Radoslav Pavlík

SOME ASPECTS OF VOT IN ENGLISH AND SLOVAK NON-FINAL PLOSIVES: A PRELIMINARY STUDY

In: Philologia XVII. Zborník Pedagogickej fakulty Univerzity Komenského. Edited by E. Tkáčiková. Bratislava: Univerzita Komenského 2005, pp. 7 – 15. ISBN: 80-223-1913-9

Radoslav Pavlík

1

Some Aspects of VOT in English and Slovak Non-Final Plosives: A Preliminary Study

SOME ASPECTS OF VOT IN ENGLISH AND SLOVAK NON-FINAL PLOSIVES: A PRELIMINARY STUDY Radoslav Pavlík Abstract The paper discusses some phonetic differences between voice onset times in English and Slovak non-final plosives occurring in a compatible functional style. The preliminary study has shown that there are more VOT categories in English than in Slovak. It has also been observed that Slovak exhibits a direct correlation between the phonetic and phonological parameter of voicing, whereas English does not. INTRODUCTION It has long been noticed that, in some languages, there is a discrepancy between the phonological and phonetic parameter of voicing. That is to say, phonologically voiced plosives may, in some contexts, exhibit no vocal-fold vibration at the moment of the burst. Nevertheless, such plosives are still distinguishable from their voiceless counterparts. To explain this phenomenon, the terms fortis and lenis have been introduced, and the distinction between these voiced – voiceless pairs was believed to have been primarily a matter of muscular tension. Later, however, a series of studies carried out by Lisker and Abramson in the 60s and 70s showed that the difference between voice and voicelessness may still be measured acoustically through the so-called voice onset time (VOT). Voice onset time may be defined as the time of the onset of the vocal-fold vibration measured with reference to the beginning of the noise component of a plosive – the burst. When the vocal-fold activity begins prior to the release, the VOT is negative and it is called ‘voicing lead’. When the vocal-fold activity begins after the release, the VOT is said to be positive and it is called ‘voicing lag’ (Lisker & Abramson 1964: 387). The VOT parameter of English plosives is well documented in a number of studies, many of which deal with comparing cross-language VOT differences (Lisker, 1957; Lisker & Abramson, 1964, 1970; Pisoni, 1977; Cho & Ladefoged, 1999; inter alia). To date, no such comparison has been made with Slovak. This preliminary study is aimed at establishing the main VOT differences between English and Slovak non-final plosives occurring in a compatible functional style. 1. SUBJECTS Five radio announcers served as subjects in this study. Three of these were native speakers of English (RP accent) and the remaining two were native speakers of Slovak (Standard Slovak pronunciation). All of them were educated adult male speakers working in national radio stations, and none of them had any speech impediments. 2. SAMPLES Recordings of connected speech have been made from national radio stations (BBC and Rádio Slovensko) directly onto the harddisk of a computer. All samples have been digitized at the frequency rate of 22,150 Hz and the bit depth of 16 bits. The overall length of all recordings made for one language is about 10 minutes, which amounts to 20 minutes of connected speech altogether.

Radoslav Pavlík

2

Some Aspects of VOT in English and Slovak Non-Final Plosives: A Preliminary Study

