17.1 Introduction
This chapter focuses on the analysis and description of intonational patterns in different sets of twentieth-century recordings in two Caribbean English-lexicon creoles, Jamaican and Trinidadian. These historically related restructured varieties have both been shaped by processes of language creation (Winford 2001; Schneider Reference Schneider2008) and are the result of European colonial expansion and West African contact in the Caribbean in the seventeenth century (Lalla and D'Costa Reference Lalla and D'Costa1990). However, the individual development of the English varieties and the related creoles in Jamaica and Trinidad is different. Differences in the particular mechanics of the language-creation process and in the local ecologies thereafter also translate to different outcomes, as the newly formed languages would not have remained static (Plag and Schramm Reference Plag, Schramm, Bhatt and Plag2006; Smith Reference Smith, Kouwenberg and Singler2008). During the forty-year period between 1680 and 1720, Lalla and D'Costa suggest that Caribbean creoles developed their distinctiveness, and the roughly 200 years of West African slave presence left its mark on the grammar and the phonology (Alleyne Reference Alleyne1980). More precisely, British presence in the Caribbean came more than 100 years after the Spanish, with the earliest being in St. Kitts in 1624. Jamaica became British in 1655 and remained under British rule until 1 August 1838. In Trinidad, the Spanish ruled from 1498 until the British came in 1797, but after 1783, French and French creole-speaking migrants from other Caribbean territories developed an indigenous French creole, which was the lingua franca on the island for several generations before gradually being supplanted by the newly developed English and associated English creole (Winford Reference Winford1997; Ferreira and Holbrook Reference Ferreira and Holbrook2002; Scott Reference Scott2011). The movement toward English was driven by the immigration of English-creole speaking workers into Trinidad, from other eastern Caribbean states such as St. Vincent and Barbados. Thus, English in Trinidad has a much shorter and perhaps more complex history than in Jamaica, the earliest reports of an English (creole) being around 1838 (Winer Reference Winer1993).
Notwithstanding historical differences, Trinidadian and Jamaican both co-exist with local Standard English varieties in similar though not identical sociolinguistic contexts (Winford Reference Winford1997). The contemporary language situation in Jamaica, for example, puts the creole variety in a diglossic situation with Jamaican English, which for most speakers is accessed through formal education and writing (Devonish and Harry Reference Devonish, Harry, Kortmann and Schneider2004). Similarly in Trinidad, the standard is the first language for some speakers, and is accessible through the education system (Winford Reference Winford1993; Youssef and James Reference Youssef, James, Kortmann and Schneider2004; Ferreira and Drayton Reference Ferreira and Drayton2014; Deuber Reference Deuber2010). Rural Jamaican has been argued to be more radical and divergent from an English superstrate than Trinidadian, which is a relatively uniform intermediate creole that is used both in rural and urban areas (Winford Reference Winford1997). The nature of the contact among the different varieties in each country has led to intricate patterns of variation, further complicated by the languages of newer migrants, especially in Trinidad.Footnote 1 Some of the speakers of these creoles have access to several social dialects that are invoked in different contexts, but the lines of demarcation between social dialects is not always clear cut. There are often clearer differences in some aspects of the grammar than in others, and having shared lexical cognates with the local English varieties feeds linguistic ideologies of the creoles as being substandard or deviant (Irvine Reference Irvine2008). This holds for diachronic aspects of the language as well as for changes over time. The presence of these kinds of sociolinguistic variation makes description of the linguistic systems challenging, and at the same time the social significance of linguistic forms can change over time, presenting another challenge for researchers.
While this diachronic change occurs in all areas of the grammar, it can be argued that changes in phonology are perhaps the most salient and most complex of the linguistic subsystems. Furthermore, given the perceived low status of the creoles relative to English, the phonological system is especially sensitive to sociolinguistic variation (see discussions on Jamaican creole in Beckford-Wassink Reference Beckford-Wassink2001; Meade Reference Meade2001; Irvine Reference Irvine2004; Devonish and Harry Reference Devonish, Harry, Kortmann and Schneider2004; and on Trinidadian in Youssef and James Reference Youssef, James, Kortmann and Schneider2004). In this regard, some areas of segmental phonology of Trinidadian and Jamaican are the same as in local English varieties, some creole features are stigmatized and avoided in English speech, and the prosodic properties show areas of overlap alongside areas of sharp differences.
Since prosody, like other areas of the grammar, can and will change over time, we can reasonably expect varieties like Jamaican and Trinidadian to develop and display changes in their stress and intonation systems when older and newer versions are compared. Thomason and Kaufman (Reference Thomason and Kaufman1988) predict that contact-induced change can affect all aspects of the linguistic system constrained only by sociohistorical factors. We will argue below that the prosody of Trinidadian (but not rural Jamaican) reveals some diachronic change, in just over fifty years, as a result of changes in the sociolinguistic dynamics, which include contemporary contact situations. This kind of “swift” change is due precisely to the ecological context in which the variety exists. These factors include contact with Indian Bhojpuri-speaking immigrants in the late nineteenth and early twentieth centuries, as well as possible influence from Trinidadian French creole right up to the first decade or so of the twentieth century (Winford Reference Winford1972; Youssef and James Reference Youssef, James, Kortmann and Schneider2004; Gooden, Drayton, and Beckman Reference Gooden, Drayton and Beckman2009). We will also show, however, that there are other prosodic patterns brought about by contemporary contact situations and the recasting of ethnic identities.
