Phonology for conversation

Journal of Pragmatics North-Holland PHONOLOGY 309 zyxwvutsrqponm 9 (1985) 309-330 FOR CONVERSATION Phonetic Aspects of Turn Delimitation in London Jamaican J.K. LOCAL, W.H.G. WELLS and M. SEBBA * Participants in conversation have at their disposal many ways of showing that their speaking-turn is complete. An important resource for achieving this interactive task is provided by phonetic features. However, the precise role of these features has been obscured because analysts have relied too heavily on their intuitions, particularly about intonational meaning. Drawing on techniques developed within Conversation Analysis we give a precise formulation of the role of phonetic features in turn-delimitation in the speech of London Jamaicans. We show that turn-delimitation in London Jamaican may be signalled by features of pitch, loudness and rhythm centred on the last syllable of the turn. In this respect, London Jamaican is different from some other varieties of English. 1. Introduction It is generally accepted nowadays that prosodic features can have a variety of interactive functions, and that the explication of prosodic systems must be a priority if we are to further our understanding of how conversations work. (Heritage (1984) Levinson (1983)). It is also recognised, by some at least, that the methods currently employed to describe these features and the interactive work they do are not altogether satisfactory. One reason is that these methods rely crucially on the linguist’s intuitions, as a native speaker of the language under description, about what constitutes a linguistically significant prosodic difference, and about what difference in meaning is to be assigned to that phonetic contrast (e.g., Brazil (1975,1978), O’Connor and Arnold (1973)). The shortcomings of an intuition-based approach to prosodic description are brought to the fore when the linguist has to describe a language or language variety other than his own. The very fact that there is a descriptive problem serves to emphasise how far, theoretically and methodologically, the study of ‘intonation’ lags behind the more traditional areas of linguistic description lexical phonology, morphology and syntax. It is perhaps for this reason that * Author’s address: J.K. Local, Department YO 1 5DD, United Kingdom. 0378-2166/85/$3.30 of Language, 0 1985, Elsevier Science Publishers University of York, Heslington, B.V. (North-Holland) York 310 J. K. Local, et al. / Phonology for conversation ‘intonational’ features of creole varieties have received relatively little scholarly attention and that where they have been studied the approach has been a comparative one, using categories that have been previously established for the input languages, rather than a descriptive one which sets up categories for the creole as an autonomous linguistic system (cf. Carter (1979,1982)). In this article we address the theoretical and methodological problems of prosodic description through a study of such a variety - London Jamaican (LJ). ’ LJ poses descriptive problems which bring into sharp focus the more general theoretical question we have raised. It is a language identifiably similar to London English in terms of its lexicon and some aspects of its syntax; at the same time it can be regarded as a local variety of Jamaican Creole and can be heard to differ from Standard English in at least some respects at all linguistic levels. 2. Categories in intonational phonology Most students of English intonation have wanted to recognise the ‘tonic’ (nucleus, nuclear tone) as the key structural unit in their description. * Crystal (1969:209) presents the received view: “Every tone unit contains one and only one nucleus, or peak of prominence, a finite number of contrasting pitch glides or sustentions on the accentual prominent word.” expounded by one of syllable of the most A dual function has been attributed to the tonic, though not always explicitly. Firstly, as the quotation above implies, the tonic serves as one of the delimitative signals for the tone unit, in as much as there can only be one tonic per tone unit. If, as Crystal and other analysts suggest, nuclear tones are phonetically distinct from non-nuclear glides (1969:221), then the occurrence of a nuclear tone serves to indicate to the hearer that he is ‘in’ a new tone unit. Secondly, the tonic has been equated with information focus, particularly by Halliday (1967:203): “The system of information focus specifies the structure number and location of the tonic components. Each point tonic component.” of the tone group, determining of information focus is realised This coincidence of functions has led to some analytical confusion English intonation, and in particular to a tendency to regard the as a in studies of phonological ’ This paper derives from research on London Jamaican in progress since 1981. The project was funded by the Nuffield Foundation and ESRC. Our thanks also to David Sutcliffe and Barbara Stanbury, and to our informants. ’ But see Brown et al (1980) for a dissenting view. J.K. Local et al. / Phonology for conversation 311 units (i.e., the tonic) which have been established in the first instance by criteria of similarity and difference, as isomorphic with semantic functions (i.e., focus) in the subsequent description. However, it has been shown that the phonetic cues which hearers respond to when identifying a focussed constituent are not restricted solely to pitch (Wells (1985)). As for the delimitative function, it has been more generally accepted that phonetic features other than pitch - rhythm, tempo, pause, phonation type - may also serve to mark the boundary of a tone unit. The need for a clear distinction between functional and phonetic criteria in phonological description is most pressing in so-called ‘intonational’ phonology. We address this issue here through an analysis of London Jamaican, 3 two aspects of which are of particular interest. Firstly, LJ does not seem to utilize a combination of phonetic events that we, as English speakers, respond to as a ‘tonic’. Secondly, the functions of focus and delimination are not jointly manifested. phonetic 3. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Pitch and focus in LJ Our first observation is that dynamic pitch movement is not an exponent of information focus in LJ but simply serves to mark the end of a turn at talk. This is shown by extracts (1) and (2). In (l), the speakers are discussing the death of a local black youth. The point of interest is the reiterated pronominal ‘him’. It is clear from the context that ‘him’ has the same reference throughout the extract. In many varieties of the lexifier, pronominals are not accented in such contexts; in particular, they are ‘deaccented’ when sentence-final, that is when in the unmarked position for sentence accent, since noticeable ‘stress’ on pronominals (possible exponents of which are major pitch movement, increased loudness, rhythmical prominence) forces the interpretation that the pronominal is not coreferential with its conventional antecedent. In (l), 3 London Jamaican, the language on which this study is based, is a language spoken by young Afro-Caribbeans born and living in London. The recordings which provide the data for this paper were made by Sebba at a senior high school in the East London Borough of Waltham Forest. They took place between two sixth-form girls (Corpus G) and two fifth-form boys (Corpus B), and were produced at the researcher’s request: in each case he asked the participants to ‘talk Jamaican’ for ten minutes or so. The conversations are therefore to a greater or lesser extent a performance, but nevertheless seem true to the character of LJ as recorded on other occasions and in other schools. The transcription used in the LJ examples makes no claim to represent segmental phonological systems. It is designed to be easily readable, whilst being constrained by the need to show the correct number of syllables (for pitch analysis). Where reference is made to phonetic features of a portion of an extract, that portion is given in impressionistic transcription. The data was transcribed impressionistically in detail by the three authors independently, discrepancies being resolved by discussion. This impressionistic transcription provides the basis for the phonological analysis. A list of transcription conventions is given in the Appendix. 312 J.K. Local, et al. / Phonology for conversation (1) [f rail. “V”” “” IL ” yeah them thump him them thump him good and proper you know (0.5) \ - --_ ““” F: [XCd. fl zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLK [mfll Lf -a” -” F: f] __-___-, “” [mfl “” nothing more them thump him bad = -- - - - -\ [ CIXSC.] ” ” ” “ .” ” I: = them thump him and kill him F: true (0.5) \ however, whilst ‘him’ clearly has the same reference throughout, and is thus conventional (rather than contrastive or deictic, for instance), it carries the major pitch movement of the sentence when sentence-final. In Z’s turn, the pitch on ‘him’ falls to the bottom of the speaker’s usual range; there is an increase in loudness towards the end of the sentence that is sustained through ‘him’; and ‘him’ is not rhythmically weaker than the preceding syllables indeed the whole utterance gives an impression of syllable-timing. These three features give an impression of saliency to the syllable displaying them; furthermore, the first two are noticeably absent from the earlier occurrence of ‘him’ in the same utterance. In zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDC F’s turn, the first ‘him’ also occurs at a sentence boundary - a major syntactic completion point which is potentially a place for turn transition. It too is as loud as the preceding word, and has on-syllable pitch movement, though not to the bottom of the range. The two remaining instances of ‘him’ in F’s turn do not display these features. This fragment thus suggests that dynamic pitch is associated with turn-ending: Z’s second ‘him’ is followed by a turn-change, unlike the other four ‘ him’s. F’s first ‘him’ has some, but not all, of the features isolated at Z’s second ‘him’. This may well have to do with the fact that although F’s first ‘him’ occurs at a major syntactic boundary it is not followed by a change of speaker. The obvious conclusion is that on-syllable pitch movement is not constrained by considerations of information structure and anaphoric reference that operate in other varieties of English. Further evidence of this is provided by extract (2) where once again a word is repeated (‘law’), in a discussion of police behaviour. J.K. Local et al. / Phonology for conversation 313 (2) [fl lfl I: because them is law they reckon (0.2) know something (0.4) _- m-y - -- _ [fl I: all them p'licemen break the law you know (1.2) \- - - __- _ \ [fl I: / - zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA [f l you call-them law but they break the law (0.4) - - - _----x F: Rules of information structure for Standard English require that once a lexical item has occurred, in its subsequent appearances in the discourse as a ‘given’ item it will not receive the sentence accent (Halliday (1967)). This is the classic motivation for de-accenting, whereby the main accent is shifted leftwards off the last lexical item of the sentence (Ladd (1980) Wells and Local (1983)). In (2), however, it is the fourth occurrence of ‘law’ which displays the biggest pitch movement of the entire turn - and it is followed by a change of speaker. Unlike this fourth occurrence of ‘law’, the second and third instances of the word, also informationally ‘given’, do not display any pitch movement and are not loud. If pitch movement were an exponent of new information focus in LJ, as it is in many varieties of English, we might expect the first occurrence of ‘law’ to display a pitch glide; but although it is relatively loud and long, on-syllable pitch movement is very slight and hard to distinguish from the intrinsic pitch change of the diphthongal glide. Extract (3) provides further evidence that in LJ, dynamic pitch is associated with delimitation but not with