The acoustics and perception of the word-initial ...

The acoustics and perception of the word-initial singleton/geminate contrast in Kelantan Malay

Hilmi Hamzah

PhD

2013

School of Languages and Linguistics The University of Melbourne

Produced on archival quality paper

i

Abstract This study takes up an acoustic phonetic investigation of the word-initial singleton/geminate contrast in Kelantan Malay (henceforth KM). Production and perception data were gathered to identify the acoustic properties that characterize the consonant contrast and to examine their potential perceptual relevance. The production experiments involved elicited materials from sixteen native speakers of KM, while the perception data were collected from thirty native listeners using natural and manipulated stimuli. The key motivation was to understand the acoustic and perceptual behaviour of the word-initial length distinction, given its unusual characteristics and the potential maintenance of the contrast in domain-initial environments. Results of the acoustic phonetic analyses examining durational correlates show that closure duration is the most robust acoustic feature of the singleton/geminate contrast in KM. Additionally, VOT duration is seen to differ significantly according to the two consonant categories. Post-consonantal vowel duration, however, signals consonant gemination in a specific context only, i.e., for the low central vowel /a/ in utterance-medial tokens. Non-durational acoustic correlates were also explored in a second experiment, focusing on several parameters: release burst amplitude; the RMS amplitude and F0 at the onset of the following vowel; and the relative values of amplitude and F0 across two syllables of disyllabic words. Results suggest that these parameters can play important secondary roles in conveying and enhancing the consonant contrast. There are also effects of manner of articulation, voicing type and utterance position on the magnitude of contrast between singletons and geminates with utterance-initial voiceless stops showing the greatest magnitude difference. The perceptual significance of some of the acoustic correlates was tested in a series of perception experiments. It was found that KM native listeners can identify correctly the natural stimuli of isolated tokens with word-initial singletons and geminates including voiceless stop tokens. Data of manipulated stimuli confirm the primary role of closure duration in geminate perception, while the RMS amplitude and F0 data show limited perceptual functions on their own, although the combined values of the two parameters do have some effect on the perception of the contrast, especially for the utterance-initial voiceless stop word-pairs. The conclusion is drawn that the KM singleton/geminate contrast can be associated with durational and non-durational acoustic correlates, lending support to the view that the consonant length distinction in word-initial position can be defined on the basis of a set of acoustic parameters alongside closure duration. This investigation represents the first major acoustic study of this linguistic phenomenon in KM, in which the results provide an empirical basis for an understudied class of sounds, such as word-initial geminate consonants.

ii

Declaration This is to certify that: (i) (ii) (iii)

the thesis comprises only my original work towards the PhD, due acknowledgement has been made in the text to all other material used, the thesis is fewer than 100,000 words in length, exclusive of tables, maps, bibliographies and appendices.

Signed:

HILMI HAMZAH (2013)

iii

Acknowledgements I wish to thank my supervisors, Associate Professor Janet Fletcher and Professor John Hajek (and, during my first year of candidature, Dr Celia Thompson), for their continuous guidance and generous support during this research process. Many thanks go to the staff of the University of Melbourne for technical support. My sincere thank goes to the management of Universiti Malaysia Kelantan for offering me access to their undergraduate students who made the task of data collection enjoyable and rewarding. I would also like to acknowledge the help and understanding of my colleagues at the Phonetics Laboratory: Simone Graetzer, Rosie Billington, Tom Harris, Olga Maxwell, Hywel Stoakes, Josh Clothier, Amos Teo, Belinda Ross, Colleen Holt, Eleanor Lewis and Debbie Loakes. Finally, my special thank goes to my family in Malaysia for their patience and constant encouragement. I thank Universiti Teknologi Malaysia and the Government of Malaysia for financial support for all four years of this research undertaking. This study has been supported by the Skim Latihan Akademik Bumiputra (SLAB) under the Ministry of Higher Education, Malaysia.

iv

Table of contents Abstract …………………………………………………………………………..………………….…….…….......................... Declaration ………………………………………………………………….……………..…………….………………………. Acknowledgements …………………………………..………………..…….……………………….………………………. Table of contents ……………………………………………………………………………………………………............... List of figures …………………………………………………...……………………………….………………………............ List of tables ………………………………………………….…………………………………………………………………..

i ii iii iv viii xiv

1.

