176, 177, 179, 180, 184, 202, 203, 204, 205, 226, 227, 234, 267 English example: X X x x x x xxxxxxxx The sun went down x x be- yond yon hills x x x x xxxxxxxx 302 (4) A- drea- yon cross (3) 0 moor ry Spanish example: X X X X X X X los pas- to- X X X X X X X X X X X X queel ni- fioes- X X X X X baen Be- 181, 182 English example: Across yon down, beyond dreary When weary yon hills 4 3 moor and lame, a boy there 4 came Up to the farmer's door. And if that thou won't One me 3 employ 4 thing I have to ask Will you 3 shelter me, till break of day From this cold 4 wintry blast? My father's dead, my 3 mother's left 4 With her five children small 3 And what is 4 worse for mother still I'm the eldest of them all 3 Spanish example: Los los la Reyes Reyes venidos, ya son ya son manana, primer fiesta del due se celebra en Esnana. 4 supieron 4 el nino estaba han dejado las ovejas Que me 4 4 que han anretado 4 ano Los pastores, que v XX X ron en Belen. a correr. quieres decir, nifio, X X X X ta- STANZA CORRESPONDENCE: went X X X X sun X X X X The X X pie- su- que res X X X X X X X X 3(4) 4 3(4) 4 303 X X len X X X X XX X con ese Me dedo pinado? 4 quieres llevar a juicio? Perddname los 4 pecados. 4 (UN)FAITHFULNESS: 203, 204, 207, 213 English example: X X X X X X X X X X X X X X The sun went down be- yond yon hills (4) X X X X X X X X X X X X A- cross yon drea- ry moor X X (3) Spanish example: X X X X X X X X X los pas- to- X X X X X X X X que res X X X X X X queel ni- noes- X X X X X ron X X X X X pie- su- X X X X X X ta- X X X X X X X baen X Be- 304 X X X X len X X X X X Appendix IV: Mismatches between stresses and beats in Manzano Alonso (2003) Notation: Song number in the collection (subgenre: dance, 'd', Time or not signature Type of mismatch (lexical or grammatical): word category; mismatched word (number of times that the word is mismatched in the mismatched 1096 (nd) lexical: noun ca-pu-chi-na 1097(nd) 3/8 lexical: noun gen-te (twice) 1098(nd) 2/4 lexical: full verb fue-ron noun hi-jo noun mon-te (twice) full verb man-da-do noun un puer-tas grammatical: preposition (prep.) de determiner (det.) el (nd) 3/8 lexical: noun same song) syllabified with syllable/s in bold and instances of synalepha underlined 3/8 1100 dance, 'nd') ca-ri-ta grammatical: conjunction (conj.) que 305 prep, de 1101(nd) 3/4 lexical: pronoun (pron.) de-nos (second semi-quaver of first beat and second beat of bar) grammatical: conj. y prep, 1102 de (twice) (nd) 3/8 lexical: noun ca-ri-dad 1104a(nd) 3/8 grammatical: prep, con prep, 1105 de (twice) (nd) 2/4 lexical: adjective (adj.) a-ma-bles (three demi-semiquavers of second beat) full verb grammatical: can-ta-mos as above) prep, con det. lo 1106 (same (twice) (nd) 6/8, 9/8 lexical: noun, full verb ca-sa es relative pronoun grammatical: (second beat of 6/8) (rel. pron.) que (second beat of 9/8) (twice) prep, a 306 1108 (nd) 2/4 lexical: full verb su-pli-car (second beat) 1110(nd) 6/8 lexical: noun ca-sa (second beat) (twice) full verb ro-ba 1111 (nd) 2/4 lexical: full verb so-li-ci-tan-do grammatical: 1112 prep, va-ra ha (nd) 2/4 lexical: full verb can-to grammatical: 1113 a prep, a (nd) 2/4 grammatical: prep, por (twice) conj. y (three times) 1114 (nd) 2/4 lexical: auxiliary (aux.) verb, full verb da-do (first and second beats) 1115(nd) 2/4 lexical: noun tri-pa noun ce-le-min 307 grammatical: det. la has prep, de 1116a(nd) 3/8, 9/8 lexical: full verb ha-bi-ta full verb que en es-ta-mos grammatical: det. demonstrative (demons.) es-te conj. que es- prep. en 1116b (nd) 3/4, 4/4 lexical: noun re-yes noun, grammatical: adj. ca-ba-lle-rohonpron. me det. los 1116c(nd) 3/8 grammatical: conj. y 1116d (nd) 3/8 grammatical: conj. y 1116e (nd) 3/8 lexical: noun ca-sa adj. con-ten-tos (twice) 308 1117(nd) 4+3+4/8 lexical: noun no-che noun