This chapter focuses on two main issues. The dominant task is a review of the phonology of intonation in Jamaican and Trinidadian focusing on intonation, prominence, and phrasing in the more recent data. We then evaluate intonation, prominence, and phrasing in the older data, so that we can compare the results from both sets of rural Jamaican creole (Jamaican) and rural and suburban Trinidadian creole (Trinidadian). A variety of early twentieth-century audio recordings spanning a period of just over sixty years are examined and are compared to more recent recordings. The aim is to see whether there are commonalities in the observed phonological patterns in each of the languages. This leads naturally to a second issue regarding the development of systems/prosodic structure in these new varieties, and might then shed light on (a) the differences in the prosodic structure in Trinidadian, which are not observed in Jamaican, and (b) differences in the alignment of the F0 on stressed syllables. In examining these varieties, we also bear in mind that the developing norms of usage for the national English varieties in each country (see Beckford-Wassink Reference Beckford-Wassink2001; Irvine Reference Irvine2004) quite likely influenced the related creoles.
The remainder of this introduction provides an overview of the role of prosody in language change and reviews the kinds of changes that have been documented in the segmental phonologies of the languages. We then present our theoretical framework for the prosodic analysis (Section 17.2.1), a brief description of the data-collection methods for the contemporary data (Section 17.2.2) and an overview of the observed prosodic and intonational patterns in the more recent data (Section 17.2.3). Section 17.3 provides the prosodic analysis for the older data, and in Section 17.4 we summarize and discuss the results.
17.1.1 Prosody in Language Change
The phonological systems of Caribbean varieties of English are comparatively under-researched, yet are very important in achieving a thorough description of these varieties. Following Irvine (Reference Irvine2004: 67), “phonology is the aspect of the language that is not reproduced in the written texts … introduced … in the school system and it is the aspect of the language that distinguishes the educated Caribbean speaker from his/her counterpart in other parts of the ‘English-speaking’ world” (paraphrased and italics added).
Prosody is a kind of structure, a “grammar” that has to be learned and/or interpreted correctly. It imposes a rhythmic structure on speech, signaling the divisions of utterances into interpretable parts (Beckman Reference Beckman1996). Intonation is then layered on top of these parts to convey a variety of discourse pragmatic meanings. Both prosodic structure and intonation work to signal information about the relatedness of constituents. The distribution of intonational features like pitch and relative prominence in a given utterance is done only in ways permitted by the prosodic structure (Ladd Reference Ladd2008). Given these factors, it is clear that changes in the prosodic system of creoles is a kind of structural change that needs to be studied, alongside our studies of grammatical change, if we are to fully understand how the languages developed or changed over time (Devonish Reference Devonish1989; Clements and Gooden Reference Clements and Gooden2009). In other words, prosody is an important part of telling the story of the development of creole languages, yet it is very often ignored in these discussions. At the same time the value of phonology (and we add phonetics) research to discussions on language change in creoles has not been completely lost on the field. Earlier works by Cassidy (Reference Cassidy1961), Cassidy and LePage (Reference Cassidy and Robert1967), LePage (Reference LePage1974), later by Singh and Muysken (Reference Singh and Muysken1995), for example, encourage a reassessment of theories of creole formation guided by findings from phonological research. Lesho's (Reference Lesho2013) dissertation and Beckford-Wassink (Reference Beckford-Wassink1999a) very clearly highlight the value of sociophonetic research to understanding variation and change in the phonological (vowel) systems of creoles. As Lesho aptly argues, even where a creole's phonological system bears resemblances to its input languages, the phonetic implementation could yield significant differences, especially given the effects of social factors and ideological stances of speakers.
Recent research by Sandler et al. (Reference Sandler, Meir, Dachkovsky, Padden and Aronoff2011) argues that prosody and its interaction with sentence structure should be incorporated into any model of language evolution. They suggest that, in a new language, prosody may be the only indicator of functions like marking constituent boundaries or relations between them, as well as marking pragmatic functions. Similarly Givón (Reference Givón and Givón1979) claims that prosody is vital to the development of “new” languages like creoles, as it signals the relations between constituents before syntactic structures are developed. For us, this means that in (early) creole formation, prosody played an important role in grammaticalization processes (see also discussions in Wichmann Reference Wichmann, Narrog and Heine2011). Certainly, there is no way to truly test this, as there is no access to these early speech recordings with which to evaluate intonational cues to prosodic structure. However, we can look to synchronic and diachronic realizations in different Caribbean creoles to provide some evidence. These can then complement the existing independent analyses of historical change. Still, we must be cautious since prosodic classification of creoles at any particular stage remains challenging (Drayton Reference Drayton2007), because the research is still in its infancy and we cannot assume that prosodic changes proceed via the same mechanisms as grammatical changes. One could construct an argument along the lines that, due to their tonal West African substrates, all Caribbean creoles were at one time tonal, and less conservative creoles have shifted from tone specification to accent specification. Our position is that this argument might be too simple because the effect of language contact on prosodic systems is complex and the outcomes are not always reliably predictable (Gooden, Drayton, and Beckman Reference Gooden, Drayton and Beckman2009). Moreover, the processes of change involved in creole formation are not unique and involve universal principles and internal processes of language change that affect all languages (see discussions in Winford Reference Winford2003).