INTRODUCTION …………………………………..…………………………………………..……………………….... 1.1. Overview …………………………….…………….………………………………….………………………… 1.2. Geminates …………………………………………………………………………………………….………… 1.2.1. Some terminology associated with geminates ……….……..……………………... 1.2.2. Distribution of geminates across languages ………….……..……………………… 1.3. Kelantan Malay ……………………………………………………………..………………….….………….. 1.3.1. Background …………………………………………………………..…………………………... 1.3.2. Some phonological characteristics of Kelantan Malay …………………..……… 1.4. Aims of the study ……………………………………………………………..……………….…….……….. 1.5. Implications of the study …………………………………………………………………………………. 1.6. Structure of the study ………………………………………………………………………………………

1 1 4 4 6 8 8 10 12 14 15

2.

LITERATURE REVIEW ………………………………………………………………………………………............. 2.1. Introduction ……………………………………………………….………..…………………………………. 2.2. Durational correlates of geminates …………..…………………………………………………….... 2.2.1. Closure duration ………………………………………………………………………………... 2.2.2. VOT duration ……………………………………………………………………………………... 2.2.3. Vowel duration ………………………………………………………………………………….. 2.3. Non-durational correlates of geminates ……………….………………………..…………………. 2.3.1. Burst amplitude …………………………………………………………………………………. 2.3.2. Vowel amplitude ………………………………………………………………………………... 2.3.3. Vowel F0 …………………………………………………………………………………………… 2.3.4. Inter-syllabic differences ……………………………………………………………………. 2.4. Perception of geminates ………………………………………………………………………………….. 2.4.1. Geminate perception in natural settings …………………………….……………….. 2.4.2. Geminate perception in manipulated settings ……………..…….………………… 2.5. Some factors affecting consonant gemination …………………………………………………… 2.5.1 Manner of articulation and stop voicing ……………………………...………………. 2.5.2 Utterance position ………………………………………………………….…..……………… 2.6. Research questions …………………………………………………………………………………………. 2.6.1. General research questions ……………………………….………………..……………… 2.6.2. Specific research questions: Durational correlates ……………….……………... 2.6.3. Specific research questions: Non-durational correlates ……….………………. 2.6.4. Perception of geminates …………………………………………………….……………….

17 17 18 20 23 26 32 34 35 36 37 39 39 40 44 44 47 51 51 52 53 54

3.

METHODOLOGY ………………………………………………………………………………………………………… 56 3.1. Introduction ………………………………………………………………………………………………........ 56 v

3.2.

Part I: Production experiment ………………………………………………………………………….. 3.2.1. Materials …………………………………………………………………………………………… 3.2.2. Speakers ……………………………………………………………………………………………. 3.2.3. Data collection …………………………………………………………………………………… 3.2.4. Data analysis ……………………………………………………………………………………… 3.2.4.1. Annotation criteria ………………….….……..……………………………..... 3.2.4.2. Measurements of acoustic parameters …………………..…………..... 3.2.4.3. Statistical analyses ………………………………………...…………………... Part II: Perception experiments ……………………………………….………..……………………... 3.3.1. Stimulus materials ……………………………………………………………………………... 3.3.1.1. Experiment 1 ……………………………….…………..………………………... 3.3.1.2. Experiment 2 ……………………………………………………………………... 3.3.1.3. Experiment 3 ………………………….…………………..……………………... 3.3.1.4. Experiment 4 ……………………….……………………..……………………... 3.3.1.5. Experiment 5 …………………….…………………….….……………………... 3.3.2. Listeners …………………………………………………………………………………………… 3.3.3. Data collection …………………………………………………………………………………… 3.3.4. Data analysis ……………………………………………………………….…………………….. Summary …………………………………………………………………………………………………………