a-mas full verb pi-den noun -na adverb es-tre-lla (adv.) cuan-do grammatical: det. demonst. es-ta det. u-naes- conj. co-mo es 1118a(nd) 2/4 lexical: noun re-yes noun es-tre-lla (twice) grammatical: conj. + det. 1118b v u-na es- (nd) 3+2/4 lexical: noun ca-sa grammatical: el prep, pa-ra det. el 1119a(nd) 2/4 lexical: full verb e-cha-do (twice) grammatical: conj. + det. para que-el (twice) det. los 1119b(nd) 2/4 lexical: adj. bue-nos 309 noun ma-na-na grammatical: prep, de a-no det. los 1119c(nd) 2/4 lexical: noun re-yes grammatical: conj. y det. los (twice) conj. + det. 1119d que el (nd) 3/8 lexical: noun se-no-res full verb sa-len grammatical: rel. pro. que (twice) det. los conj. v u-na 4+3+3/8 lexical: noun re-yes full verb lu-ce 2/4 lexical: full verb e-cha-do 1120a (nd) 2/4 grammatical: prep.+ det. del 1120d(nd) 2/8, 3/8 grammatical: prep, de (nd) 1120f 2/4 lexical: noun re-yes 1120g (nd) 2/4 grammatical: 1120h prep, de (twice) prep. det. del (nd) 2/4 lexical: noun li-cen-cia noun pas-to-res noun na-ci-mien-to full verb 1120i de-jan-do (nd) 3/8 lexical: noun per-jui-cio (twice) noun Be-len noun grammatical: ja-rri-ta prep. + det. del (twice) prep, a (twice) 1120j (nd) 2/8, 3/8 lexical: noun O-rien-te full verb ba-jan noun u na grammatical: es-tre-lla prep, por det. u-na 311 1120k (nd) 3+5/8 lexical: noun re-yes (twice) full verb ce-le-bra en grammatical: det. la 1121(nd) 4/8, 5/8 noun a-le-sri-a grammatical: pron. nos lexical: 1122 v (nd) 2/4, 3/4 lexical: noun O-rien-te noun Je-ru-sa-len noun u-tta es-tre-lla grammatical: conj. + prep. y_a det. 1123 u-na (nd) 2/4 lexical: noun re-yes noun per-jui-cio grammatical: 1124a prep. + det. del (nd) 6/8 lexical: 1124b noun re-yes (twice) (nd) 6/8 lexical: full verb que ha-bi-ta en grammatical: prep, en 1125a(nd) 3/8 lexical: noun li-cen-cia full verb que ha-bi-ta en full verb can-tar grammatical: rel. pron. que ha- prep. -ta en prep, a conj. que det. los 1125b (nd) 3/8 grammatical: prep. + det. del (twice) conj. y prep, a prep, en det. los 1125c(nd) 6/8 lexical: noun re-yes (three times) full verb ce-le-bra grammatical: det. los 1126 (nd) 2/4 lexical: noun ca-sa noun he a-le-srl-a grammatical: conj. v co-mo en 1127 (nd) 4+5/8 grammatical: det. el 1128 (nd) 2/4 lexical: full verb ha-bi-ta en full verb na-ci-do grammatical: det. e-sos (second beat) (second beat) prep, a conj. que es-ta prep, en 1129 (nd) 2/4 lexical: full verb noun grammatical: 1130 ba-ja na-ci-mien-to prep, a (nd) 2/4 lexical: full verb grammatical: 1131 me ex-tien-das pron. meex- (nd) 3/8 lexical: noun cho-ri-zos noun man-za-nas grammatical: conj. si 314 1132a(nd) 3/4 lexical: full verb grammatical: 1132b va-sa-ra us- pron. lo (nd) 3/4 lexical: noun A-gue-da grammatical: rel. 1133 pron. que (nd) 3/4 lexical: noun adj., vis-pe-ra noun San-ta A-eue-da (3rd beat) full verb sa-li-do grammatical: 1134 prep, de (nd) 6/8 lexical: noun re-gi-dor full verb duer-me noun ni-ha full verb, pronoun duer-ma us-ted full verb des-can-se full verb vie-ne noun grammatical: 1135 cua-dri-lla prep, de (nd) 4/4 lexical: full verb te-ne-mos 315 1136a(d) 6/8 lexical: adj. San-to (second beat) noun Fran-cia noun ce-ba' a noun ni-nos (second beat) grammatical: (second beat) de (three times) prep, rel. pron. que (twice) pron. nos det. la (second beat) (three times) (three times), det. los, det. el conj. + prep. + det. v al conj. y (second beat) prep, para (second beat), prep, 1136c 6/8 lexical: noun grammatical: Se-vi-lla a pron. nos det. el prep, (second beat) (twice) (second beat) de (second beat) prep, a 1137 6/8 lexical: noun po-co full verb com-para-mos grammatical: pron. nos det. el, (twice) det. lo 1138 2/4 lexical: noun se-ho-res 316 pa' (second beat) noun ca-pi-tan grammatical: conj. si 1139 7/4 grammatical: det. nues-tro 1140 3/8 lexical: noun