17.1.2 Phonological Variation in the Vowel Systems
Although this chapter focuses on higher-level phonological structure, segmental phonology also sheds light on the development of creoles. Smith and van de Vate (Reference Smith, van de Vate, Bhatt and Plag2006), for example, argue that “Suriname type” vowel systems are to be found in Eastern Maroon varieties in Jamaica, such as in Moore Town, such that there are no long vowels and there is preconsonantal monopthongization. In contrast, other varieties of Jamaican are said to preserve diphthongs and vowel-length contrasts.Footnote 2 Devonish and Harry (Reference Devonish, Harry, Kortmann and Schneider2004) argue for the presence of implosives in Jamaican, and Smith and Haabo (Reference Smith and Haabo2007) argue the same for Saramaccan. In both cases, researchers use this to support their argument for substrate influence on the phonology of the languages. Phonetic analyses of creole phonologies are rare, however, with some varieties like Jamaican receiving more treatment than others (see Lesho Reference Lesho2013 for a review). We highlight a few examples of sociophonetic analyses of creole vowel phonology as the area that has received most treatment in the literature.
The vowel systems of Jamaican and Trinidadian have been the subject of sociophonetic analyses examining variation according to sociolinguistic factors like geography, gender, social class, and networks of interaction. To date, Beckford-Wassink's work (Reference Beckford-Wassink1999a, Reference Beckford-Wassink1999b, Reference Beckford-Wassink2001, Reference Beckford-Wassink2006) is the most comprehensive sociophonetic analysis of the vowel system of contemporary Jamaican data. To our knowledge, there is no analogous analysis of earlier Jamaican data that would be comparable to say Leung's (Reference Leung2013) analysis of 1970s Trinidadian. Beckford-Wassink (Reference Beckford-Wassink2001, Reference Beckford-Wassink2006) examined tense vowels in the beat, bath, boot class and lax vowels in the bit, trap, book class in both a rural conservative Jamaican creole variety and Jamaican English. By using a three-dimensional metric (F1, F2, and duration) to analyze the degree of vowel overlap, Beckford-Wassink showed that in these cases the rural creole variety showed partial spectral overlap, meaning that vowel length rather than spectral differences distinguished between vowel categories. The vowel quality metric was distinctive in Jamaican English, however. This lends some support to Smith and van de Vate's (Reference Smith, van de Vate, Bhatt and Plag2006) hypothesis that vowel-length contrasts are retained in more conservative varieties.
Leung (Reference Leung2013) examined variation in monophthongs in Trinidadian English with some comparison with Trinidadian English creole data from Winford's Reference Winford1972 PhD dissertation. The three male speakers were born in 1898, 1911, and c. 1932 and therefore give samples of speech of persons born in the late nineteenth century and early twentieth century. Leung's findings prove very interesting, revealing some shifts in the production of certain vowels over time. The nurse lexical set, for example, is produced by the informant born in 1898 as an open-mid back rounded vowel with variations [ɔː ∼ oː ∼ ɘː], due to its close acoustic proximity to the vowels in goat and north. Leung points out that this is different from contemporary productions of the vowels where the acrolectal productions of nurse involve the close-mid central unrounded [ɘː] and the creole speaker's open back rounded vowel [ɒ]. The other speaker from the rural area (born 1911) also displayed a large degree of variation in his vowel tokens as well. While the north and nurse vowels are not as close for the speaker born in 1911 as they are for the speaker born in 1898, Leung noted a tendency for the [+high] vowels like thought, strut, and lot to cluster together as did the [+low] vowels found in trap, bath, and start. There was a wide degree of variation shown by this speaker, for example [ʌ ∼ ɒ ∼ ɔ] for the strut and lot lexical sets. The speaker from the urban area, born in 1932, exhibited a different vowel pattern from these other two, which Leung described as closer to contemporary acrolectal Trinidadian. Leung describes this speaker's vowel set as being similar to Wells's (Reference Wells1982) description of Trinidadian English.
These data prove very interesting as they highlight a vowel pattern which existed at one time in Trinidad, but which no longer exists except perhaps in very old, rural speakers of more conservative forms. The current situation, as Leung shows in her examination of rural and urban speakers (recorded in 2009), is that of a stabilized system in which there are mergers and the amount of vowel variation has been reduced. Our analysis suggests that there are changes in the prosodic systems of the languages, analogous to those documented in the segmental system.
17.2 Theoretical Background
The prosodic analysis is done in the Autosegmental Metrical (AM) framework (Pierrehumbert Reference Pierrehumbert1980; Beckman and Pierrehumbert Reference Beckman and Pierrehumbert1986; Ladd Reference Ladd2008) with ToBI-type transcriptions of intonational tunes (Beckman, Hirschberg, and Shattuck-Hufnagel Reference Beckman, Hirschberg, Shattuck-Hufnagel and Jun2005), and we assume a Strict Layer Hypothesis of prosodic structure (Selkirk Reference Selkirk1984). The AM model assumes that pitch accents, which consist of either a single tone or a sequence of tones, are phonologically associated with metrically prominent syllables (Pierrehumbert Reference Pierrehumbert and Horne2000). We examine two major parameters of prosody, prominence and phrasing, at both the lexical and postlexical levels since both determine the prosodic property of an utterance. Prominence marking at the word level for Trinidadian (Drayton Reference Drayton2013) and for Jamaican (Gooden Reference Gooden2003) is cued by lexical stress. At the postlexical level, prominence marking is cued by the head of a phrase (marked by a nuclear pitch accent), or by a tone at the phrase edge, or by both (see Table 17.1). We follow Ladd's (Reference Ladd2008) typology in defining the intonational differences between the varieties. Although the typology was designed to describe intonational differences among languages, it is also applicable to varieties within the same type, or in this case putative differences due to changes within the system. We have argued elsewhere (Gooden, Drayton, and Beckman Reference Gooden, Drayton and Beckman2009) that Jamaican and Trinidadian exhibit type IV differences, i.e. phonotactic differences, specifically in the timing of F0 patterns relative to the sequence of speech segments in an utterance. For instance, the languages show differences in the alignment of the rise-fall patterns on prominent syllables in broad focus declaratives.
