Research ArticleClosure duration as an acoustic correlate of the word-initial singleton/geminate consonant contrast in Kelantan Malay
Introduction
In experimental phonetic investigations of consonant gemination, the focus is usually on geminates that occur in word-medial position (e.g., Aoyama & Reid, 2006 on Guinaang Bontok; Ham, 2001 on Lebanese Arabic; Hansen, 2004 on Persian), while much less attention is given to word-initial geminates, which are claimed to be rare in the world׳s languages (Ladefoged & Maddieson, 1996). Indeed, some researchers (e.g., DiCanio, 2008; Dmitrieva, 2012; Kawahara, 2005, Kraehenmann, 2011, among others) argue that word-initial geminates are more phonologically dispreferred than word-medial geminates. This could be potentially due to the difficulties of perceiving clear contrasts in utterance-initial and word-initial contexts. For example, Padgett (2009) suggests that any consonant contrast is better situated in an intervocalic context that offers a clear beginning and an end point for the perception of the target consonant. By way of contrast, in word-initial position, there may be poorer perceptual cues to closure duration, especially in the case of voiceless stops where there are no acoustic cues to occlusion at the start of an utterance (see, e.g., Kraehenmann & Lahiri, 2008). On the other hand, native listeners of languages with word-initial geminates can reliably differentiate the contrasting phonemes even for voiceless stops in utterance-initial position, as shown for Pattani Malay (Abramson, 1986) and also for Cypriot Greek (Muller, 2001). These results indicate that additional parameters, such as intensity and f0, likely function as critical secondary cues in this context, as Ladefoged and Maddieson (1996:94) point out.
In this study, we aim to contribute to a greater understanding of consonant gemination across the world׳s languages with a detailed acoustic phonetic analysis of word-initial singleton/geminate duration in Kelantan Malay (KM), spoken in Malaysia. We focus specifically on the closure durations of voiced stops and sonorants (in which closure duration can be reliably measured in all contexts) as well as on those of voiceless stops embedded in a carrier sentence. Our main motivation is to identify the degree to which closure duration underlies the word-initial consonant length distinction in KM and also, to the extent that it is possible, compare results for KM with those reported by Abramson (1986) for closely related Pattani Malay (PM) and with relevant findings for other languages with geminate consonants (e.g., Ridouane, 2007 on Tashlhiyt Berber). The current experimental investigation will also examine the extent to which closure duration in KM singletons and geminates is further conditioned by factors such as utterance position (i.e., utterance-initial or utterance-medial positions), manner of articulation (e.g., stops or sonorants), voicing type (i.e., voiceless or voiced stops) and phoneme type within each manner category. This is drawn from the findings of past studies (e.g., Engstrand & Krull, 1994) which have shown that consonant gemination may vary in different contexts and environments (Section 1.4 will describe these factors in detail).
It is worth noting that, to date, no experimental phonetic research has been conducted on KM beyond a small number of short research papers directly associated with this study (i.e., Hamzah, 2010; Hamzah et al., 2011, Hamzah et al., 2012, Hamzah et al., 2014, Hamzah et al., 2015, Hamzah, Hajek, & Fletcher, 2012a). The current study is therefore novel in that it is the most detailed investigation yet of the word-initial consonant contrast in KM which, like closely related PM,1 only has geminate consonants in word-initial position. Our goal is to examine whether word-initial closure duration is a robust acoustic marker in defining geminate consonants, in line with other languages with medial geminates (see Section 1.2). More broadly, the findings will make a significant contribution to our understanding of KM and less-well studied language varieties as well as to the field of acoustic phonetics dealing with a relatively rare cross-linguistic phenomenon such as word-initial consonant gemination.
KM is part of the larger Malay language family spoken in the Malay archipelago. In Malaysia, KM is the primary lingua franca used by more than 1.5 million speakers in the state of Kelantan, which is located in the north-eastern part of the Malay Peninsula in Malaysia. This Malay variety stands out among most other such varieties in Malaysia because of the rare presence of geminate consonants in word-initial and, by extension, utterance-initial position. While KM and PM are mutually intelligible, it is crucial to note that PM in Thailand is different from KM in Malaysia in that the former is in intense contact with Thai, a lexical tone language, while the latter is in close contact with Standard Malay, a non-tonal language that is also the official and national language of Malaysia.