se-no-res grammatical: prep, pa-ra 1141 5/4 lexical: noun ge-ne-ral noun e-jer-ci-to 1142a 2/4 lexical: full verb noun grammatical: can-tar di-a pron. no-so-tros det. mu-cha 1142c 2+3+3+2/4 lexical: noun li-cen-cia (second and third beats) full verb me-ter-nos grammatical: (first and second beats) prep. + det. del prep, a (second beat) pron. nos (second beat) 317 1142d 3/4, 4/4 grammatical: pron. le 1143 2/4, 3/4 lexical: noun ga-lli-to full verb da-do grammatical: prep. + (twice) det. del 1144a 3/4, 2/4 lexical: full verb pa-gar full verb sa-li-do a full verb de-be-mos grammatical: pron. dl-ga-le us- 1144b 6/4 grammatical: conj. aun-que pron. nos conj. pe-ro 1146 2/4 grammatical: det. la (twice) 1147 6/8 lexical: noun gi-ta-nas full verb adj .fe-as so-mos full verb na-ci-do noun chi-me-ne-a grammatical: conj. det. que he-mos u-na 1148a 2/4, 3/4 lexical: full verb vie-nen grammatical: conj. y det. la pron. nos 1148b 4/4 lexical: full verb pe-di-mos grammatical: conj. y, conj. mas 1149 6/8 grammatical: det. el 1150 2/4 grammatical: pron. se, pron. 1151a 6/8 grammatical: det. las 1152b 2/4 lexical: noun Ca-ta-lu-na le noun cor-te-si-a ha grammatical: prep, pa-ra (five times) det. to-do conj. si (six times) 1153 2/4 grammatical: conj. si (five times) 1154b 3/4 lexical: noun ban-de-men noun tri-eo v prep, de 1154c 3/4 lexical: grammatical: conj. 1154f 3/4 lexical: noun ban-de-men 1155a 2/8 lexical: full verb re-su-ci-to grammatical: prep, pa-ra det. u-na 1155b 2/8 (2/4) lexical: full verb res-su-ci-to (second beat) 320 1158 2/4 lexical: full verb re-su-ci-ta 1159 2/4, 3/4 lexical: noun Jue-ves noun ca-sa grammatical: prep, de (three times) conj.j; 1160 3/8 lexical: noun cho-ri-zo noun es-cue-la grammatical: prep, v de, prep, a, prep, pa- pron. nos, pron. conj .y det. o-tro a-ho 1161 2/4 lexical: noun Se-hor full verb ve-ni-mos full verb can-tan-do full verb se-ri-a noun cho-ri-zo full verb noun grammatical: v que-de-se lon-ga-ni-za prep, de prep, a pron. nos le la pron. conj .y conj. que det. o-tro a-ho 1164 3/8 lexical: full verb que-re-mos 1167 2/4 grammatical: prep, pa-ra 1168 2/4 lexical: full verb re-su-ci-to grammatical: prep, con, de 1169 2/4 lexical: noun an-ge-les 1170 3/4 lexical: full verb pi-dien-do-le a noun to-do noun ca-be-za full verb tra-e-mos adj. car-ga-do grammatical: det. u-na det. las (twice) prep, pa-ra (twice), prep, pa' 322 prep. + pron. det. al le, pron. nos rel. pro. que conj.^, conj. que 1171 2/4 lexical: full verb pa' a-pa-sar 1172 6/8 grammatical: det. det. a es-ta u-nos pron. den-nos 1173a 3/8 lexical: noun li-cen-cia noun an-gel 1173d 3/8 lexical: 1173e 3/8 lexical: noun mar-zo 1173g lexical: noun mar-zo (twice) 1173h 3/8 lexical: noun mar-zo 1173j 3/8 lexical: noun mar-zo full verb flo-re-cen 1173k 3/8 lexical: noun mar-zas 11731 2/8,3/8 grammatical: pron. nos 1173m 2/8, 3/8 lexical: noun Ma-ri-a 1174a 3/8, 2/4 lexical: noun se-no-res noun me-dia noun me-dia 1174b 3/8 lexical: grammatical: det. tu 1174c 2/4 lexical: noun mu-jer noun li-cen-cia noun me-dia noun li-cen-cia 1175a 3/8 lexical: (twice) 1175b 3/8, 2/4 lexical: (twice) full verb -che en-tra-ra (second beat of 2/4) 1175c 3/8 lexical: noun se-no-res noun (twice) li-cen-cia grammatical: prep.+ det. del (twice) 1175d 3/8 lexical: noun se-no-res (twice) noun li-cen-cia (twice) noun li-cen-cia 1175e 3/8 lexical: 1175f 3/8 lexical: 325 1175g 3/8 lexical: noun li-cen-cia (twice) noun li-cen-cia (twice) noun se-fio-res (twice) noun li-cen-cia noun li-cen-cia noun per-mi-so (twice) noun li-cen-cia (twice) noun li-cen-cia (twice) 1175h 3/8 lexical: 1175i 3/8 lexical: 1175J 3/8 lexical: 1175k 3/8 lexical: (twice) 11751 3/8 lexical: 1175m 3/4 lexical: 1175n 3/4 lexical: full verb -che en-tra-ra (twice) noun me-dia noun li-cen-cia noun me-dia noun li-cen-cia noun