information focus. Lexically and syntactically, ‘stylers’ and ‘slickers’ (two groups of young blacks) represent a clear case of contrastive focus, in terms of the information structure of the utterance, yet the prosodic marking of this contrastive function is strikingly different from its marking in those varieties of the lexifier that have been investigated to date. In many varieties of English, contrast of this kind is associated with maximal pitch movement (e.g., fall from top to base of pitch range: cf. Wells (1985)): here there is no movement at all on ‘slickers’ and barely any on ‘stylers’. Instead, the largest pitch movements occur on ‘know’ - a sentence boundary J. K. Local, et al. / Phonology zyxwvutsrqponmlkjihgfedcbaZYXWVUTSR for conoersatton 314 (3) I B: not s:tylers accel. go there s:o -_ you ] zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJI know -- B: A: - / -- \ - \ / and therefore a potential turn transition point - and on the final syllable of ‘wicked’, after which the turn does actually change. As for the contrasted items, they occur very high in the pitch range, but on a relatively level pitch. Other noteworthy features of these items are the long initial consonants and the following acceleration in tempo: the exponents of focus and contrast will not, however, be further discussed in the present article, which is concerned with delimitation. On the basis of data such as this, we adopt the working hypothesis that in LJ the system of delimitation and the system of information focus are distinct. LJ would therefore differ from other varieties of English spoken within the same geographical area and possibly even by the same speakers. This might be responsible for the strong impression of English speakers that LJ is distinctly non-English prosodically, and at times difficult to understand, in spite of a good deal of shared lexis and syntax. This hypothesis therefore seemed to be worthy of further investigation, not only for its intrinsic linguistic interest as a contribution to the study of accentual systems, but also for its possible social implications. The findings we present below are restricted to declarative sentences. 4. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Turn-delimitation and phonology Since Trubetzkoy’s programmatic remarks (1969, Part II) the phonology of delimitative systems has remained a relatively neglected area of linguistic research, particularly with regard to pieces longer than the word or phrase. Some of the problems for the analysis of conversational data that result from this neglect are mentioned in Brown, Currie and Kenworthy (1980). Trubetzkoy’s view is illustrated by the celebrated comparison of phonological J.K. Local et al. / Phonology forconversation 315 delimitation to traffic lights, which are not necessary at every junction: “(. . . ) each language possesses specific, phonological means that signal the presence or absence of a sentence, word or morpheme boundary at a specific point in the sound continuum. But these means are only ancillary devices” (our emphasis). This position has recently been adopted, without apparent modification, by Brown et al. in their treatment of phonological delimitation in Edinburgh English (1980:158). The idea that delimitative signals have an ancillary status presumably arises from the assumption that their function is to delimit syntactically defined units, such as the sentence. It is clearly the case that not every sentence, as defined by its syntactic structure, has its boundary signalled phonetically, but some do. Instead of therefore concluding that there is a system of sentence delimitation, which is optional, we propose that the domain of those delimitative features sometimes associated with the sentence is not in fact the sentence itself but a higher structural unit - the turn - and that the sentence will be delimited just in the case where it is coextensive with a turn. The ‘optional’ nature of sentence delimitation will then be accounted for by the facts that one sentence may constitute a turn, but a turn may consist of an indefinite sequence of sentences. By taking the turn - a category to be defined interactively - as the domain over which to examine delimitation, we aim to free our analysis from the problem which besets virtually all intonational studies that we know of: reliance on intuition. To achieve this end, we base our analysis on detailed phonetic observation, in conjunction with techniques associated with conversation analysis. We take it as axiomatic that it is not sufficient to rely on the intuitions of the native speaker as to which intonational distinctions are meaningful and which are not, and then to use these judgements, in conjunction with distributional regularities in the phonic data, as the basis for establishing categories. Analytic categories should be established not only on distributional grounds; they should also be shown to have relevance for the conversational participants. It is only by attempting systematically to ground analytic claims in the observable orientations of participants, that one can place constraints upon theorizing about functional categories and their exponents. By making use of speaker behaviour as an analytical resource we have access to a non-intuitive warrant for the functional category of delimitation. We shall now use this resource in the phonological analysis of turn-delimitation in LJ. The structural unit turn is identified in the first instance as a spate of talk by one speaker followed by a change of speaker in the clear (i.e., not in overlap). 4 We take as primary evidence for the delimitative function, points in the talk where the listener/coparticipant treats the speaker’s turn as complete, and demonstrates that he is doing so. As primary evidence for an instance of 4 See Goodwin (1981) on the problem of defining ‘turn’. J. K. Local, et al. / PhonologV for conversatron 316 (4) Imfl 1: it true that them p'licemen [mfl them fresh you know (') __----- F: \ yeah turn delimitation, we use conversational criteria: at a completion point in first speaker’s turn. 5. Turn-delimitation second speaker starts to talk in LJ 5.1. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA You know (I) We begin our investigation of delimitation by considering turns which end with the tag you zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA know. There are three main reasons for this: (i) (ii) (iii) Turns ending in you know are extremely common in the data. The differences in the realisations of the tag between LJ and British English (BE) highlight the fact that LJ, whilst having a good deal in common with BE, differs markedly in its use of pitch and other prosodic features. The tag could thus provide a precise focus for the investigation of systemic mismatch. Within LJ itself, the tag takes different phonetic shapes, whilst retaining a degree of phonetic identity. By choosing to concentrate on this tag, as it occurs at turn-endings, we are able to highlight more clearly the phonetic features relevant to turn delimitation. (This is not the only possible procedure for the investigation of turn-delimitation. In a similar study (Local, Kelly and Wells (forthcoming)), we looked at turn-endings more generally, at the first pass; this does, however, present additional problems of comparing like items with like.) Turns ending in you know form two distinct analytic groups, henceforth Group I and Group II. Group I comprises the bulk of the instances of you zyxwvutsrqponmlkjihgfedcbaZYX know in the boys’ conversation and a few of those in the girls’ conversation. The phonetic characteristics of Group I are: (a) the first syllable (you) regularly occurs on the same pitch as the syllable(s) preceding, or at a lower pitch, and is rhythmically integrated with them. The two syllables of the tag are rhythmically 317 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPO J.K. Local et al. / Phonology for conversation short, and, insofar as the notion of rhythmical feet (Abercrombie (1964)) can be evoked impressionistically in the description of this non-standard variety, our impression is that the tag in neither case constitutes a separate foot, but is linked to the preceding syllables, narrow falling pitch movement to the bottom of the speaker’s pitch range on the second syllable (know), with accompanying creaky phonation, the starting point of this pitch movement is never higher than the preceding syllable: it may be below or at the same level, absence of decrescendo on the second syllable, in spite of frequent decrescendo over the preceding portion of utterance, leading to an impression of a resurgence of loudness, and absence of greater dynamic pitch movement earlier in the utterance. (b) (c) (d) (e) Extract (4) exemplifies these features: there is a narrow fall to the base of the speaker’s normal range on ‘know’, which starts at the same height as ‘you’ and is the final word of Z’s turn; there is no greater on-syllable pitch movement earlier in the turn; and there is no drop in loudness over the final part of the turn. Z’s turn is followed by a change of speaker, and the fact that Z’s turn is (5) zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA [f f l ” I: __: ” them a:lways ” ” ” ” come and stop me -- -- - _---_ [fl F: true enough (6) [fl [f l : li,” --$” me used see him [fl -: F: yeah - - - you know - \ ” (0.5) ” ” for nothing $ ” you know CO.41 318 (7) J. K. Local, et al. / Phonology for conoersatron zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA If1 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA If1 ” I: ” -“” them(.) all - -_: p'licemen - - -+ them s:oft you know= 7- - - TP \ PI ““““” F: = yeah them soft you know - - - - \ designed and treated as complete is further attested by the pause that intervenes between the two turns: I does not continue talking after ‘you know’. which indicates that the features displayed at the end of his turn are not being used to project further talk and thus hold the turn for the current speaker. The same sequence can be observed in extracts (5) and (6). Extracts (4) (5), and (6) represent clear instances where turn transition is achieved without problem: for the participants and where the phonetic features provisionally associated with turn delimitation are displayed. In extract (7) we find the same phonetic (8) [fl If1 _“” F: some ” I: “v-” of them p'lice -- [fl [fl - [fl -w is the biggest -N_---- IL -: thief agoin: --y ” you know _ -1 - me know = -\ [fl [fl -““-” F: - them is the biggest thief -_--- y- If1 F: [fl “WV_:: and them a try stop it = them is a ras WV_ -- [fl " IL thief you know _ I: IfI “_---- - 7 = yunu could -- - just walk down a street _ - -- - so . . . - - --. J. K. Local et al. / Phonology 319 zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPO for conversation zyxwvutsrqponmlkjihgfedcbaZY features. In I’s turn, ‘know’ has the greatest on-syllable pitch movement; it starts at the same height as ‘you’ and falls to the base of the speaker’s normal range. As in (4) to (6), there is then a change of speaker, but this time with no intervening pause. In fact, the two turns are ‘latched’: they could not be closer together without being in overlap. Yet there is no indication that the transition is in any way problematic for the participants: I does not attempt to come in again while F’s turn is in progress. This suggests that in (4) to (6) it was not simply the pause that indicated to the second speaker that the first speaker had completed his turn, but that features (a)-(d) above are involved in projecting turn completion. Further evidence is provided by the changes of turn in extract (8). In both F’s turns, ‘know’ carries the greatest on-syllable pitch movement in the utterance; this begins at the level of the preceding ‘you’ and reaches the bottom of the speaker’s normal range. In the second turn, there is a clear resurgence of loudness on ‘know’. Following both turns there is a latched transition, yet in neither case does the transition cause any problems for the participants, as can be seen from the fact that F does not attempt to continue his talk during I’s turns. In