As mentioned earlier, there has been no experimental research that can offer empirical data for KM word-initial geminates beyond our own small-scale studies. Earlier works (e.g., Aziz & Ahmad, 2008; Karim, 1965; Mahmood, 2006) were only based on impressionistic evaluation, in which word-initial geminates were described phonologically, sometimes within a particular theoretical framework (e.g., within Optimality Theory by Aziz and Ahmad (2008)) but with no actual phonetic evidence provided. Karim (1965) observes that geminate consonants in KM occur only in word-initial position - across all places and manners of articulation. She notes that word-initial consonant gemination in KM is typically the “compensated” form of original syllables, words or phrases that are retained in Standard Malay. As shown in Examples 1.1 to 1.3 below, consonant gemination in KM can be accounted for by, for example, the “loss” of prepositional phrases (Example 1.1), and syllable reduction and assimilation in trisyllabic nouns (Example 1.2) and in reduplicated nouns (Example 1.3).
Geminates are easily found to contrast with singletons, as seen in a few examples here, and more copiously in our experimental data set shown in Table 3 in Section 2.1:
Traditionally, the term ‘geminates’ is used to describe a class of consonants that are produced with longer constriction duration than their short counterparts. Geminate consonants are therefore often defined phonologically as being long since the phonetic signature most commonly associated with them is increased duration relative to their singleton counterparts (e.g., Hankamer, Lahiri, & Koreman, 1989). It is well established across languages that the singleton/geminate contrast, either word-initially or word-medially, is distinguished primarily by a difference in consonant closure duration. Table 1 shows a list of 39 languages for which previous experimental investigation has confirmed closure duration to be a reliable and consistent acoustic parameter of the singleton/geminate contrast. It is worth noting that all the languages reported here exhibit longer closure duration for geminates relative to singletons regardless of position in a word. Crucially, however, only a handful of languages surveyed here contrast singletons and geminates in word-initial position, and, of these, only PM, like KM, has geminates in word-initial position only. In PM, for instance, word-initial geminates, across contexts, have an average closure duration of 186 ms, while their corresponding singletons have an average closure duration of only 66 ms.2 Further, the magnitude of contrast between word-initial singletons and geminates in PM is consistent across utterance-initial and utterance-medial positions, although the durational contrast is smaller in the former than in the latter.3 In any case, the word-initial singleton/geminate contrast in PM is clearly marked via closure duration with considerably longer duration for geminate consonants compared to singletons.
In other languages with word-medial geminates, a similar pattern can also be observed. In Malayalam, for example, the mean closure durations are 52 ms for singleton consonants and 175 ms for geminate consonants (Local & Simpson, 1999), in line with findings for many languages including for Turkish and Bengali (Lahiri & Hankamer, 1988). A number of studies have also established that closure duration is maintained across different speaking rates so that the consonant contrast is preserved, such as in Tarifit Berber (Bouarourou et al., 2011), Egyptian Arabic (Kawahara, 2005) and Cypriot Greek (Muller, 2001). With regard to utterance-initial voiceless stop contrasts where there is no acoustic cue to the onset of occlusion in word-initial position, experimental articulatory research using electropalatography (e.g., Kraehenmann & Lahiri, 2008; Ridouane, 2010) has confirmed reliable duration differences in the closure phase of the word-initial singleton/geminate contrast for voiceless stops. That is to say, articulatorily, closure duration differences also contribute to the singleton/geminate contrast in the case of utterance-initial voiceless stops.