56 56 58 60 61 61 70 76 79 79 79 80 85 88 90 96 97 99 102

DURATIONAL CORRELATES ………………………………………………………………………………………. 4.1. Introduction ……………………………………………………………………………………………………. 4.1.1. Hypotheses ………………………………………………………………………………………... 4.1.1.1. Closure duration ………………….………...…………………………………... 4.1.1.2. VOT duration ……………………………………………………………………... 4.1.1.3. Vowel duration …………….………………………………...………………….. 4.2. Part I: Closure duration ………………………….………………………………...……………………… 4.2.1. Overall results …………………………………………………………………………………… 4.2.2. Closure duration: Utterance position effects ………………….……………………. 4.2.3. Closure duration: Manner of articulation/voicing effects ……..………………. 4.2.4. Closure duration: Phoneme-specific analysis ………………………………….…… 4.2.5. Summary of the closure duration results ……………………………...……….…….. 4.3. Part II: VOT duration ……………………………………………………………………………………….. 4.3.1. Overall results …………………………………………………………………………………… 4.3.2. VOT duration: Utterance position effects ……………………………..……………… 4.3.3. VOT duration: Place of articulation effects ……………………………...…………… 4.3.4. Burst duration in voiced stops ……………………………………………………………. 4.3.4.1. Overall results …………………………………………….………...…………… 4.3.4.2. Burst duration: Place of articulation effects ………………………… 4.3.5. Summary of the VOT and burst duration results …………….…………………… 4.4. Part III: Vowel duration …………………………………………………………………………………… 4.4.1. Overall results …………………………………………………………………………………… 4.4.2. Vowel duration: Utterance position and vowel height effects …….…………. 4.4.3. Vowel duration: Manner of articulation/voicing effects ……………..………… 4.4.4. Summary of the vowel duration results …………………….………………...………. 4.5. Discussion ……………………………………………………………………………………………………….

103 103 104 104 105 106 107 107 110 111 114 117 118 118 124 126 131 132 135 137 138 138 140 142 143 144

3.3.

3.4. 4.

vi

4.5.1. Key findings ………………………………………………………………………………………. 4.5.2. Closure duration ………………………………………………………………………………... 4.5.3. VOT duration ……………………………………………………………………………………... 4.5.4. Vowel duration ………………………………………………………………………………….. Conclusion …………………………………………………………………..……………………..……………

144 146 150 153 156

5.

NON-DURATIONAL CORRELATES ………………………………………………………………………………. 5.1. Introduction ……………………………………………………………………………………………………. 5.1.1. Hypotheses ………………………………………………………………………………………... 5.1.1.1. Burst amplitude ………………………..…….…………………………………. 5.1.1.2. Vowel onset amplitude and F0 ……………………………………............ 5.1.1.3. Inter-syllabic differences ……………………..….…………………………. 5.2. Part I: Burst amplitude ……………………………………………………....……………………………. 5.2.1. Overall results …………………………………………………………………………………… 5.2.2. Burst amplitude: Utterance position effects …………….………………….……….. 5.2.3. Burst amplitude: Place of articulation effects …………..…………….…………….. 5.2.4. Summary of the burst amplitude results ……………….…………………………..... 5.3. Part II: Vowel onset amplitude ……………………………………………...…………………….…… 5.3.1. Overall results …………………………………………………………………………………… 5.3.2. Vowel onset amplitude: Utterance position effects ……….……………………... 5.3.3. Vowel onset amplitude: Manner of articulation/voicing effects ….………… 5.3.4. Summary of the vowel onset amplitude results …………………………………… 5.4. Part III: Vowel onset F0 …………………………………………………………………………………… 5.4.1. Overall results …………………………………………………………………………………… 5.4.2. Vowel onset F0: Utterance position effects ………………………….……….……… 5.4.3. Vowel onset F0: Manner of articulation/voicing effects …………..…………… 5.4.4. Summary of the vowel onset F0 results ……………………………………………..... 5.5. Part IV: Inter-syllabic differences …………………………………………………………………….. 5.5.1. Overall results …………………………………………………………………………………… 5.5.2. Inter-syllabic differences: Utterance position effects …………………..……….. 5.5.3. Inter-syllabic differences: Manner of articulation/voicing effects ….……... 5.5.4. Summary of the inter-syllabic differences results ………………………………... 5.6. Discussion ………………………………………………………………………………………………………. 5.6.1. Key findings ………………………………………………………………………………………. 5.6.2. Burst amplitude …………………………………………………………………………………. 5.6.3. Vowel onset amplitude and F0 ……………………………………………………………. 5.6.4. Inter-syllabic differences ……………………………………………………………………. 5.7. Conclusion ………………………………………………………………………………………………………

157 157 158 158 159 161 161 161 167 168 173 174 174 177 179 181 182 182 187 188 191 192 192 198 200 203 204 204 206 208 212 213

6.