| H* | A tone target on the accented syllable of a word or phrase, which is in the higher part of the speaker's pitch range during production of the phrase. |
| L* | A tone target on the accented syllable of a word or phrase, which is in the lower part of the speaker's pitch range during production of the phrase. |
| < | An early pitch accent diacritic indicating an F 0 peak that precedes the accented syllable. |
| > | A late pitch accent diacritic indicating an F 0 peak that aligns after the accented syllable. |
| ↑ | An upstepped or elevated pitch accent diacritic indicating an unusually high F 0 target. |
| H+L* | A bitonal accent sequence consisting of a relatively high rise in F 0 followed by a low F 0 target on the accented syllable. |
| H*+L | A bitonal accent sequence consisting of a high F 0 target on the accented syllable followed by a relatively low F 0. |
| !H* | A high tone that is realized in a downstepped pitch range, that is, the pitch is still relatively high, but is measurably lower than that of the previous H* tone in the IP. |
| L-, H-, M- | A phrasal tone which marks an intermediate level (ip) intonational boundary. L- marks a boundary with a low F0 target at the right edge of the ip; H- marks a boundary with a high F0 target at the right edge of the ip; M- marks a boundary with a mid-level F0 marking the right edge of the ip. |
| L%, H%, M% | A phrasal tone which marks every full intonation phrase (IP) boundary. L% marks a boundary with a low F0 marking the right edge of the IP; H% marks a boundary with a high F0 marking the right edge of the IP; M% marks a boundary with a mid-level F0 marking the right edge of the IP. |
17.2.1 Prosody of Trinidadian and Jamaican
The prosodic systems of the Caribbean English creoles are perhaps best described as a smorgasbord, reflecting varying degrees of hybridity. This variation is due in part to the contact history of the varieties and in part to the multiple functions of prosody. The variation is also loosely aligned with a classification of more conservative to less conservative languages, shifting from lexical tone usage to no lexical tone (see discussions in Clements and Gooden Reference Clements and Gooden2009; Gooden, Drayton, and Beckman Reference Gooden, Drayton and Beckman2009). There has been some controversy over the characterization of the prosodic systems of Trinidadian and Jamaican as having lexical tone and/or stress. Recent research on the prosody of Jamaican (Gooden Reference Gooden2003, Reference Gooden and Jun2014) and Trinidadian (Drayton Reference Drayton2013), however, show that both varieties have lexical stress systems with intonationally marked prominences. This means that the F0 aligned with prominent syllables is not contrastive. This is the case for all the data we examine, including the ultra conservative Eastern Maroon Jamaican variety.Footnote 3 This is in contrast to the situation in other varieties, like Saramaccan (Good Reference Good2009), Curaçaoan Papiamentu (Remijsen and van Heuven Reference Remijsen and van Heuven2005; Kouwenberg Reference Kouwenberg2004; Rivera-Castillo and Pickering Reference Rivera-Castillo and Pickering2004; Rivera-Castillo Reference Rivera-Castillo2009), or Ndjuka (Huttar and Huttar Reference Huttar and Huttar1994), which clearly show contrasting F0 patterns that can be linked to a lexical or grammatical tone system.
17.2.1.1 Phonology of Word Level Stress
Both Jamaican (Wells Reference Wells1973; Alderete Reference Alderete and Benedicto1993; Gooden Reference Gooden2003, Reference Gooden, Huber and Velupillai2007) and Trinidadian creoles (Drayton Reference Drayton2013) are weight sensitive systems with trochaic foot structure. Native speaker judgments of stressed syllables correlated strongly with those identified by Allsopp (Reference Allsopp1996) for Trinidadian and Cassidy and LePage (Reference Cassidy and Robert1967) for Jamaican. Main stress in Jamaican generally falls on the leftmost or only heavy syllable in a word, but there is no weight-based preference for stress among CVV, CVVC, CVCC, and CVC syllables. Secondary stress falls two syllables away from the main stress (in any direction). When the primary stress is on the initial syllable in a trisyllabic word, secondary stress falls on the word-final heavy syllable. However, when primary stress is on the penultimate syllable, there is no secondary stress. In Trinidadian, main stress also falls close to the left edge of the word, and the computation of stress takes syllable weight into account, with closed syllables and syllables with long vowels and diphthongs being heavier than open syllables. Secondary stress is found two syllables away from the main stress, and is found in words longer than three syllables, since the final syllable according to Drayton's (Reference Drayton2013) analysis is not typically stressed in Trinidadian. These stress analyses are also supported by acoustic analyses of the contemporary data sets, targeting words of different lengths in different prosodic positions in sentences and in isolation.