The singleton/geminate duration ratio is particularly relevant in the context of geminate production and perception given that increased duration ratios are usually associated with geminates that are perceptually challenging, such as word-initial geminates. The cross-linguistic diversity of geminate-to-singleton duration ratios is exemplified in Table 2 (rightmost column). The duration ratio results presented in the table are of course indicative only, given the significant methodological differences across the listed studies, i.e., there are differences in terms of, among others, segments, utterance context, number of repetitions and speakers employed in each study reviewed (see columns 3–5 in Table 2). Notwithstanding these caveats, it can be seen in Table 2 that there is a wide range of duration ratios across these languages, ranging from 1.25 (Cypriot Greek, word-initial) to 3.40 (Malayalam, word-medial). Researchers have put forward different proposals to explain why the ratios of geminate-to-singleton durations vary so much across languages. First, Engstrand and Krull (1994) claim that duration contrasts are maintained more consistently in quantity-based languages (e.g., Finnish and Estonian), i.e., languages in which durations are primarily used to distinguish word meaning, as Lehiste (1970) points out (see also Šimko, O׳Dell, and Vainio (2014); Yoshida, de Jong, Kruschke, and Päiviö (2015)). Ham (2001) further reasons that duration ratios between singletons and geminates are potentially larger for mora-timed languages than for syllable-timed languages. More recently, Pajak (2009) reasons that the variation in duration ratios may be perceptually driven, i.e., increased duration ratios may probably reflect an attempt to compensate for otherwise lesser perceptibility of word-initial geminates, as suggested by, e.g., Kawahara (2007).
Based on these earlier studies of a range of different languages (i.e., Engstrand & Krull, 1994; Ham, 2001; Pajak, 2009), those showing relatively higher duration ratios in Table 2 (like PM, Malayalam, Turkish and Ingrian) have phonological quantity contrasts (Engstrand & Krull, 1994) or have been analyzed as mora-timed (Ham, 2001). In addition, in light of Pajak׳s (2009) proposal, it could also be hypothesized that these languages may possess perceptually-challenging geminates in their inventories. The case of PM is of critical importance here as this language, closely related to KM, only allows word-initial geminates which, as noted earlier, are more phonologically dispreferred than word-medial geminates (cf. Ladefoged & Maddieson, 1996). Further, Goedemans (1998) also argues that an extremely high duration ratio is essential for the length distinction in syllable onset position to be audible from a perceptual perspective.4 It should also be noted, however, that duration ratios are also extremely high for some languages with word-medial geminates such as Malayalam (Local & Simpson, 1999) and Turkish (Lahiri & Hankamer, 1988), although only a limited range of contrasts were explored in these studies compared to the current investigation of KM (Section 2.0 will describe the research design in KM in detail).
As noted earlier in Section 1.0, the magnitude of durational contrast due to gemination may be conditioned by other factors such as utterance position (i.e., utterance-initial or -medial positions), manner of articulation (e.g., stops or sonorants), voicing status (i.e., voiceless or voiced stops) and also phoneme type within each manner category. In many languages, consonants of all manners and voicing types can usually be geminated (Ladefoged & Maddieson, 1996). In Moroccan Arabic (Heath, 1987), for example, the entire segment inventory, including stops and sonorants, may be geminated although stops are considered as the most common type of geminate across languages (see, e.g., Aoyama & Reid, 2006; Kochetov, Sreedevi, Kasim, & Manjula, 2014).
Within stops, variation in consonant gemination due to voicing status has been reported in many studies (e.g., Engstrand & Krull, 1994). Voiceless stop geminates are of particular interest, especially with regard to word-initial consonant contrasts (see, e.g., Abramson, 1998). As noted earlier, the lack of clear acoustic marking of initial occlusion of voiceless stops in utterance-initial position has led many researchers (e.g., Ridouane, 2007) to claim that secondary cues, beyond closure duration, may contribute to the contrast in this context. As for voiced stop geminates, they occur less frequently than voiceless stop geminates even in medial position (see, e.g., Kawahara, 2005 for Japanese). It has also been claimed that voiced stop geminates are cross-linguistically more marked than voiceless stop geminates. For example, Jaeger (1978), and Ohala and Riordan (1979) propose that this is due to the aerodynamic difficulty in sustaining voicing within obstruent closure. That is, the longer prevoicing that is required during the initial closure of word-initial voiced stop geminates in order to maintain the contrast with their singleton counterparts is more effortful than in other segments, e.g., sonorants. In many cases, voicing during the closure phase of voiced stop geminates is reported to be weakened and absent (e.g., Taylor, 1985).