ve-ci-no noun me-dia 1175ft 3/4 lexical: (twice) (twice) 1175o 3/8 lexical: noun e-so 1175p 3/8 lexical: noun li-cen-cia noun me-dia (twice) grammatical: prep. + det. del conj.j> 1175q 3/8 lexical: (twice) noun me-dia noun li-cen-cia noun me-dia 1175r 3/8 lexical: (twice) 1175s 3/8, 2/4 lexical: (twice) grammatical: det. vues-tras (twice) 1175t 3/8 noun li-cen-cia (twice) norm li-cen-cia (twice) grammatical: pron. nos lexical: 1175u 3/8 lexical: 1175v 3/8 lexical: noun li-cen-cia norm me-dia 1175x lexical (noun): li-cen-cia he 1175z 2/4, 3/8, 2/4 lexical: full verb se en-fri-a grammatical: det. tu pron. se en-fri-a pron. va-mo-nos 1179 3/8, 2/4, 3/8 lexical: noun se-no-ras noun co-rri-llo (twice) full verb sa-li-do noun vi-lla-no (twice) (twice) full verb -la a-e-no-ia (twice) full verb pa-sa noun Za-ra-go-za (twice) noun vi-to-res noun ma-yo noun Fer-nan-do grammatical: prep, pa-ra prep, (twice) (twice) (twice) ha-cia 1180 3/8 lexical: full verb di-ces (twice) noun va-cas grammatical: prep, con, prep, en det. los conj. por-que (twice) 1181 6/8 lexical: (twice) noun ma-yo noun ca-lo-res noun ce-ba-das noun cier-nen noun e-na-mo-ra-dos 1182 6/8 grammatical: det. los (twice) 1185 5/8 lexical: adj. en-ra-ma-di-ta 329 noun ma-ha-na grammatical: det. u-na pron. la ban 1186 2/4 lexical: noun mo-zas full verb va-mos 1187 3/4, 2/4, 2/4, 2/4 lexical: noun ti-na-ja (three times) full verb su-be v (three times) grammatical: det. la prep, con, prep, a prep. + (three times) det. del conj. si adv. ya conj. su-be v (three times) 1188 4/4, 6/8 grammatical: det. las, det. la (twice) prep, a 1189 6/8, 4/4 lexical: adj. San adverb cuan-do grammatical: det. la (twice) prep, a, prep, con 330 1190 6/8,2/4, 3/4 lexical: nounpu-che-ro (three times) (three times) noun ri-ve-ro noun dl-a noun Jua-ni-to noun a-mor (twice) grammatical: det. la prep, con, prep, rel. pron. que sin, prep, a (twice) (twice) conj. si 1192 Recitado lexical: nounper-mi-so noun se-no-res noun to-do el grammatical: det. el 1193 3/4 lexical: adj.po-bre grammatical: prep, en rel. pron. que 1194 2/4 lexical: full verb co-mer grammatical: det. el 1195,1196 2/4 331 lexical: adj. po-bre-ci-tas adv. co-mo full verb que-dan grammatical: det. las conj. que, pron. les conj. y 1200 6/8 lexical: noun co-ra-zon full verb he en-tra-do 1201 6/8 grammatical: prep, a det. las 1202a 6/8, 2/4-3/4, 3/8 lexical: full verbpo-ne-mos noun li-cen-cia (four times) full verb ve-ni-mos grammatical: pron. (three times) la 1202b 2/4-3/4, 3+2+3/8, 3+2+3/8 lexical: full verb po-ne-mos full verb des-pe-di-ras adj. sol-te-ras (twice) full verb se_a-pa-gan noun to-dos grammatical: conj. y (three times) 332 (twice) prep, con 1202(II)c 2/4, 3/4 lexical: noun li-cen-cia full verb pi-do grammatical: (second beat of 2/4) prep, pa-ra, prep, a det. es-te 1202(II)d lexical: noun li-cen-cia (twice) 1202(II)e lexical: full verb ten-eo a grammatical: det. to-do det. mis det. mi (twice) (second beat of 2/4) prep, ten-eo a 1202(II)f lexical: quie-res 1202(II)g lexical: noun la en-ho-ra-bue-na grammatical: det. laenconj .y 1203 3/4, 5/8 grammatical: det. tu 333 1204 2/4 lexical: noun tre-bol full verb flo-rez-ca noun grammatical: sra-cia prep, en -cia en-tre 1205 5/8 lexical: noun in-ten-cion noun li-cen-cia noun la en-tra-da 1206 3/4/7/8 lexical: adj. Bien-ve-ni-dos grammatical: det. la en- 1207a 5/4 lexical: noun grammatical: vuer-ta he-mos prep, a, prep, de 1207b 6/8 lexical: noun no-via noun ya-dri-no noun ma-dri-na grammatical: rel. v pron. que conj. pa-dri-no v 1207c 3/4 lexical: lull verb lle-ga-do cua-dri-lla noun full verb sen-te-mos numeral vein-ti-cin-co noun si-llas (twice) (twice) 1207d 3/4 lexical: noun cu-ra noun pa-dres noun I 'a-com-pa-fia-mien-to grammatical: conj. que v (twice) (twice) 1208a 5/4 lexical: noun hier-bas 