the absence of even the slightest intervening pause, change of turn is effected smoothly, indicating that the features associated with ‘you know’ are projecting turn completion. The transition from F’s first turn to I’s second turn in extract (9) displays the same features, once again demonstrating the precision with which speakers orient to the projected him can't hold me you know (0.4) him can't hold me : ‘;phhdow* _ - you I: know = = s:lap up him: b: umberklaat one time you know (0.8) Lure -- I: -a- - , \-- \ --x \ h 1 mhm 320 J. K. Lad, et al. / Phonology for conoersarion delimitation of the turn: ‘know’ begins at a lower pitch than the preceding syllable, falls to the base of the speaker’s range, and there is no earlier on-syllable pitch movement in the utterance; Z’s incoming, ‘slap’, is latched onto ‘you know’, and there is no indication from zyxwvutsrqponmlkjihgfedcbaZYXWVU F that this is unexpected or unwanted. The first transition in (9) presents a rather more complex situation. Once again, the delimitative features are present at the end of I’s turn, which is treated as complete by F, who comes in after a brief pause. However, I resumes talk almost simultaneously with F, only to drop out and relinquish the floor to F. At first glance, this may appear to contradict the claim that features (a)-(d) project turn-completion, since I does not relinquish the turn after displaying the features, even though F starts to talk. Closer examination suggests, however, that Z’s talk in overlap with F’s turn (‘him.can’t hold me’) does not constitute a claim to hold the floor, but is presented as a footnote to his original talk. Firstly, it is lexically and syntactically an exact repetition of the original turn, and as such adds no new information at all to the discourse. Secondly, it seems to be subordinated prosodically to what precedes: the pitch range of the repeated phrase is markedly lower overall than that of the preceding talk, and it is quieter. Thirdly, these prosodic features do not constitute I’s talk as turn-competitive, i.e., as an attempt to regain the floor, if the interruptive strategies of LJ are at all comparable to those of Standard English as described by French and Local (1983), who identify high pitch and extra loudness as markers of turn-competitive incomings. Fourthly, F does not them thump him and / everything 1: you know ifI I: two (0-J me Idrin dem awatching -- ‘___L_ it you know (.) __ \ [ F: accel. yeah they thump him them thump him good an' proper '--, _ 1 you know _---_-_ \ J.K. Local et al. / Phonology for conversation 321 treat Z’s talk as an attempt to hold the floor by speaking slowly and loudly in order to fend off the competition (French and Local (1983:26)). Rather, F seems to proceed normally, as if I’s talk did not present any problem for the changeover of turn that has just taken place. Fifthly, the brief pause that precedes F’s talk and the indistinctness of his first syllable (transcribed ‘em’) suggests that I’s resumption of talk may be a response to F’s failure to take up the offer of a turn as quickly and as clearly as he might. In extract (10) we find a different type of overlapping talk, and this too points to the precise orientation of participants to the turn-delimitative features. I comes in during the course of F’s first turn, but breaks off (at a syntactically incomplete point), only resuming after F’s ‘you know’. This ‘you know’ has the features identified with delimitation (fall to base of range, not starting higher than the preceding syllable; no bigger pitch glide in the utterance; lack of diminuendo), and is followed by a no-pause transition, as in extracts (7) and (8). The fact that I breaks off his original talk without even completing it syntactically, in conjunction with the fact that when he does resume, he repeats and continues his original phrase, indicates that he treats his original talk as lost, and thus his original incoming as illegitimate, i.e., placed at a point where change of turn is dispreferred (cf. Jefferson and Schegloff (1975)). Extracts (4)-(10) are representative of coparticipants’ treatment of Group I you know in our data. Apart from instances such as these, where second speaker displays through his talk that he is treating first speaker’s turn as complete, there are occasions where current speaker uses this type of you know but it is not taken up by the coparticipant; instead, current speaker continues talking, generally after a pause. Such instances are null evidence for the hypothesis that Group I you know is turn-delimitative, conforming to Rule (c) of the turn-taking rules proposed by Sacks et al (1974): ‘If the turn-so-far is so constructed as not to involve the use of a ‘current speaker selects next’ technique, then current speaker may, but need not continue, unless another self-selects.” 5.2. You know (2) To this point we have not shown that it is the phonetic parameters rather than the tag you know which constitute turns as complete. Clear evidence that phonetic features are constitutive of turn-endings is provided by an examination of the second group of you know turns. These (Group II) occur almost exclusively in the girls’ conversation. The phonetic characteristics of Group II are: Either step up in pitch from the first syllable (you) to the second (know), with pitch on the second level or level plus slight rise; Or step up to the second syllable and rising pitch on second syllable. syllable Whereas Group I you know never elicits a response in ouerfup, in Group II the majority do. In extract (ll), B has been talking about a male friend whom she 322 J.K. Local, Ed 01. / Phonology for conuersaclon (11) [ B: accel. ] tell him to coo:1 himse:lf: you know ---\,- - (.) hh A: - A: (.) ?ah (2.5) \ zyxwvutsrqponmlkjihgfedcbaZYXW v' vou seen Benita -\_ talked to at a recent party, and concludes her account of the meeting in the way shown in this extract. The fall on ‘self’ is the biggest on-syllable pitch movement in B’s utterance, and it is the only syllable to reach the base of the speaker’s usual range, having