With respect to sonorant geminates, Payne (2005) argues that there is an “in-built relationship” between singleton and geminate sonorants, calling sonorants particularly “successful geminates”. As shown in Table 2 earlier, word-medial sonorants in Malayalam (Local & Simpson, 1999) display the highest duration ratio (i.e., 3.40) among other languages listed. Further, a number of studies (e.g., Khattab, 2007 for Lebanese Arabic) have also shown that geminate-to-singleton duration ratios are always higher for sonorants than for other manners of articulation. By contrast, some studies (e.g., Kawahara, 2013 for Japanese) have shown that the singleton/geminate duration ratios are much smaller for nasals than for stops. Variation in consonant gemination has also been observed for word-medial fricatives in Japanese (Kawahara, 2015) for which the singleton/geminate duration ratios are usually smaller for fricatives than for stops. Crucially for the current study of KM, it remains to be seen exactly how this contrast is realized acoustically in consonants that vary in manner and phoneme type.
Besides the consonant variables of manner/place and voicing, utterance position is also another potential conditioning factor in the realization of the word-initial singleton/geminate contrast. There is a well-known association between utterance position and consonant contrast enhancement which is often referred to in the literature as domain-initial strengthening (see, e.g., Cho, 2001; Cho & Keating, 2001; Fougeron & Keating, 1997; Keating, Cho, Fougeron, & Hsu, 2003, among others). Cho and Jun (2000) suggest two types of domain-initial strengthening effects: (1) syntagmatic contrast enhancement and (2) paradigmatic contrast enhancement. The former refers to the contrast between the initial segment and the following vowel, i.e., the consonant-vowel (CV) contrast at the beginning of a prosodic domain, e.g., an intonational phrase, while the latter deals with the enhancement of a contrastive phonemic distinction such as the singleton/geminate consonant contrast which is the main focus of the current study of KM. Under this account, it is hypothesized that both singleton and geminate consonants for voiced stops and sonorants, for example, can be expected to undergo domain-initial syntagmatic CV contrast enhancement by way of durational lengthening, making these initial segments more distinct from the following vowel. We might also argue that, in utterance-initial position, the KM word-initial singleton/geminate contrast for voiced stops and sonorants, for which there are clear cues to the occlusion phase during voicing/resonance, may also be paradigmatically enhanced in this context as compared to that in utterance-medial position. However, in a related study, Cho and Keating (2009) claim that a consonant contrast may also become less distinctive at higher prosodic domains, suggesting that paradigmatic contrast is not necessarily enhanced domain-initially, but rather it becomes “blurred” at the expense of the syntagmatic CV enhancement. With regard to geminates, Kraehenmann and Lahiri (2008) report that the difference between word-initial singleton and geminate stops in Swiss German weakens rather than strengthens in domain-initial position. They suggest that consonant articulation is strengthened and lengthened within both categories, but the contrast between them is not enhanced. This suggests enhancement of the CV contrast is paramount if there is an equal degree of strengthening and lengthening of the initial singleton and geminate consonant in domain-initial position at higher prosodic levels.
As outlined in the literature review presented above, closure duration is considered to be a universal acoustic parameter for consonant gemination. Our first research question pertains to the extent of this acoustic feature in the context of the word-initial singleton/geminate contrast in KM:
- 1.
How robust is closure duration in defining the word-initial singleton/geminate contrast in KM?
It is expected that closure duration will be a highly reliable acoustic property for the consonant contrast in KM, as also observed in many languages with consonant gemination (see Section 1.2), including the closely related language PM. With regard to a position in a prosodic domain (i.e., utterance-initial or utterance-medial positions) and its potential interaction with the singleton/geminate contrast, the second research question asks:
- 2.
To what extent does utterance position affect the closure duration for both singletons and geminates?
Given the fact that the singleton/geminate contrast in KM only occurs in word-initial position, we expect that the contrast will be enhanced in utterance-initial position, either syntagmatically or paradigmatically, as predicted by Cho and Jun (2000). For example, if both singletons and geminates are lengthened in utterance-initial position compared to utterance-medial position, we suggest that syntagmatic contrast enhancement may be in operation here. However, if the durational contrast between the two categories is stronger in utterance-initial versus utterance-medial contexts, then we may be dealing with paradigmatic contrast enhancement. There is of course the third option that both paradigmatic and syntagmatic contrast enhancements may be observed if we see lengthening in both singletons and geminates in utterance-initial position but greater lengthening in geminates, for example.