17.2.1.2 Methods
The contemporary Jamaican data are from Gooden (Reference Gooden2003). Speakers included men and women (ages 29–80+ at the time of the recording, i.e. birthdate between 1973 and approximately 1918) from a rural area, Top Alston, Clarendon. The majority of speakers were basilect-dominant and reported using Jamaican creole in most informal social settings and English in most formal settings. The data are semi-spontaneous speech elicited using a combination of an elaborated interview style and a picture-task, in addition to conversational style data using a traditional sociolinguistic style interview. The methods mimic carefully controlled laboratory speech while avoiding read speech. The elicitation protocol yielded different prosodic contexts as follows: broad focus statement, yes-no question, wh- questions, narrow focus constructions, complex sentences with several subordinated clauses, sentences with multiple foci. Examining these different contexts was important since the realization of pitch accents is lexically specified and is therefore subject to influence from the discourse pragmatics.
Two recent acoustic phonetic studies of Trinidadian provide a considerable amount of data (Leung Reference Leung2013; Drayton Reference Drayton2013). The data described here were collected by Drayton in 2006–2007 and cover a wide geographical area including the areas sampled by Winford in 1970. The data collection methods are similar to those used for Jamaican. These data yielded target words in final and nonfinal position in statements and questions and in a focus condition. The complete data set includes sixteen speakers, both male and female, in conversation at their homes or a community site. Two speakers are from the suburbs of Port of Spain and two from a similar geographical location to Mayo (rural). The speakers ranged in age from twenty-four to fifty-nine years at the time of the recordings (birthdates 1947–1982). Finally, speakers self-identified as African, Indian, or Mixed. Although most speakers could approximate Trinidadian (Standard) English with varying degrees of success, they were primarily speakers of Trinidadian creole.
These recent Jamaican and Trinidadian data were digitized at a sampling rate of 44.1 kHz and were analyzed in Praat (Boersma and Weenink Reference Boersma and Weenink2010). For the intonational analysis, the data were analysed auditorily, assisted by visual inspection of the waveform and F0 contour, and were crosschecked by both authors following the initial coding of the individual data sets. For the vowel-duration analyses, vowel onset and offset were marked on a wideband spectrogram at points when the F2 ceased transition into and out of the vowel respectively (DiPaolo, Yaeger-Dror, and Beckford-Wassink Reference DiPaolo, Yaeger-Dror, Beckford Wassink, Paolo and Yaeger-Dror2010). Details of the other measurements are given below. Statistical analyses were done in R (R Development Core Team 2010).
17.2.1.3 Acoustic Cues to Stress
Given that both Trinidadian and Jamaican are stress languages, lexical prominence is cued indirectly by the F0 (Fry Reference Fry1958), and in both varieties this is realized as a fall (Lawton Reference Lawton1963; Wells Reference Wells1973; Gooden Reference Gooden2003). Since the F0 by itself is unreliable for marking prominence, other acoustic cues signal stress, i.e. duration, intensity, and vowel quality as expressed by F1/F2 differences (Arvaniti Reference Arvaniti2000). Stress in Jamaican is cued by pitch prominence and duration in addition to the reduction or deletion of unstressed syllables (Gooden Reference Gooden and Jun2014). In fact in words like MAda “mother” and maDA “female religious leader,” that are often cited as examples of lexical tone contrast, it is syllable duration rather than vowel duration that differentiates between stressed and unstressed items. For Trinidadian, Drayton (Reference Drayton2007) reported results from an analysis of 275 words of varying lengths. A series of t-tests revealed that there was a significant difference in the mean duration for stressed vs. unstressed syllables (p < 0.01) and vowel quality F1 (p < .05) and F2 (p < 0.01), such that unstressed vowels tended to be reduced, but intensity ratios were not significantly different. Stressed vowels were on average 26 ms longer than unstressed vowels.
17.2.2 Intonational Phonology
As noted earlier, the more recent Jamaican and Trinidadian data show tonal alignment differences. For example, the HL sequence observed in broad-focus utterances is analyzed as binary pitch accent sequences H+L* or H*+L in Jamaican, whereas in Trinidadian the H is associated with an accentual phrase (AP) phrase boundary and a unary L* pitch accent. (An AP is an intermediate level prosodic unit, smaller than an IP and larger than a phonological word (see Jun Reference Jun1998 for detailed discussion). APs are assigned language specific phonetic markings of prominence to delineate boundaries. The IP in languages with APs is therefore made up of one or more APs which pattern the same way in terms of phonological features like phrase length, tonal prominence and tonal alignments.) The languages also show differences in the LH pattern associated with focused items (in non-final position). Jamaican creole has a L+H* pitch accent on emphatic focused items whereas Trinidadian has a L* and a very prominent H marking the AP boundary. In other words, the data suggests that Jamaican marks focus with pitch accents and pitch range expansion and Trinidadian marks it through F0 manipulation only. Additional details on the individual intonational systems are summarized just below.