started at the same level as the preceding syllable. These features, which occur on the word immediately preceding the you know tag, were associated with the final word of the tag itself in Group I fragments. Conversely, the tag itself is quite different from Group I you know: there is a step up in pitch, and a level pitch on ‘know’ rather than a fall to the base of the speaker’s range. Furthermore, B’s turn elicits a different type of response, namely a supportive token which is uttered by A in ooerhp with the tag. This overlapping response does not, however, cause any apparent problems for B: she does not try to regain the turn, but allows A to pause and then change the topic. There is therefore no evidence that B construes A’s response as competitive or that B has not designed her own turn as complete. Some of the same features can be noted in extract (12), where B is talking about an upcoming party. Again, B reaches the base of her range on the last word of her turn (‘sure’) prior to the tag, and there is no greater on-syllable pitch movement earlier in the turn. The tag itself has level syllables stepping up. As in (11). the tag is overlapped by A’s response, which is not treated as competitive: after the overlapping talk, there is a substantial pause before B continues her talk, which can be taken as evidence that she does not feel constrained to reclaim her turn. The non-problematic nature of talk overlapping with the tag is also evident in (13). At the end of her first turn, B reaches the base of her range with ‘time’, the only syllable having a dynamic fall. The tag, preceded by a micropause, is reduced to one syllable, mid-level, which is overlapped by A’s talk. B does not attempt to regain the floor, allowing further talk in the clear from A and a substantial pause. J.K. Lmal et al. / Phonology for conversation (1-4 B: [ this either Saturday or Sunday I'm not quite know/(l.O) hm but should be qood hhh I hope so - A: 1 accel. sureryou A: B: 323 - -\ (1.2) --_ me I've nowhere to go . . . . In (ll)-(13) we can see that the participants in the talk treat Group II you know as a quite distinct interactional event from Group I you know as illustrated in (4)-(10). Thus, Group II you know, whilst differing somewhat (13) A: v' you seen Benita B: B'nita come mu':: --- last night .\ see her phone her all the time : B: (0.2) Ieah tch (.I dread \ (.) tch y’know (0.2) 324 J. K. Local, et al. / Phonology for conversation from standard varieties of English in the phonetic detail of its exponents, can be identified functionally with the use of this tag in standard varieties, as characterised by Sacks et al. (1974:707, 708, 718): by employing a tag such as you know, speaker transforms his turn into a locus of ‘current speaker selects next’ zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA after an initial transition relevance place (i.e., the syntactic completion point immediately preceding the tag). Significantly, the phonetic features associated with the talk immediately preceding you know in Group II, after which overlapping talk begins, are the same as the feature (a)-(d) which we identified earlier as signalling turn delimitation in Group I. Thus, in addition to signalling to hearer that the turn is over (by delimitative signals at the first completion point), speaker adds a tag which directly selects a respondent, thus requiring response. As might be expected, such tags are often overlapped by incoming talk (cf. Sacks et al. (1974)). With Group I you know, on the other hand, we cannot identify two interactive components in the same way. Instead, the next-selecting potential of you know is integrated into the delimitation of the turn: the option is not available to the speaker to manipulate these two interactive components separately. The consequences this has for the management of talk, and particularly of talk between speakers using the Group I system and speakers using a ‘standard’, two-component system such as Group II or the variety described by Sacks et al., could be such as to constitute a potential source of misunderstanding. ’ 6. The phonetics of turn-delimitation The hypothesis that turn-delimitation is associated with the phonetic features isolated for Group I you know, is supported by a further, quite general property of the data: the features are also associated with turn-final items other than you know, in the speech of all four informants. For instance, in Z’s turn in extract (14), the only on-syllable pitch movement is the narrow fall on the final word, ‘time’, which starts at the level of the preceding word, ‘one’, and reaches the base of the speaker’s range, without trailing off in loudness. F’s response is latched onto Z’s turn, which indicates that F orients finely to these phonetic features as marking the end of the turn. Similarly, in extract (15) F’s turn displays the delimitation features. There is no on-syllable pitch movement apart from the narrow fall on ‘bad’, starting at the same level as the preceding syllable and reaching the base of the range; an immediate change of speaker 5 Misunderstandings would presumably be most likely in a situation where neither participant in the interaction has the other’s system in his competence, and less likely where the participants have both system available, as seems to be the case with the girls represented in our data. Sociolinguistic implications of the prosodic systems we have identified will form the subject of a separate study. 