Finally, it is important to investigate the extent to which segmental factors are important in determining closure duration in the realization of the KM word-initial singleton/geminate contrast. As noted in Section 1.4, geminate consonants vary according to manner of articulation, voicing and phoneme type. Therefore, it is asked:
- 3.
To what extent do manner of articulation (i.e., stops or sonorants), stop voicing distinctions (i.e., voiceless or voiced stops) and phoneme influence the closure duration differences between word-initial singletons and their geminate counterparts in KM?
It is predicted that there will be variation as a result of these segmental and prosodic factors, although we also hypothesize that closure duration will be a consistent and robust acoustic marker for the KM word-initial consonant contrast. We are aware also that concomitant acoustic and articulatory cues are potentially relevant here in the realization of the singleton/geminate contrast. That is, the importance of a single acoustic parameter should ideally be determined through assessing its importance relative to other concomitant acoustic parameters. Nevertheless, this study will make two important contributions to phonetic studies of gemination: (1) It will assess the “reliability” of closure duration as a proxy for the defining property of the contrast regardless of other conditioning factors as mentioned in the second and third research questions above; and (2) it will serve as a strong foundation for further experimental work in KM with regard to other acoustic and articulatory parameters alongside closure duration in defining the word-initial singleton/geminate contrast in this Malay variety.
Section snippets
Materials and speakers
An acoustic phonetic experiment was designed to investigate the robustness of closure duration in defining the word-initial singleton/geminate contrast in KM. A list of thirty-eight tokens was prepared consisting of nineteen minimal pairs (shown in Table 3).
All tokens were disyllabic words with either C(C)VCV or C(C)VCVC structures. Twenty phonemes were chosen and they were grouped according to voicing profile and manner of articulation: voiceless stops (/p/–/pp/, /t/–/tt/, /k/–/kk/); voiced
Overall results
The overall closure duration results for singletons and geminates are demonstrated in Fig. 2 and the mean G/S ratio is summarized in Table 4. Note, however, that the data is unbalanced across the two positions since utterance-initial voiceless stop tokens were excluded, given that acoustic closure duration measurement for these was not possible. The boxplots in Fig. 2 show a combined simplified distribution and outliers for closure duration for voiceless stops (utterance-medial position only),
Discussion
In this study, we have presented the results of an acoustic analysis of the typologically rare word-initial consonant contrast in KM with special regard to a durational parameter deemed reliable across languages, i.e., closure duration. The first research question laid out in Section 1.5 asked the degree to which closure duration primarily defines the word-initial contrast in KM. The overall acoustic results presented in Section 3.1 show that the KM singleton/geminate contrast is indeed
Conclusions
This study has documented the durational characteristics of relatively rare word-initial geminates in an understudied language, Kelantan Malay. More specifically, it has explored the production of the word-initial singleton/geminate contrast in KM in order to evaluate the contribution of closure duration to such a length distinction in this language. Taken together all the closure duration results in KM, it appears that they point to a general trend of high duration ratio values across
Acknowledgments
This study was supported by the Skim Latihan Akademik Bumiputra (SLAB) under the Ministry of Education Malaysia and Universiti Teknologi Malaysia awarded to the first author. Many thanks go to the staff of the University of Melbourne for their generous assistance during the recording of the Kelantanese speakers in Melbourne. We are also grateful to the management of Universiti Malaysia Kelantan for offering access to their undergraduate students who made the task of data collection enjoyable
References (84)
Complex prosodic focus marking in Finnish: expanding the data landscape
Journal of Phonetics
(2016)- et al.
Mixed-effects modeling with crossed random effects for subjects and items
Journal of Memory and Language
(2008) - et al.
Articulatory and acoustic studies on domain-initial strengthening in Korean
Journal of Phonetics
(2001) - et al.
Effects of initial position versus prominence in English
Journal of Phonetics
(2009) - et al.