PERCEPTION OF GEMINATES …………………………………………………………………………………….. 6.1. Introduction ……………………………………………………………………………………………………. 6.1.1. Hypotheses ………………………………………………………………………………………... 6.2. Experiment 1: Control test of isolated tokens ……………………...……………………………. 6.2.1. Overall results …………………………………………………………………………………… 6.2.2. Factors affecting the identification rates of natural stimuli …………..………. 6.2.3. Summary of Experiment 1 …………………………………………………………………..

215 215 216 218 218 220 224

4.6.

vii

6.3.

Experiment 2: Closure duration in utterance-medial contexts …………………….…….. 6.3.1. Lengthened singletons vs. shortened geminates ………….…………………….… 6.3.2. Summary of Experiment 2 ………………………………………………………………….. Experiment 3: Vowel onset amplitude in utterance-initial tokens ……………………… 6.4.1. Voiceless stop /p/-/pp/ ………………………………………………….………………….. 6.4.2. Voiced stop /g/-/gg/ and sonorant /m/-/mm/ …………………………………… 6.4.3. Summary of Experiment 3 ………………………………………………………………….. Experiment 4: Vowel onset F0 in utterance-initial tokens …………………………………. 6.5.1. Voiceless stop /p/-/pp/ …………………………………………………….……………….. 6.5.2. Voiced stop /g/-/gg/ and sonorant /m/-/mm/ …………………………………… 6.5.3. Summary of Experiment 4 ………………………………………………………………….. Experiment 5: Combined amplitude and F0 in utterance-initial tokens ……………… 6.6.1. Voiceless stop /t/-/tt/ …………………………………………………………...….……….. 6.6.2. Voiced stop /d/-/dd/ …………………………………………………………………………. 6.6.3. Sonorant /n/-/nn/ ………………………………………………………………....………….. 6.6.4. Summary of Experiment 5 ………………………………………………………………….. Discussion ………………………………………………………………………………………………………. 6.7.1. Key findings ………………………………………………………………………………………. 6.7.2. Geminate perception in natural stimuli ……………………………….……………… 6.7.3. Geminate perception in manipulated stimuli ………………………….…………… 6.7.3.1. Closure duration as the primary cue to KM geminates …………. 6.7.3.2. Amplitude and F0 as secondary cues to KM geminates ………… Conclusion ………………………………………………………………………………………………………

225 227 233 234 236 238 241 242 244 246 249 250 253 257 260 264 265 265 268 271 271 274 278

GENERAL DISCUSSIONS AND CONCLUSIONS ……………………………………………………………… 7.1. Introduction ……………………………………………………………………………………………………. 7.2. The primary and secondary acoustic properties of the Kelantan Malay wordinitial consonant contrast ………...………………………………………………….……….…..……… 7.2.1. Durational aspects of Kelantan Malay geminates ……………..………………….. 7.2.2. Non-durational aspects of Kelantan Malay geminates ……….………….……… 7.3. Assessing factors that affect the magnitude of contrast between singletons and geminates …………………………………………………………………………………………….…………. 7.3.1. The effects of manner of articulation and voicing status ………..……………... 7.3.2. The effect of utterance position …………………………………………………..……… 7.4. The perceptual weight of acoustic cues associated with the Kelantan Malay word-initial consonant contrast ……………………………………………………………………….. 7.5. Potential fortitional process in the Kelantan Malay word-initial consonant contrast ………………………………………………………………………………………………………….. 7.6. Closing remarks ………………………………………………………………………………………….……

279 279

6.4.

6.5.

6.6.

6.7.

6.8. 7.

280 282 289 295 295 302 307 314 317

BIBLIOGRAPHY …………………...……………………………………………………………………………………………. 319 APPENDICES …………………………………………….…………………………...……………………………………......... 338

viii

List of figures Figure 1.1. Figure 1.2. Figure 2.1. Figure 3.1.