To demonstrate more clearly the differences in F0 alignment between the varieties, Figure 17.1 shows overlaid F0 contours on a normalized timeFootnote 4 axes for the target word alligator in non-final position in a broad-focus declarative. There are a total of nine speakers, four for Trinidadian and five for Jamaican, all from the more recent data sets. We employ normalization because individual differences may greatly affect the description of the F0 contour since it is a speaker-dependent physiological characteristic. Hertz values were converted to an auditory scale, Equivalent Rectangular Bandwidth (ERB), which provides a more accurate representation of speakers’ pitch perception.
Figure 17.1 Mean interpolated F0 contours on a normalized time axis. The contours are averaged across repetitions from a portion of the sentence, Him want wan || alligator and (some) yam. “He wants an alligator and some yams.”
The highlighted portion of the graph is the pretonic and tonic syllables [lɪˈɡɛ]. The fall starts in the pretonic syllable and ends on the tonic syllable, demonstrating that there is a consistent fall in F0 on the stressed syllable for all the speakers in both languages.
Postlexical pitch accents in Jamaican are at least of two types, monotonal accents (e.g. H*) and bitonal accents (e.g. H+L*). Two larger prosodic constituents above the word, an intermediate phrase (ip) and an intonational phrase (IP) are marked with tones at the right edges. IPs can have H%, L%, or M% tones and ips can have L- or H- (Gooden Reference Gooden and Jun2014). Intonational marking of focus occurs alongside syntactic means since, as with other creoles, Jamaican creole typically marks focus syntactically (Christie Reference Christie and Christie1998; Patrick Reference Patrick, Kortmann, Schneider, Upton, Mesthrie and Burridge2004; Durrleman Reference Durrleman2005, Reference Durrleman2007). However, Gooden (Reference Gooden and Jun2014) showed that while double foci constructions are prohibited via syntactic reorganization, it is permissible when combining prosodic and syntactic strategies. Narrow/emphatic focus constructions are marked with an L+H*, and the research so far suggests that there is no de-accenting in the post-focal domain, even when the focused item occurs early in the IP. Table 17.2 provides a summary of pitch accents and boundary tones.
Table 17.2 Summary of pitch accent and boundary tones in contemporary Jamaican
| Word level | L*, H*, H+L*, H*+L, L+H* with emphatic focus, possibly (!H+L*) | |
| Intermediate Phrase (ip) | L-, M- or H- on medial ip or final ip preceding IP boundary tones | |
| Intonational Phrase (IP) | L% on final or medial IP in broad focus statements, emphatic focus statements, Wh-questions
H% on final IPs in yes-no questions, emphatic focus yes-no questions M% on final IPs in yes-no questions, continuations rises |
In an analysis of recent data, Drayton (Reference Drayton2007, Reference Drayton2013) recognized two larger prosodic constituents above the word, an Intonational Phrase (IP) as well as an intermediate level phrase: an Accentual Phrase (AP). IPs can have H% (e.g. in yes-no questions or continuation rises) or L% edge tones. The AP consists of a word or group of words delimited by H boundary tones at the right edge of the constituent, and has at least one L* pitch accent anchored to stressed syllables (see Figure 17.2). These L* pitch accents are argued to be present in these recent data regardless of the ethnic identity of the speaker and is the result of contact with Indo-Trinidadians who themselves have been influenced by Trinidadian Bhojpuri.Footnote 5 Support for this comes from the observation that other “Indian Englishes” have L* accents on prominent syllables as well (Harnsberger Reference Harnsberger1999; Pickering and Wiltshire Reference Pickering and Wiltshire2000). Of note is that the data from Afro-Trinidadians recorded in the 1970s that we have looked at do not have this pattern.
Figure 17.2 F0 contours of statements: (a) broad focus, He wants a banana and yam, (b) narrow focus (non-final), He wants a banana (not mango) and yam, (c) narrow focus (final), The calabash has a banana (female Afro-Trinidadian speaker from Princes Town).
Items in focus are made prominent through F0 manipulation and duration. Consequently, as shown in Figure 17.2(b), the F0 in the AP preceding the focused item is reduced compared to the analogous AP under broad focus in Figure 17.2(a). The L* pitch accent on the focused item is also realized at a lower F0 and the H tone associated with the postfocal AP is realized much higher, but is not obligatory since it is not observed in absolute phrase final position (see Figure 17.2(c)).
The L* pitch accent is also seen in the spontaneous speech data in Figure 17.3. The H% boundary tone here marks a continuation rise and is preceded by the H tone of the final AP.
Figure 17.3 F0 contour showing L* H (male Afro-Trinidadian speaker from Dabadie).
Figure 17.4 F0 contour showing bitonal pitch accents L+H* L* H in younger speaker (male Afro-Trinidadian speaker from Dabadie).
17.3 Older Recordings
The F0 contours presented in this section are more directly comparable with the interview-style data we collected, although patterns seen in the elicited semi-spontaneous speech data are also referred to.
17.3.1 1950s Jamaican
The older Jamaican recordings were digitized (sampling rate 48 kHz) from David DeCamp's (Reference DeCamp1957–1958) original reels,Footnote 6 recorded in Jamaica between November 1957 and August 1959. The time period overlaps a bit with data reported in Lawton (Reference Lawton1963), who, as noted above, identified L F0 on stressed syllables in Jamaican and was the first to demonstrate that it was also important for prosodic phrasing. There is also some overlap with data reported in Wells (Reference Wells1973), but the bulk of these speakers were Jamaicans residing in London. The interviews described are from a wide range of parishes and include rural and urban locations as well as Maroon settlements. We focus on data from Banana Ground-border of Manchester and Clarendon, Trout Hall in Clarendon, and one Maroon settlement, Moore Town in Portland. These were chosen as they are all rural and so can be compared with the more recent Jamaican data described above. The Maroon settlement was chosen as it represents a more conservative form of creole, the Jamaican Spirit Language and Kramanti (Bilby Reference Bilby1983; Devonish Reference Devonish2005) and might reasonably show prosodic features not seen in other rural varieties.