Prosodic strengthening on the /s/-stop cluster and the phonetic implementation of an allophonic rule in English
Journal of Phonetics
(2014) - et al.
Perception of consonant length: Voiceless stops in Turkish and Bengali
Journal of Phonetics
(1989) Emphatic gemination in Japanese mimetic words: a wug-test with auditory stimuli
Language Science
(2013)- et al.
Spatial and dynamic aspects of retroflex production: an ultrasound and EMA study of Kannada geminate stops
Journal of Phonetics
(2014) - et al.
The timing of geminate consonants
Journal of Phonetics
(1988) Medial stop gemination in Rembarrnga: a spectrographic study
Journal of Phonetics
(1980)
Voice onset time in Spanish–English spontaneous code-switching
Journal of Phonetics
Emergent consonantal quantity contrast and context-dependence of gestural phasing
Journal of Phonetics
Cross-language similarity and difference in quantity categorization of Finnish and Japanese
Journal of Phonetics
The perception of word-initial consonant length: Pattani Malay
Journal of the International Phonetic Association
Word-initial consonant length in Pattani Malay
The complex acoustic output of a single articulatory gesture: Pattani Malay word-initial consonant length
Cross-linguistic tendencies and durational contrasts in geminate consonants: an examination of Guinaang Bontok geminates
Journal of the International Phonetic Association
On the articulatory bases of prominence in Italian
Onset mora dalam dialek Kelantan: Analisis tatatingkat kekangan
Jurnal e-Bangi
Word structure in Ngalakgan
Praat, a system for doing phonetics by computer
Glot International
Gemination in Tarifit Berber: Doing one or two things at once
Effects of prosody on articulation in English
Prosodic strengthening and featural enhancement: Evidence from acoustic and articulatory realisations of /a u/ in English
Journal of the Acoustical Society of America
Domain-initial strengthening as featural enhancement: aerodynamic evidence from Korean
Chicago Linguistics Society
The phonetic realization of singleton-geminate contrasts in three languages of Indonesia
Criteria for acoustic-phonetic segmentation and word labelling in the Australian National Database for Spoken Language
Introductory Statistics with R
The phonetics and phonology of San Martín Itunyoso Trique
Geminate typology and the perception of consonant duration
Durational correlates of quantity in Swedish, Finnish and Estonian: cross-language evidence for a theory of adaptive dispersion
Phonetica
The duration of some Norwegian speech sounds
Phonetica
Articulatory strengthening at edges of prosodic domains
Journal of the Acoustical Society of America
Weightless segments: a phonetic and phonological study concerning the metrical irrelevance of syllable onsets
Word-Initial geminates and stress in Pattani Malay
The Linguistic Review
Phonetic and phonological aspects of geminate timing
Durational properties of initial geminate consonants in Kelantan Malay
Durational correlates of word-initial voiceless geminate stops: the case of Kelantan Malay
A taste of prosody: possible effects of the word-initial singleton-geminate contrast on post-consonantal vowel duration in Kelantan Malay
An acoustic analysis of release burst amplitude in the Kelantan Malay singleton/geminate stop contrast
Amplitude and F0 as acoustic correlates of Kelantan Malay word-initial geminates
Cited by (11)
Not all geminates are created equal: Evidence from Maltese glottal consonants
2018, Journal of PhoneticsCitation Excerpt :It is generally accepted that long consonants, or geminates, are longer than short consonants, or singletons. This difference in duration is the primary distinction made in production and used in perception (for a recent overview, see Table 1 in Hamzah, Fletcher, & Hajek, 2016). However, there is considerable discussion regarding the status and origin of secondary cues that are non-durational (e.g., release burst amplitude in stops, see Hamzah, Fletcher, & Hajek, 2012).
The acoustics of word-initial and word-internal voiced stops in Somali
2023, Journal of the International Phonetic AssociationThe perception of rearticulated and single-articulated geminates in Polish
2022, Canadian Journal of LinguisticsNon-durational acoustic correlates of word-initial consonant gemination in Kelantan Malay: The potential roles of amplitude and f0
2020, Journal of the International Phonetic AssociationInitial laryngeal neutralisation in Tyrolean
2018, Phonology