Figure 3.2.

Figure 3.3.

Figure 3.4.

Figure 3.5.

Figure 3.6.

Figure 3.7.

The distribution of Malay speakers in contemporary Southeast Asia (from Collins, 1998, p. 84) …………………………………………………………………………………….. The state of Kelantan in the Malay Peninsula, Malaysia (left panel; retrieved from http://en.wikivisual.com/index.php/Malaya) and the main districts in Kelantan (right panel; retrieved from http://www.dromoz.com) ………………….. An example of a model of a prosodic hierarchy (from Keating et al., 2003; after Beckman & Pierrehumbert, 1986) …………………………..…………………………… Waveforms and spectrograms of the initial syllables of the disyllabic wordpair /kabo/ “blurry” (left panel) versus /kkabo/ “beetle” (right panel) produced in utterance-initial (upper panel) and -medial (lower panel) positions by a KM male speaker, MS7. The blue vertical lines indicate the intervals of (1) closure phases for the word-initial singleton [k] and geminate [k:], (2) VOT (marked as [h]) and (3) the following vowel [a]. The utteranceinitial closure phases for [k] and [k:] are only arbitrary and therefore labeled in parentheses …………………………………………………………………………………….... Waveforms and spectrograms of the initial syllables of the disyllabic wordpair /gaʝi/ “salary” (left panel) versus /ggaʝi/ “sawing tool” (right panel) produced in utterance-initial (upper panel) and -medial (lower panel) positions by a KM male speaker, MS8. The blue vertical lines indicate the intervals of (1) closure phases for the word-initial singleton [g] and geminate [g:], (2) release burst (marked as [burst]) and (3) the following vowel [a] …….. Waveforms and spectrograms of the initial syllables of the disyllabic wordpair /nikɔh/ “marriage” (left panel) versus /nnikɔh/ “married” (right panel) produced in utterance-initial (upper panel) and -medial (lower panel) positions by a KM male speaker, MS8. The blue vertical lines indicate the intervals of (1) closure phases for the word-initial singleton [n] and geminate [n:] and (2) the following vowel [i] ………………………………………………………………. Illustration of annotated waveform and spectrogram in the Praat TextGrid for the male speaker, MS8, /ppitu/ “at the door” in utterance-medial position, showing four major annotation tiers: (1) the word tier (top tier); (2) the syllable tier (second tier); (3) the phonemic tier (third tier); and (4) the phonetic tier (bottom tier) …………………………………………………………………………... Illustration of the hierarchy view in the EMU Graphical User Display for the male speaker, MS8, /ppitu/ “at the door”, showing four major tiers (i.e., word, syllable, phonemic and phonetic) and associated labels with hierarchical relationship specified ………………………………………………………………………………….. Waveforms and spectrograms of the initial syllables of the disyllabic wordpairs /kiɣi/ “left” versus /kkiɣi/ “to the left” (upper panel) and /gaʝi/ “salary” versus /ggaʝi/ “sawing tool” (lower panel) produced in utterance-initial position by MS7 and MS8, respectively. The red arrows indicate the points of burst amplitude measurements …………………………………………………………………… Illustration of the same distributions of energy between singletons (bottom line in blue) and geminates (top line in red). Full results of burst amplitude

ix

8 9 47

62

64

66

67

69

73

Figure 3.8. Figure 3.9.


Figure 3.12.

Figure 4.1.


Figure 4.4.