3. PROCEDURE The VOTs of non-final plosive consonants (word-initial and word-medial plosives) found in a discourse of approximately ten minutes have been measured in milliseconds, and placed into several categories. The samples have been taken from national radio stations, and the same style of discourse for both languages (news-reporting) has been chosen. In order to achieve higher objectivity, both spectrographic and oscillographic visualizations have been used when segmenting the speech signal. The signal in the oscillograms has been segmented at zero crossings (whether positive or negative). The programs used for recording, editing, and analyzing samples were the SIL’s Speech Analyser 1.5, and the Sonic Foundry’s Sound Forge 4.0. For each English and Slovak plosive, ten tokens per each group (specified below) have been selected and analyzed for VOT. The data collected for both languages have been tabulated using basic statistical methods. Prior to the actual measuring of VOTs, these groups of plosives were selected: (a) English: 1. phonologically voiceless plosives – word-initial (aspirated, preceded by silence, followed by a vowel) – word-medial (unaspirated, intervocalic) 2. phonologically voiced plosives – word-initial (preceded by silence, followed by a vowel) – word-medial (intervocalic) (b) Slovak: 1. phonologically voiceless plosives – word-initial (preceded by silence, followed by a vowel) – word-medial (intervocalic) 2. phonologically voiced plosives – word-initial (preceded by silence, followed by a vowel) – word-medial (intervocalic) All phonetic symbols used for plosives come from the International Phonetic Alphabet. 4. RESULTS The following tables show VOT measurements for both individual plosives and groups of plosive consonants. VOT of English plosives in ms Phonologically voiced plosives Phonologically voiceless plosives Word-initial Word-medial Word-initial Word-medial aspirated unaspirated Mn. Md. Rg. Mn. Md. Rg. Mn. Md. Rg. Mn. Md. Rg. 54.9 49.5 49 23.2 22 15 69.1 60.5 65 50.6 50 27 65.6 68.5 77 42.1 43.5 31 10.6 10.5 8 -50 -47 40 14.2 14.5 11 -42.1 -43.5 42 21.7 22.5 17 -45.1 -43.5 32

/p/ /t/ /k/ /b/ /d/ // Table 1. The VOT of the individual English plosives in milliseconds (Mn. = Mean, Md. = Median, Rg. = Range).

Radoslav Pavlík

3

Some Aspects of VOT in English and Slovak Non-Final Plosives: A Preliminary Study

VOT of Slovak plosives in ms Phonologically voiced plosives Phonologically voiceless plosives Word-initial Word-medial Word-initial Word-medial Mn. Md. Rg. Mn. Md. Rg. Mn. Md. Rg. Mn. Md. Rg. 24.2 24.5 18 15.5 15 8 21.9 22.5 15 23.9 24 8 39.8 39.5 14 52.3 53.5 21 36.5 35.5 15 41.5 42.5 18 -61.1 -51.5 73 -55.6 -54 25 -58.6 -56.5 69 -45.9 -46.5 20 -51.6 -55 44 -43.7 -43.5 13 -60.6 -63.5 39 -45.5 -46 16

/p/ /t/ /c/ /k/ /b/ /d/ // // Table 2. The VOT of the individual Slovak plosives in milliseconds (Mn. = Mean, Md. = Median, Rg. = Range).

VOT of English plosives in ms VOT of Slovak plosives in ms / p, t, k / / p, t, c, k / / b, d, / / b, d, ,  / Initial Medial Initial asp. Medial unasp. Initial Medial Initial Medial Mean Mean Mean Mean Mean Mean Mean Mean 15.5 -45.733 63.2 38.633 -57.975 -47.675 30.6 33.3 Table 3. The mean values of VOTs of English and Slovak groups of plosives in milliseconds. The voice onset time has been divided, for practical purposes, into two main categories: (1) Negative VOT – voicing lead (2) Positive VOT – voicing lag Prior to placing the VOT measurements into individual categories, the duration difference limen (dDL) had to be determined for plosive consonants in order to draw the line between perceptually significant categories. This matter, however, is complicated in several respects. First of all, the duration of the individual types of plosive consonants exhibits immense variation. Secondly, it is a moot point whether the formant bends of the neighbouring vowels should be taken as part of the consonant duration (formant bends are not taken into account in this study). Finally, the difference limens and Weber ratios depend both on testing conditions and the method used (Stott, 1935; Henry, 1948; Ruhm et al., 1966). The reference duration for the dDL in this study has been determined as the highest duration value of the sum of voicing-lead (if any) and the following noise component (up to the beginning of the regular vibration of the following sound). The highest duration values were found for Slovak // and //, and they were of the order of 110 ms. Most plosives, however, fell significantly below this level. Nevertheless, the dDL based on this reference duration may be considered suitable for our purposes, since smaller dDL would break up the VOT continuum into too many categories, and the perceptual status of such categories would become more questionable. According to Henry (1948), the dDL for 110 ms reference duration is approximately 20 ms. The VOT values of the individual groups differing by about 20 or more ms have been placed into separate categories. If the difference was smaller than, or approximately, 20 ms, the groups have been merged into one category. One exception was made in the case of English word-initial /b,