325 zyxwvutsrqponmlkjihgfedcbaZYXWVUTS J.K. Local et al. / Phonology for conversation (14) I: (s'mone should - F: = mhm come blow'p - - (laugh) Leyton I: - P'lice -- - a) p'lice -- station one ti:me: = - - station - _ / mm 1 follows. Another instance was discussed earlier: in I’s turn in extract (1) the final occurrence of ‘law’ is contextually given; nevertheless it carries a falling pitch movement to the base of the range, is relatively loud, and is followed by a change of turn. The only earlier pitch fall that reaches the base of the range is on ‘you know’, which also elicits a response. I’s first turn ‘me know’ in (8) also displays features (a)-(d), and is followed by a change of speaker; indeed, the association is overwhelmingly characteristic of the boys’ talk. It is regularly present in the girls’ conversation too. In extract (16), there is a narrow fall to the base of the range on the final syllable of B’s turn. This fall starts below the level of the preceding syllable; there is no earlier on-syllable pitch movement and no diminuendo. The change of speaker indicates that A orients to B'sturn as having been completed. In extract (3) above, the features [mfl F: nothing more them thump him bad = --c- -\ (crest. I: ] = them thump him and kill him - F: zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONML true - -\ (0.5) 326 J. K. Local, et al. / Phonologv for conversation (16) B: on Saturday A: yeah night me see the one Winston (0.2) / B: yeah me see him Saturday \ _- _ __- night .. .. \ are again observable, on B’s final phrase ‘it was wicked’. These two examples are of special interest since they exhibit a common phenomenon in our LJ data: turn-final polysyllabic words that in standard English are conventionally assigned non-final stress (Winston, wicked) we perceive as being ‘stressed’ on the final syllable. This suggests that the delimitative system identified here for turns can ‘override’ lexical stress assignment rules - a phenomenon which would repay further investigation, since it implies a further quite fundamental phonological difference between LJ and other varieties of English. Further evidence for the claim that the phonetic features we have isolated are exponents of the turn delimitative function is furnished by the recurrent failure of next speaker to come in at points in current speaker’s turn where the syntax is complete but which do not display features (a)-(d). Notably, in Group I utterances with you know, such as (4)-(10) above, second speaker never comes in immediately prior to the tag, although syntactically a potential sentence termination has been reached. In Group I, features (a)-(d) are never associated with this point, whereas in Group II, where these features are present immediately before the tag, next speaker frequently starts to talk in overlap with the tag, as we have seen. Extract (8) presents striking exemplification of this ‘disjunction’ between phonetics and syntax. In F’s second turn there is a major syntactic break - an unambiguous sentence boundary between ‘stop it’ and ‘them is’. Since the sentence is the turn-constructional unit par excellence at the syntactic level of description, one might anticipate a change of speaker following ‘stop it’, but this does not happen: in fact, F latches his two sentences together. Furthermore, the delimitative features are not in evidence at ‘stop it’: the final syllable is not at the base of the speaker’s range, nor does it have falling pitch, yet there is earlier on-syllable pitch movement, on ‘thief’. In spite of its syntactic completeness, the first part of 327 J. K. Local et al. / Phonology for conversation F’s utterance is not marked prosodically as a turn-constructional not treated as a complete turn by the coparticipant in the talk. unit and is 7. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Phonology for conversation? We have shown that turn delimination in LJ is realised by particular pitch characteristics of the final syllable of a syntactically complete piece, in conjunction with other phonetic features of the piece involving loudness as well as pitch. The findings indicate that LJ differs from its principal lexifier, English, in which the occurrence of dynamic pitch at the end of a syntactic piece is constrained by considerations of information focussing. The claims we make have been substantiated by examination of the interactive and phonetic characteristics of two types of you know identified in the data, then of other turn-final pieces. By using such evidence, the analyst can show which phonetic features are functionally relevant for the participants in the conversation, and thus provide a warrant for categories used. This approach draws upon procedures developed within the ethnomethodological discipline of Conversation Analysis (C.A.). From the CA. standpoint, the present paper can be regarded as a first attempt to accede to the request made by Sacks et al. to linguists to investigate ‘how projection of unit types is accomplished’ (1974: p. 703, fn. 12): by using linguistic techniques, the analyst can set up systems of turn-completion vs. non-completion for various syntactic types where turn-transition is possible. The present analysis might thus be considered a tentative step towards a ‘phonology-for-conversation’ on the lines of the ‘syntax-for-conversation’ discussed by Schegloff (1979). From the standpoint of phonology, the analysis represents an attempt to appropriate some of the methodological procedures of C.A. in the hope that they will eventually permit a more rigorous formulation of functional categories than has been achieved hitherto in the field of ‘intonation’. The approach to phonology taken here is essentially an extension, to longer pieces of discourse, of the type of analysis developed to handle junctural phenomena within Firthian prosodic analysis (e.g., Sprigg (1957)). Specifically, the present authors share with prosodic analysts the view that it is inappropriate to impose a phonological system set up for one part of the language onto another part. A coherent theory of delimitation, and thence of discourse, requires that each delimitative function - of words, of longer syntactic pieces, or of discourse pieces - be analysed on its own terms, on the basis of non-intuitive evidence of the type used here, and without prejudice as to the phonic features involved. Only after such an analysis has been made does it become meaningful to make statements about phonological systems. Our analysis suggests that London Jamaican differs from other varieties of English in two important ways: the domain over which the bundle of delimitative features operates, and the role