1208b grammatical: rel. pron. que conj. co-mo, conj. co-mo 1208c 10/8 lexical: noun grammatical: Ma-ri-a prep, pa-ra prep, sin 1208e 2/4 lexical: noun li-cen-cia a 335 grammatical: aux. verb he-mos prep, con, prep, de, prep, pa-ra 1208g 3/4, 2/4 lexical: noun bal-co-nes noun la- en-ho-ra-bue-na grammatical: conj. que det. la- en-ho-ra-bue-na 1208h 10/8 lexical: noun po-sa-da adj. gui-a-dos adj. pu-li-da 12081 2/4 lexical: noun Ma-ri-a full verb di-ce noun puer-ta adj. con-ce-bi-da grammatical: prep. + prep, det. al sin, prep, de pron. nos 1209 3/4 lexical: numeral vein-ti-cin-co grammatical: prep, a conj. que 336 Im 1210 3/4 lexical: noun ma-ri-do grammatical: prep.+ det. al 1211 4/4 lexical: full verb des-pi-de-te tonic pron. de e-lla (twice) grammatical: det. la (twice) 1212 3/4 lexical: noun pa-re-des adv. don-de 1213 3/4 grammatical: det. la conj. por-que conj. que prep, por 1214a 3/4 lexical: rvo\mA-ve noun grammatical: pe-ca-do prep, pa-ra 1214b 3/4 lexical: noun ga-lan 337 tonic pron. que grammatical: pron. e-lla la 1215 3/4, 3/8, 5/8 lexical: noun cu-cha-ras grammatical: conj. y 1216a 3/4 lexical: noun to-mi-llo noun la-do noun tie-rra noun to-mi-llo 1216b 6/4 lexical: ad), flo-ri-do noun no-via noun la-do 1217a 2/4, 3/4 lexical: full verb ten-gan noun grammatical: com-pa-ne-ras prep, a (twice) det. mis 1217b 5/8 lexical: adj. bue-nas noun com-pa-ne-ras 1217c 2/4 lexical: adv. bien 1217d 3/4 lexical: noun li-cen-cia noun no-via full verb des-a-ho-sue-mos full verb grammatical: -ve a-co-bar-dar prep, pa-ra a- det. nues-tras particle no 1217e 4+3+2+3/8 grammatical: prep, de conj. que, conj. ^ 1217f 4+5/8 lexical: adj. bue-nas (three times) 1217g 4+4+3/8/4+5+3/8 grammatical: conj. que det. la 1217h 6/8 lexical: adv. bien 1217i 4+2+3/8 lexical: ra-mi-tos noun grammatical: (twice) pron. se prep.a det. las 1217J 3/4, 7/8 lexical: full verb de-ja 1217k 3/4, 7/8 grammatical: pron. se (twice) 12171 2/4 lexical: noun com-pa-ne-ra noun pa-nue-lo full verb grammatical: des-pi-das pron. te (twice) 1217m 2/4, 3/4 lexical: noun in-ten-cion full verb po-de-mos full verb em-pe-zar full verb si grammatical: es aux. verb he-mos prep, de (twice) conj. si es iiiii 1217n 2/4 lexical: noun in-ten-cion full verb pe-di-mos grammatical: det. a es-ta aux. verb he-mos prep, de, pron. les prep, pa-ra 12172 4/4 lexical: noun grammatical: A-ve (twice) prep, pa-ra 1217o 2/4 lexical: noun puer-ta adj. li-bre v grammatical: det. el (twice), det. la rel. pron. que (twice) 1217p 6/8 (3/4) lexical: noun in-ten-cion grammatical: det. dem. a es-ta prep, pa-ra 1217q 6/8 lexical: noun ra-mi-tos 341 1217r 4/8, 3/8 lexical: li-cen-cia noun (twice) full verb em-pe-ce-mos adj. fe-li-ces adv. que grammatical: a-si prep, con (twice) conj. a-si que 1217s X lexical: noun se-nor grammatical: det. la pron. nos 1217t 2+3/8 lexical: adverb mas que a-ver grammatical: det. la (twice) rel. pron. conj. don-de mas que a-ver 1217u 3/4 grammatical: prep, con, prep, conj. de (twice) que 1217v 3/4 lexical: full verb des-pi-de-te, des-pl-de-te adj. pri-me-ra adj. ul-ti-ma 342 :i! 1218 2/4 lexical: noun Ma-ri-a full verb (twice) lle-gar 1220 6/4 nounpa-dri-no lexical: noun con-vi-da-dos 1221 3/4, 2/4 lexical: noun ma-na-na noun a-mor (twice) 1222a 6/8 lexical: noun full a-rro-yo verb, noun vi-san-do es-vi-nas 1222b 6/8 grammatical: det. tu det. to-das 2/4 lexical: full verb a-bran-se grammatical: prep, de 1222c 6/8 lexical: full verb noun de-ja (twice) gen-te (twice) 343 full verb pa-sar noun la-do full verb tees-ta grammatical: (twice) pron. te es-ta pron. (twice) lo 1222d 2/4 lexical: full verb ten-gan 1222e 3/4, 7/8 lexical: full verb po-ner grammatical: prep, de 1224a 3/4, 5/8 lexical: full verb in-co-mo-dar grammatical: prep, con prep.