Radoslav Pavlík

4

Some Aspects of VOT in English and Slovak Non-Final Plosives: A Preliminary Study

d, /, where the dDL was lowered to 15 ms. This was done because a dDL of 15 ms is more than enough to make the /b, d, / perceptually distinct from the rest of the categories. The application of the above method has yielded the following results: (1) Negative VOT: The vocal-fold vibration starts before the burst of a plosive. All Slovak phonologically voiced non-final plosives, and all English phonologically voiced intervocalic non-final plosives belong in this category. The mean VOT for this category is -50.5 ms.

Fig. 1. An example of a negative VOT. (An oscillogram and spectrogram of the Slovak word ‘desať’). The VOT of the plosive // is -89 ms. (2) Positive VOT: The vocal-fold vibration starts after the burst of a plosive. (2a) Early: The vibration starts early after the beginning of the burst. All English voiced word-initial plosives preceded by silence belong in this category. The mean VOT for this category is 15.5 ms.

Fig. 2. An example of an early positive VOT. (An oscillogram and spectrogram of the English word ‘game’). The VOT of the plosive // is 17 ms. (2b) Medium: The vocal-fold vibration starts after the beginning of the burst. All Slovak voiceless non-final plosives, and all English voiceless unaspirated plosives in nonfinal positions belong in this category. The mean VOT for this category is 34.2 ms.

Radoslav Pavlík

5

Some Aspects of VOT in English and Slovak Non-Final Plosives: A Preliminary Study

Fig. 3. An example of a medium positive VOT. (An oscillogram and spectrogram of the Slovak word ‘potom’). The VOT of the plosive /p/ is 32 ms. (2c) Late: The vocal-fold vibration starts late after the beginning of the burst. All English aspirated word-initial plosives in stressed positions belong in this category. The mean VOT for this category is 63.2 ms.

Fig. 4. An example of a late positive VOT. (An oscillogram and spectrogram of the English word ‘patch’). The VOT of the aspirated plosive /p/ is 78 ms. 5. DISCUSSION Examining the results of the above analysis points to several important facts. First of all, the number of VOT categories in English and Slovak is not the same. Secondly, there is a certain overlap of VOT categories in English and Slovak. Finally, the functional use of the VOT categories of the two languages differs. There are four distinct VOT categories in English. These are phonetically manifested as parameters of allophonic variations of English non-final plosives. Phonemically, however, these four categories are condensed into two. The phonological feature [+ voiced] is realized in speech by negative and positive early VOTs, and the feature [– voiced] is manifested by positive medium and positive late VOTs. The discrepancy between the phonetic and phonological parameter of voicing in English is evident in the case of the word-initial (phonologically) voiced plosives /b, d, /, where, at the moment of the burst, there is generally no voicing present at all. In Slovak, there are only two VOT categories – negative and positive medium. What is more important, there exists a direct one-to-one relationship between the phonetic and phonological parameter of voicing, that is to say, the presence of voice at the moment of the burst indicates voicing (both phonetically and phonologically), and the absence of voice at the moment of the burst indicates voicelessness (both phonetically and phonologically). Comparing the VOT data of Slovak with some other languages (especially the one-to-one correlation between the phonetic and phonological parameter of voicing) places the Slovak

Radoslav Pavlík

6

Some Aspects of VOT in English and Slovak Non-Final Plosives: A Preliminary Study

language in the same category as Dutch, Spanish, Hungarian, Tamil, Czech, and many other languages (Lisker & Abramson, 1957, 1964, 1970; Cho & Ladefoged, 1999; inter alia).