Generally, the patterns are consistent with those seen in the more recent data, although there are some differences. As of yet, there are no observed differences in prosodic structure. IPs have L% boundary tone as in Figure 17.6 and a medial boundary tone M% (Figures 17.5, 17.7, 17.11). The M% appeared in continuation rises and also in yes-no questionsFootnote 7 in the more recent Top Alston data. The M% boundary is at the mid range of the speaker's pitch range and appears most consistently in the speech of older speakers in both the older and more recent recordings. The next level of phrasing, the intermediate phrase, has a smaller perceived degree of disjuncture than IPs, and we have marked only the clear cases. The L-phrase accent marking it is typically scaled lower than the L% boundary tones (e.g. Figure 17.6). Figure 17.5 shows the same speaker as in Figure 17.6, with M-phrase tones and M% boundary tones. There are several instances, however, which might be better analysed as utterance-internal IPs rather than as ips (see Figures 17.7, 17.9). In all of these cases there is a measureable or discernable pause between phrases and the degree of perceived disjuncture is not as small as at the ip boundary.
Figure 17.5 F0 contour showing syntactic marking of focus and an M% and M- in the sentence, Whereas, the man ROBBED you, lit. “whereas, its rob that the man robbed you” (Carter, male speaker from Banana Ground). Continued in Figure 17.6.
Figure 17.6 F0 contour showing phrase accents in sentence continued from Figure 17.5, … and he is gone, because after all he can't move it (Carter, male speaker from Banana Ground).
Figure 17.7 F0 contour showing a variety of pitch accents in sentence, Until after you bury, after you BURY the dead (upper panel), then they have the, the wake (lower panel) (Barnett, male from Trout Hall).
Figure 17.8 F0 contour showing focus in the phrase, …this man's house and THAT house (Harris, male from Moore Town).
Figure 17.9 F0 contour showing utterance internal IPs, in the sentence, After they BURY the dead, then they do the singing (Barnett, male from Trout Hall).
Four types of pitch accents were observed: falling (H+L*, H*+L), rising (L*+H, L+H*), high (H*), and low (L*). Figure 17.7 nicely illustrates the pragmatic function of pitch accents in Jamaican. The first occurrence of bury has broad focus (upper panel) and has a falling F0 (H+L*). The second occurrence is focused and has a rising F0 (upper panel). This is the familiar LH pattern seen in the focus contexts in more recent recordings, but there are alignment differences. In this case, there is an L*+H as opposed to an L+H*, so the low F0 is aligned with the initial stressed syllable of the word and the peak is realized later. The L*+H pitch accent is also seen in Figure 17.9 on another focused rendition of the word bury, and again in Figure 17.8 on the monosyllabic word dat “that.” Figure 17.10 has focus on the word wan “none” and is realized as L+H*.Footnote 8 In Figure 17.11, the word lean is focused and has the same L+H* pitch accent.
Figure 17.10 F0 contour …but of the land they gave the maroons, not me … they, they did not give me ANY. I (alone) should have gotten more (Harris, male from Moore Town).
Figure 17.11 F0 contour showing utterance internal IPs with both L% and M%, in the sentence, But however, we are called ambadasha;Footnote 9 (Kramanti term) (upper panel), and we are LEAN, and we will not fall (lower panel) (Harris, male from Moore Town).
17.3.2 1970s Trinidadian
These older data were collected by Donald Winford in 1970 for his PhD study on phonetic-phonological aspects of sociolinguistic variation in two communities (urban and rural). The original analogue recordings were digitized (sampling rate 44 kHz) and made available to us. The complete data set includes thirty-six male speakers in conversation at their homes or workplaces. Twenty-one speakers were from St. James, a suburb of Port of Spain, and fifteen were from Mayo, a small village in the south of Trinidad. These speakers ranged in age from thirty-five to eighty (birth dates 1890–1935). Unlike in the contemporary data, speakers self-identified only as either African or Indian, and the third category “mixed” was not used explicitly. The implications of these self-identifications for the observed prosodic patterns are discussed further below.
The speaker in Figure 17.12 shows the canonical L* pitch accent on stressed syllables, and the IP is organized in terms of APs marked by H boundary tones at their right edges. The data from Afro-Trinidadian speakers is interesting in several ways. First, it clearly shows bitonal pitch accents on stressed syllables. In Figure 17.13, there is a rising accent (L+H*) on the word see and a falling accent (H*+L) on the word anything. Figure 17.14 is from the same speaker and shows an L+H* on working and !H*+L on conditions. In addition, there appears to be APs bounded by a right-aligned H tone. These appear to be far less frequent than seen in the speech of Indo-Trinidadians of the same age cohort and younger Afro-Trinidadians (as seen in Figure 17.3). This is taken as evidence for earlier influence from contact with Trinidadian Bhojouri speakers.
Figure 17.12 F0 contour showing L* pitch accents in APs. Indo-Trinidadian from the Mayo area (rural).