are presented in §5.2 …………………………………………………………………………………... Schematics of potential differences in the RMS amplitude and F0 values across two syllables, showing ratios in positive values (left) and negative values (right). ‘S1’ and ‘S2’ refer to the first and second syllables, respectively. Illustration of a perception experiment conducted in a Praat window, showing the response categories written in Standard Malay. The yellow boxes at the top panel consist of a word beginning with a singleton consonant /kabo/ ‘blurry’ (left) and a word beginning with a geminate consonant /kkabo/ ‘beetle’ (right). The box at the bottom panel is an instruction for listeners, meaning “click here to listen again” ……………………. Illustration of a response file extracted to a table file in Praat, showing the subject (first column), stimuli (second column) and responses (final column). Illustration of the calculation of correct identifications in Experiment 1 (natural stimuli) using Excel, showing the listeners (first column), the scores for the minimal pair /pitu/ vs. /ppitu/ (second and third columns) and the scores for the minimal pair /pagi/ vs. /ppagi/ (fourth and final columns). The scores are summed up and converted into percentages (bottom row) …………... Examples of responses curves taken from the results of Experiment 2 (manipulated closure duration) for the voiceless stop pair /kabo/ ‘blurry’ versus /kkabo/ ‘a beetle’, showing geminate responses to two series of stimuli, i.e., the stimuli made from the original words with the singleton /k/ (blue line) and the geminate /kk/ (red line). The original closure duration values are shown on the leftmost step of the duration continuum (for the singleton /k/) and on the rightmost step of the duration continuum (for the geminate /kk/). The horizontal line shows the crossover zone between singletons and geminates at 50%, while the vertical lines indicate the 50% crossover points between the two series of stimuli. Full results are presented in §6.3.1 in Chapter 6 …………………………………………………………………………………... Distribution of closure duration values (ms) for singletons (left) and geminates (right) across the entire corpus. Data are collapsed across voiceless stops (utterance-medial position only), voiced stops (all utterance positions) and sonorants (all utterance positions) ……………………………………….. Mean closure duration values (ms) for singletons and geminates according to utterance position: utterance-initial (left) and utterance-medial (right) positions. Voiceless stop tokens were excluded from the analysis …………………. Distribution of closure duration values (ms) for singletons and geminates according to manner of articulation/voicing and utterance position; voiceless stops (left), voiced stops (middle) and sonorants (right). Utterance-initial tokens are shown in the upper panel, while utterance-medial tokens in the lower panel. The closure duration for voiceless stops was measured in utterance-medial position only ……………………………………………………………………. Distribution of closure duration values (ms) for singletons and geminates according to phoneme type and utterance position. The phoneme pairs, starting from the leftmost column, are /p/-/pp/, /t/-/tt/, /k/-/kk/, /b/-/bb/, /d/-/dd/, /g/-/gg/, /m/-/mm/, /n/-/nn/, /l/-/ll/ and /ŋ/-/ŋŋ/. Here, /ng//nng/ represent /ŋ/-/ŋŋ/, while ‘S’ singletons and ‘G’ geminates. Utterancex

74 75

98 99

100

101

107 110

112

Figure 4.5. Figure 4.6. Figure 4.7.


Figure 4.11.

Figure 4.12.


Figure 5.4.

initial tokens are provided in the upper panel, while utterance-medial tokens in the lower panel. The closure durations for voiceless stop pairs (i.e., /p//pp/, /t/-/tt/ and /k/-/kk/) were measured in utterance-medial position only ……………………………………………………………………………………………………………. Mean VOT duration values (ms) for singletons and geminates according to stop type; voiceless stops (left) and voiced stops (right). Error bars represent standard deviations …………………………………………………………………………………….. Mean VOT values (ms) for singletons (blue lines) and geminates (red lines) according to utterance position for voiceless stops (left) and voiced stops (left) …………………………………………………………………………………………………………… Illustration of the effect of Place of Articulation on the mean VOT values (ms) of singletons (blue lines) and geminates (red lines) according to utterance position. Positive VOT values in voiceless stops are shown in the upper panel, while negative VOT values in voiced stops are shown in the lower panel ……..... Illustration of the presence of release bursts in voiced stop singletons (blue bars) and geminate (red bars), showing the percentage for each place of articulation in utterance-initial (left) and utterance-medial (right) positions … Distribution of burst duration values (ms) for voiced stop singletons (left) and geminates (right) pooled across three voiced stop series: /b/-/bb/, /d//dd/, /g/-/gg/. Data include utterance-initial and -medial tokens ………………… Distribution of burst duration values (ms) in voiced stop singletons and geminates for each stop place of articulation: /b/, /d/ and /g/. Utteranceinitial tokens are shown in the upper panel, while utterance-medial tokens in the lower panel …………………………………………………………………………………………… Mean vowel duration values (ms) following singletons and geminates according to utterance position and vowel height: singleton + /i/ (blue line); geminate + /i/ (red line); singleton + /a/ (green line); geminate + /a/ (purple line) …………………………………………………………………………………………………………… Distribution of vowel duration values (ms) following singletons and geminates according to manner of articulation/voicing: voiceless stops (top), voiced stops (bottom left) and sonorants (bottom right). The data were extracted from the low central vowel /a/ in utterance-medial tokens only ……. Spectra of release burst for singletons (blue lines) and geminates (red lines) averaged across the entire corpus: voiceless stops (left) and voiced stops (right) ………………………………………………………………………………………………………… Mean energy sum values (dB) of release bursts for singletons and geminates according to stop type [voiceless stops (left) and voiced stops (right)] and utterance position ………………………………………………………………………………………. Spectra of release burst for singletons (blue lines) and geminates (red lines) according to place of articulation: /p/ (left column), /t/ (middle column) and /k/ (right column). Utterance-initial tokens are shown in the upper panel, while utterance-medial tokens in the lower panel ………………………………………… Spectra of release burst for singletons (blue lines) and geminates (red lines) according to place of articulation: /b/ (left column), /d/ (middle column) and /g/ (right column). Utterance-initial tokens are shown in the upper panel, while utterance-medial tokens in the lower panel ………………………………………… xi