+ det. del pron. nos 1224b 3/4, 5/8 lexical: full verb su-bes grammatical: prep. + det. al prep, por 1225 6/8 grammatical: det. la conj. que 1226a 2/4, 4/4 lexical: noun grammatical: per-mi-so aux. verb he-mos prep, con, prep, de, prep, pa-ra 1226c X grammatical: conj. y (three times) conj. cuan-do 1226d 4+3+3/8 lexical: full verb ve-ni-mos grammatical: conj. y prep, a det. la en-ho-ra-bue-na es-ta 1229 3/4 lexical: noun la en-ho-ra-bue-na (twice) grammatical: det. la en-ho-ra-bue-na (twice) prep, a es-tos det. to-dos prep. dos (twice) + det. al (twice) 1230 6/8, 3/4 lexical: noun grammatical: se-ho-res prep, pa-ra em- particle no 345 1231 6/8 lexical: adv. a-qui grammatical: pron. les prep. + det. al prep a prep. + conj. v a 1232 6/8 grammatical: det. la (twice) 1233 3/4 lexical: noun po-sa-da full verb ve-ni-mos (twice) adj. pu-li-do (twice) grammatical: det. la prep, a (twice) 1234 6/8 grammatical: prep, pa-ra pron. que 1235 6/8 lexical: full verb su-bir (twice) grammatical: det. es-ta (twice) 1238 2/4 lexical: adj. don-de-te grammatical: pron. cd-me-le det. la det. Su 1239 2/4 grammatical: det. la, det. el 1240 3/4 grammatical: det. su 1241a 2/4 lexical: adj. San grammatical: pron. le 1241b 2/4 lexical: adj. San grammatical: pron. le 1242 2/4 lexical: full verb mar-cho grammatical: prep, mar-cho det. tus prep. 1243 X a En a lexical: adj. tem-pra-na grammatical: pron. duer-me-te 1244 2/4 grammatical: prep. + det. al (three times) prep, por 1245 3/8 grammatical: det. es-te 1246 3/4 grammatical: conj. que 1247 2/8, 3/8 lexical: full verb duer-me grammatical: conj. + prep, coni. co-mo en por-aue al 1248 2/4 grammatical: det. el 1249 2/4 grammatical: prep, sin 1252 2/8 348 lexical: full verb Ila-man-do grammatical: conj. conj. que (twice) a det. los pron. (twice) (twice) duer-me-te (twice) 1253 2/4 lexical: adj. dor-mi-do grammatical: pron. duer-me-te pron. duer-me-te pron. duer-me-te 1254 2/4 grammatical: 1256 3/4 grammatical: det. la (twice) If. 1257a 2/8 grammatical: prep, de, prep, con (twice) 1257b 2/8 lexical: adj. fres-aui-ta grammatical: v prep, de, conj. que pron. de-ja-la prep, con 1259 3/4 349 grammatical: det. mi det. las 1261a 3/8 lexical: nounpa-jaen grammatical: conj. + det. si el prep, de prep, va-ia en 1261b 3/8 lexical: full verb vie-nen (three times) grammatical: det. los (twice) 1262a 2/4 grammatical: pron. me prep, a 1262b 2/4 grammatical: pron. me prep, a 1263 2/8 lexical: full verb que es grammatical: det. la pron. me rel. pron. que es 350 1267 3/4 lexical: noun a-ve-lla-no noun bue-yes noun la-bra-do-res noun hier-ba en grammatical: det. mi prep, con, prep, de, prep, hier-ba det. los 1268 2/4 grammatical: prep, con, prep, de 1269 2/4 lexical: adj. di-chas (twice) noun Je-ru-sa-len (twice) grammatical: det. los (twice), det. las 1270 2/4 lexical: numeral, noun do-ce a-vos-to-lo-rum full verb man-da grammatical: det. la 1271 2/4, 3/4 grammatical: det. la 1272 2/4 351 grammatical: det. las (three times) 1273a 2/4, 3/4 lexical: numeral la numeral u-na on-ce mil grammatical: conj. >> (three times) conj. que det. la (four times) u-na 1273b 2/4, 3/4 lexical: numeral on-ce mil grammatical: conj. y (four times) det. las 1274a 2/4, 3/4 lexical: noun vir-gen (twice) 1274b 2/4, 3/4 grammatical: det. las (three times) 1274c 3/8, 2/8 grammatical: rel. pron. que (twice) 1275 2/4 grammatical: prep, de (twice) 1276a 3/8 lexical: full verb es-ta-ban 1277a 2/4 lexical: noun rue-da noun rue-da 1277b lexical: grammatical: det. la 1278 2/4 (twice) lexical: full verb te-ner grammatical: conj. y (twice) 1279 2/4 lexical: full verb ca-sar 1281 2/4 grammatical: pron. me 1282a 2/4 grammatical: prep, en 1282b grammatical: prep, en (three times) 1283 2/4 lexical: noun mier-co-les noun jue-ves full verb ha-cer noun sa-ba-do noun do-min-go full verb pue-de noun ma-ha-ni-ta grammatical: det. el (five