References Abramson, A. S. and Lisker, L. (1970). ‘Discriminability along the voicing continuum: Crosslanguage tests’, in B. Hála, M. Romportl and P. Janota (eds.) Proceedings of the 6th International Congress of Phonetic Sciences, Prague: Academia, pp. 569–573. Auzou, P., Özsancak, C., Morris, R. J., Jan, M., Eustache, F. and Hannequin, D. (2000). ‘Voice onset time in aphasia, apraxia of speech and dysarthria: a review’, Clinical Linguistics and Phonetics 14: 131–150. Baart, J. (1999). A Field Manual of Acoustic Phonetics, SIL International. Brady, S., and Darwin, C. J. (1978). Range effect in the perception of voicing. Journal of the Acoustical Society of America 63: 1556–1558. Brohn, M. et al. (1996). Sound Forge 4.0 User Manual. Sonic Foundry, Inc. Carney, A., Widin, G., and Viemeister, N. (1977). Noncategorical perception of stop consonants differing in VOT. Journal of the Acoustical Society of America 62: 961–970. Fry, D. B. (1979). The Physics of Speech. Cambridge University Press. Handbook of the International Phonetic Association (1999), Cambridge University Press. Henry, F. (1948). ‘Discrimination of the duration of a sound’, Journal of Experimental Psychology 38: 734–743. Cho, T. and Ladefoged, P. (1999). ‘Variation and universals in VOT: evidence from 18 languages’, Journal of Phonetics 27: 207–229. Isačenko, A. V. (1968). Spektrografická analýza slovenských hlások. Vydavateľstvo Slovenskej akadémie vied Bratislava. Keating, P. (1984). Phonetic and phonological representation of stop consonant voicing. Language 60: 286–318. Kráľ, Á. and Sabol, J. (1989). Fonetika a fonológia. Bratislava: Slovenské pedagogické nakladateľstvo . Laver, J. (1994). Principles of Phonetics. Cambridge University Press. Lehiste, I. (1970). Suprasegmentals. Cambridge: MIT Press. Lenhardt, J. (1981). Konfrontačná fonetika anglických a slovenských hlások. Bratislava: Filozofická fakulta UK. Lisker, L. and Abramson, A. (1964). ‘A Cross-Language Study of Voicing in Initial Stops: Acoustical Measurements’, Word 20: 384–422. Lisker, L. (1957). ‘Closure duration and the intervocalic voiced-voiceless distinction in English’, Language 33: 42–49. Lisker, L. and Abramson, A. (1970). ‘The voicing dimension: Some experiments in comparative phonetics’, in B. Hála, M. Romportl and P. Janota (eds.) Proceedings of the 6th International Congress of Phonetic Sciences, Prague: Academia, pp. 563–567. Pisoni, D. (1977). Identification and discrimination of the relative onset time of two component ones: implication for voicing perception of stops. Journal of the Acoustical Society of America 61: 1352–1361. Potter, R., Kopp, G. and Green, H. (1947). Visible Speech. New York: D. Van Nostrad Company, Inc. Ruhm, H. B. et al. (1966). ‘Differential sensitivity to duration of acoustic stimuli’, Journal of Speech and Hearing Research 9: 371–384. Sabol, J. and Zimmermann, J. (1986). Základy akustickej fonetiky, 3rd edition. Košice: Edičné stredisko UPJŠ.

Radoslav Pavlík

7

Some Aspects of VOT in English and Slovak Non-Final Plosives: A Preliminary Study

Stott, L. H. (1935). ‘Time-order errors in the discrimination of short tonal durations’, Journal of Experimental Psychology 18: 741–766. Vachek, J. (1973). Chapters from Phonology of Modern English. Praha: Státní pedagogické nakladatelství. Vachek, J. and Firbas, J. (1963). Zvukový rozbor současné angličtiny. Praha: Státní pedagogické nakladatelství. Súhrn V štúdii skúmame rozdiely medzi hodnotami VOT anglických a slovenských explozív v pozícii na začiatku a v strede slova. Predbežný výskum problematiky ukazuje, že angličtina má viac VOT kategórií ako slovenčina. Zároveň sme zistili, že v slovenčine existuje priamy vzťah medzi fonetickým a fonologickým parametrom znelosť – neznelosť. V angličtine, ako je známe, tento vzťah neplatí.