played by dynamic pitch as an exponent of J.K. Local, et al. / zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLK Phonology for conversation zyxwvutsrqponmlkjihgfedcbaZYXWVU 328 delimination. Recent work on localised Tyneside English confirms this (Local, Kelly and Wells (forthcoming)). Varieties of English such as Tyneside, Edinburgh (Brown et al. (1980)), and RP differ from each other in the phonetic exponents of turn delimitation but not in the phonological domain of those exponents: the final zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFED foot of the piece. London Jamaican, by contrast, differs from these varieties in both respects: the phonetic features associated with turn-delimitation are again different, but more importantly they are centred on the final syllable of the turn, for it is here that the major pitch movement is invariably located. It can thus be seen that at the phonological level of analysis LJ is typologically different from those varieties of English from which its lexicon is largely derived and with which it is in continued contact. zyxwvutsrqponmlkjihg Appe ndix Transcription Conventions Pitch marked by syllable, between staves representing the normal limits of the speaker’s range, below the orthographic transcription. Other prosodic features Rhythm marked by syllable, above the orthographic v = short; -: = extra long; $ = extra short. The domain orthographic Loudness Tempo are indicated of other prosodic transcription: only when discussed features is indicated in text. transcription: by square brackets, - = long; above the absence of symbols signifies normal loudness level for speaker; louder or quieter portions by mj, j, jj, p, pp (mezzo-forte, forte, fortissimo, piano, pianissimo); gradual increase by crest, decrease by dim. acceleration is indicated by accel, slowing by rail. Unusual segmental length and aspiration are marked in the orthographic transcription by IPA symbols. The remaining notations used derive from the transcription system developed by Dr Gail Jefferson: (0.4) (.) = [ / pause within or between turns, given in tenths of a second. pause of one tenth of a second or less. marked at the end of one speaker’s turn and at the beginning of the next indicates that the final segment of the first turn and the initial component of the second are almost, but not quite, simultaneous. square brackets represent the point at which simultaneous speech begins. obliques indicate the point at which simultaneous speech ends. J. K. L.ocal et al. / Phonology for conversation (aidrin im awatching) (**) hhh 329 parentheses indicate that the transcription of the enclosed portion is open to some doubt. Where asterisks appear in parentheses, we have been unable to assign any representation to the enclosed portion. The number of asterisks corresponds to the number of syllables we hear the speaker as having produced. audible outbreaths or breathiness in words. The number of ‘h’s’ corresponds to the duration of the outbreath measured in tenths of a second. zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA References Abercrombie, D., 1964. ‘Syllable quantity and enclitics in English’. In: D. Abercrombie et al., eds., In honour of Daniel Jones. London: Longman. pp. 216-222. Brazil, D., 1975. Discourse intonation. Discourse Analysis Monographs 1. Birmingham: English Language Research. Brazil, D., 1978. Discourse intonation II. Discourse Analysis Monographs 2. Birmingham: English Language Research. Browin, G., K.L. Currie and J. Kenworthy, 1980. Questions of intonation. London: Croom Helm. Carter, H., 1979. Evidence for the survival of African prosodies in Caribbean Creoles. Society for Caribbean Linguistics, Occasional Paper 13. School of Education, University of the West Indies, St. Augustine, Trinidad. Carter, H., 1982. The tonal system of Jamaican English. Paper presented at the Fourth Biennial Conference of the Society for Caribbean Linguistics, Paramaribo, Suriname, September 1982. Crystal, D., 1969. Prosodic systems and intonation in English. London: Cambridge University Press. French, P. and J.K. Local, 1983. Turn-competitive incomings. Journal of Pragmatics 7: 17-38. Goodwin, C., 1981. Conversational organisation: Interaction between speakers and hearers. New York: Academic Press. Halliday, M.A.K., 1967. Notes on transitivity and theme in English, Part 2. Journal of Linguistics 3: 199-244. Heritage, J., 1984. Recent developments in conversation analysis. Warwick Working Papers in Sociology 1. University of Warwick. Jefferson, G. and E.A. Schegloff, 1975. Sketch: Some orderly aspects of overlap in natural conversation. Paper delivered to the 74th Annual Meeting of the American Anthropological Association, San Francisco. Ladd, D.R., 1980. The structure of intonational meaning. Bloomington, IN: Indiana University Press. Levinson, S., 1983. Pragmatics. London: Cambridge University Press. Local, J.K., J. Kelly and W.H.G. Wells, (n.d.) ‘Some phonetic aspects of turn-delimitation in the speech of urban Tynesiders’. Journal of Linguistics. (Forthcoming.) O’Connor, D.J. and G.F. Arnold, 1973. Intonation of colloquial English. London: Longman. Sacks, H., E.,A. Schegloff and G. Jefferson, 1974. A simplest systematics for the organization of turn-taking for conversation. Language 50: 696-735. Schegloff, E.A., 1979. ‘The relevance of repair to syntax-for-conversation’. In: T. Given, ed., Syntax and semantics 12: Discours and syntax. New York: Academic Press. pp. 261-286. Sprigg, R.K., 1957. Junction in Spoken Burmese. Studies in Linguistic Analysis. Special Publication of the Philological Society, Oxford. Trubetzkoy, N., 1969. Principles of phonology. (Trans. C.A.M. Baltaxe). Berkeley, CA: University of California Press. [1939] 330 J. K. Local, et al. / Phonology for conuersatlon Wells, W.H.G., 1985. ‘An experimental approach to the interpretation of focus in spoken English’. In: C. Johns-Lewis, ed., Studies in intonation and discourse. London: Croom Helm. Wells, W.H.G. and J.K. Local, 1983. Deaccenting and the structure of English intonation. Linguistics 21: 701-715.