Figure 17.13 F0 contour showing H*, H*+L, and L+H* pitch accents on stressed syllables (male Afro-Trinidadian from Mayo).Footnote 10
Figure 17.14 F0 contour showing !H*+L and L* pitch accents on stressed syllables (male Afro-Trinidadian from Mayo).
17.4 Discussion and Summary
The period between the late 1890s and the 1940s brought some changes to Caribbean creoles. Trinidadian shows changes in prosody, mainly in the marking of prominent syllables and in phrasing since there is a more consistent marking of APs in the speech of Afro/Mixed speakers in the more recent data. Using the oldest speakers from both data sets, this would have taken place between 1890 and 1947, so that speakers born after that period would have prosodic features that are similar to those used by the contemporary speakers described here. As discussed earlier, contemporary Trinidadian intonation features include L*H sequences, with L* anchored to the stressed syllable of a content word, and an H tone marking the boundary of an Accentual Phrase. This pattern holds for speakers across various ethnic identities and urban-rural areas. In the older data, especially in the African-identified speakers, there was a greater presence of binary pitch accents (e.g. L+H*) and H* tones, in addition to L* on stressed syllables, and a less consistent use of APs. At the same time, the fact that APs were minimally present in the speech of older Afro-Trinidadians suggests that this prosodic category had started to creep into the prosodic structure, and is now further along in the speech of younger speakers. The data from the Indo speakers more closely resembled that of the ubiquitous modern AP pattern of L* followed by H.
The oldest speakers from both sets of Jamaican data would have been born between the late 1890s and 1918, the youngest between the 1930s and 1970. This is roughly the same time period as in Trinidad, i.e. among speakers born after the 1930s or 1940s. Among the older speakers, there are no changes in the prosody, and, as discussed earlier, the message of maintenance rather than shift is also seen in the vowel system, as rural speakers maintain “older” length contrasts while other non-rural speakers make use of spectral properties (Beckford-Wassink Reference Beckford-Wassink2001). There is not sufficient data from younger speakers in the more recent Top Alston data to tell if there are age-related changes in that cohort. Still, the data from these younger speakers is consistent with those from the older speakers in the community. One clear difference is that the M% in continuation rises is observed most consistently in the speech of the older speakers across both sets of data. Older speakers also appear to be making more use of utterance-internal IPs. Marking focus prosodically involves increased pitch range, duration, and differences in pitch accent marking, a strategy used by both older and younger speakers. The older speaker from the more recent Top Alston data did not show post-focal de-accenting, and we see a similar pattern in the older data as well.
The changes in Trinidadian are very possibly a marker of a new “Trinidadian” identity among younger Trinidadians (see similar discussions in Winford Reference Winford1972). Winford highlighted clear ethnolinguistic differences between many older Indo-Trinidadians and older Afro-Trinidadians, including differences in their pitch and intonation patterns. Our hypothesis that a “big” change is in progress in Trinidadian phonology is also supported by Leung's (Reference Leung2013) work. She suggested that, compared to Winford's results, there is now less vowel variation than earlier. The same sociohistorical factors that can account for this decrease in variation are likely the same ones that drove the prosodic changes. Three factors contribute to these changes. First, a shift in identity that resulted in a larger “mixed” category in speakers born around the 1940s and later (as documented, for example, in the 2000 Trinidad census data).Footnote 11 The census reports 40 percent Indo-Trinidadians, 37 percent Afro-Trinidadian, and 20 percent of mixed ethnicities for the country. The second factor is the increasing amalgamation of formerly disparate groups, and their participation in mainstream culture (see Mohan Reference Mohan1990 for some discussion). Thirdly, there is increased contact between Afro-Trinidadians and Indo-Trinidadians, for example through marriage and internal migration and (re)settlement patterns. As speakers redefine themselves in their changing ecologies and create new sociopolitically driven identities, the linguistic shifts become a reflection of the associated changing identities (Schneider Reference Schneider2003).
The combined results from our research and the research on vowel variation suggest that both Trinidadian and Jamaican are likely to be between the last two stages of emergence: the fourth stage, i.e. endonormative stabilization,Footnote 12 and the fifth stage, i.e. differentiation (Schneider Reference Schneider2003). The differences are determined by the local ecologies in which each variety exists. So, the kind of ethnolinguistic differences seen in Trinidad are not observed in Jamaica since these rural communities are far more ethnically homogenous (at 91.6 percent Afro-Jamaicans island-wide).Footnote 13
In summary, two different processes of language change have affected the varieties in the time period we examined. While we observe convergence in Trinidadian prosody, Jamaican shows retention. The changes in Trinidadian center on ethnolinguistic contact, the outcomes of which are varied and complex, and may also interact with other influences on the construction of identity, such as age. As such, we must ideally seek to understand the context in which speakers use the language and how they construct their identities within these contexts.Footnote 14 At the same time, ethnic groups are not static, and speakers’ identities might shift due to social context (Fought Reference Fought2006, Reference Fought and Hickey2010). This is demonstrated here, through the shifting ethnic affiliation of younger Trinidadians. We see a consequent convergence of the type of intonational pitch accents in both of the major ethnic groups as well as in the speech of mixed identity speakers, and a strengthening of the AP as a prosodic category below the IP. The maintenance of the features in Jamaican is facilitated by the rural context in which speakers reside.