times) prep, pa-ra (twice) 1284 2/4 grammatical: conj. si (twice) 1285 3/4 lexical: full verb vi-nien-do adj. so-li-ta noun pa-ti-tas full verb se-ra u-na full verb ve-ras (four times) tonicpron. sial-gu-no adv. don-de grammatical: pron. se, pron. me conj. + det. v el. conj. v un conj. sial-gu-no conj. + det. oal-gun prep, con, prep, a det. la, det. los 354 1286 2/4 grammatical: det. el 1287 3/8 grammatical: conj.por-que pron. me 1292 2/4 lexical: noun pa-tro-nes adj. lo-cos noun lu-gar grammatical: det. la prep, de (twice) prep, con det. el conj. + det. y el 1293 2/4 grammatical: prep, en (twice) rel. pron. que 1294 3/4 lexical: 1295 2/4 noun pro-vin-cia noun par-ti-do lexical: noun fron-te-ra grammatical: det. la 1296 2/4 grammatical: pron. se pron. las 1297 2/4 lexical: noun huer-to noun na-bos (twice) 1298 2/4 lexical: noun ca-so noun grammatical: mi a-bue-la pron. les prep, a det. mi a-bue-la 1299 2/4, 3/4 lexical: noun des-gra-cias full verb e-cha-ras 1300 2/4 grammatical: conj. si pron. lo 1302 2/4, 3/4 lexical: noun da-mas (twice) 1304 2/4 grammatical: prep, con 1305a 2/4 lexical: full verb bo-rran-do a (twice) grammatical: det. la prep, con 1305b 2/4 lexical: full verb grammatical: es-tar prep, en Ik 1308 2/4 lexical: (three times) noun vi-no noun Na-va-rra noun to-dos (three times) full verb a-gra-da full verb a-lar-ga 1309 2/4 lexical: noun can-ti-nas noun Chi-na adj. me-jor 357 noun grammatical: to-da prep. + det. del 1310 2/4 grammatical: det. la conj. que (twice) prep, a 1311 2/4 lexical: full verb ti-ra-mos grammatical: det. el 1313 6/4 lexical: adj. pre-pa-ra-da noun noun sar-ten ja-rra 1314 lexical: full verb va-ya grammatical: conj. mien-tras (twice) Appendix V: Mismatches between stresses and beats in Kennedy (1984) Notation: Song number in the collection Time signature Word class, word instance (in italics) with mismatched syllable (in bold), beat in the bar where the mismatched syllable falls (times the mismatch occurs in the same song). Explanation for lexical mismatches. 124 3/4 preposition un-to on the last beat of a bar and the first beat of the next bar. 126 4/4 preposition in on a first beat. 129 6/8 conjunction and on a second beat. 130 3/4 determiner my on a first beat. 133 3/2 lexical mismatch Reasons: (full verb): be-came, third beat on crotchet and crotchet. transcription forces it (a single metrical arrangement is repeated throughout the whole piece); singer's own memory word; not really important, the weakest part of the bar. as it falls 359 on (maybe it was not that 134 4/4 conjunction and on a third beat. 137 3/4 lexical mismatch (noun): mor-ning, third beat of a bar and first two of another (last). Characteristic of English folk song; it would not occur in art 139 3/4 preposition a on a first beat. 141 3/4, 4/4 preposition to on a first beat. 142 6/8 preposition at on a first beat, preposition of on a first beat. on a first beat. 154 4/4 preposition in 156 4/4 determiner (article) the, determiner a on a third beat. 161 3/4 360 song. 1« preposition of on a first beat, auxiliary verb was on a first beat. 162 3/4, 4/4 preposition o-ver on a determiner my on a first beat (twice), first beat. 163 6/8 determiner my on a first beat. 168 3/4 preposition with determiner his on a on a first beat, first beat. 170 * 6/8 preposition of on a first beat. 172 6/8 auxiliary verb pronoun / on was on a a first beat, first beat. 175 3/4 preposition on on a first beat (twice). 176 6/8 361 preposition first beat. on on a Ill 4/4 preposition but on a third beat (twice). preposition in third beat. on a 178 3/4 preposition in-to on first and second beats. 181 3/4 preposition to on a first beat. on on a first beat. 184 4/4 preposition 186 4/4 preposition of on a first beat, determiner my on a pronoun she on a first beat, first beat. 188 6/8 lexical mismatch (noun): Coun-te-rie at the end of a line, preposition of on a first beat, on a first beat. preposition in 205 6/8 362 preposition first beat. on on a 209 6/8 preposition in pronoun I on first beat, on a a first beat. 211 6/8, 9/8 conjunction if on a first beat; conjunction when preposition pronoun first beat, up on a I on a first beat. 212 3/4 preposition by on first beat. a 214 2/4 auxiliary verb is on a first beat. 215 4/4, 6/4 pronoun /, pronoun her on first and third beats. 217 4/4 conjunction when on a first beat. 218 6/8 preposition to on a first beat (three times). 363 on a first beat, 219 4/4 conjunction if on a first beat, particle not on a first beat. 225 4/4 lexical mismatch fourth beat (noun): shoe-ma-kers on a third beat (demisemiquaver) and (dotted semiquaver). It is fine because it is a compound, so the main stress falls 226 4/4 preposition pronoun on on a I on first beat, first beat (twice). a 228 6/8 preposition at pronoun modal I on on a first beat. first beat (end of line and pause). a can on a first beat. 229 6/8 conjunction if on a first beat. 230 4/4 pronoun / on a first beat. 238 4/4 364 on shoe. a conjunction when on a first beat. conjunction if on a third beat. auxiliary verb + particle don't on a first beat. 242 4/4 preposition for on pronoun a first beat, I on a first beat. 248 4/4 preposition to pronoun / on on a a first beat, first beat. 253 6/8 preposition on on a first beat. 255 4/4 on a first beat. determiner my on a first beat, preposition in 258 6/8 determiner pronoun (demonstrative) this on a first beat, / on a first beat twice, pronoun we on first beat. lexical mismatch (noun): game-keep-er on first crotchet, first semiquaver, and second crotchet. It is fine because it is a compound, 365 so the main stress falls on game-. 261 4/4 preposition on on a third beat. 266 4/4 relative pronoun which on a first beat. 267 4/4 preposition in on a first beat. preposition un-til on a first beat. pronoun + auxiliary verb we '11 on a first beat. 268 3/4 preposition to on a first beat. 273 3/2, 2/2 preposition to, preposition of on a first beat. 280 6/8 preposition to on a lexical mismatch first beat. (noun): onepennie on first crotchet, first semiquaver, and second crotchet of the last bar. This (what I call 'country mismatch') is idiomatic of folk happens at the end of a line penny; or stanza. song; it usually Look at the re-spelling of pennie < other cases of this phenomenon are countrie or companie. The unstressed, lax (neutralised) vowel /i/ becomes tense, and therefore, it carry some degree of stress. 366 can lexical mismatch (noun): North Coun-te-rie, on a first and a second beat. 289 6/8 determiner his determiner pronoun first beat. on a (demonstrative) this / on a on a first beat. first beat (twice). 298 6/8 lexical mismatch (noun): la-dy on third semiquaver of first beat and crotchet of second beat. This is not dance song. a As happened with penrtie, the lax, neutralised vowel HI gets tense. 305 4/4 preposition for on a first beat. auxiliary modal verb + particle couldn't on a 314 3/4, 4/4 pronoun it on a first beat, pronoun he on a preposition of on first beat, a first beat. on on a first beat. 315 4/4 preposition 317 6/8 367 first beat. preposition of on a first beat. 320 6/8 auxiliary verb + particle won't on a first beat. 321 4/4 preposition to first beat, on a preposition be-tween relative pronoun on a which first beat, on a first beat, auxiliary verb + particle won't on a first beat. 354 2/4 preposition for on a first beat (twice), conjunction if on a first beat, conjunction as on a first beat, determiner my on a first beat, pronoun you on a first beat. 359 6/8 preposition in on a first beat, melismatic. preposition in on a third beat, pronoun I on a first beat (twice). 368