114 118 125

127 131 132

135

140

142 161 167

169

170

Figure 5.5.

Figure 5.6.

Figure 5.7.

Figure 5.8.

Figure 5.9.

Figure 5.10.

Figure 5.11.

Figure 5.12.

Figure 5.13.

Figure 5.14.

Distribution of RMS amplitude values (dB) at vowel onset after singletons (left) and geminates (right) across the entire corpus. Data are collapsed across manners of articulation/stop types (voiceless stops, voiced stops and sonorants) and utterance positions (utterance-initial and -medial positions) ... Mean RMS amplitude values (dB) at vowel onset after singletons and geminates according to utterance position: utterance-initial (left) and utterance-medial (right) positions. Data are collapsed across manners of articulation/stop types (i.e., voiceless stops, voiced stops and sonorants) ……... Distribution of RMS amplitude values (dB) at vowel onset after singletons and geminates according to manner of articulation/voicing: voiceless stops (left column), voiced stops (middle column) and sonorants (right column). Utterance-initial tokens are provided in the upper panel, while utterancemedial tokens in the lower panel …………………………………………………………………. Distribution of F0 values (Hz) at vowel onset after singletons and geminates across the entire corpus, showing results for male (left) and female (right) speakers. Data are collapsed across manners of articulation/voicing types (voiceless stops, voiced stops and sonorants) and utterance positions (utterance-initial and -medial positions) ……………………………………………………… Mean F0 values (Hz) at vowel onset after singletons and geminates according to utterance position as produced by male speakers (left) and female speakers (right). Data are collapsed across manners of articulation/stop types (i.e., voiceless stops, voiced stops and sonorants) ………………………………………… Distribution of F0 values (Hz) at vowel onset after singletons and geminates produced by male speakers according to manner of articulation/voicing: voiceless stops (left column), voiced stops (middle column) and sonorants (right column). Utterance-initial tokens are provided in the upper panel, while utterance-medial tokens in the lower panel ………………………………………… Distribution of F0 values (Hz) at vowel onset after singletons and geminates produced by female speakers according to manner of articulation/voicing: voiceless stops (left column), voiced stops (middle column) and sonorants (right column). Utterance-initial tokens are provided in the upper panel, while utterance-medial tokens in the lower panel ………………………………………… Distribution of ratios of the first and second syllables of disyllabic words beginning with singletons and geminates averaged across the entire corpus: amplitude ratio (left) and F0 ratio (right). Data are collapsed across manners of articulation/voicing types (voiceless stops, voiced stops and sonorants) and utterance positions (utterance-initial and -medial positions) …………………. Mean ratios of the first and second syllables of disyllabic words beginning with singletons and geminates according to utterance position. Amplitude ratio is shown on the left, while F0 ratio on the right. Data are collapsed across manners of articulation/stop types (i.e., voiceless stops, voiced stops and sonorants) ……………………………………………………………………………………………. Distribution of amplitude ratios of the first and second syllables of disyllabic words beginning with singletons and geminates according to manner of articulation/voicing: voiceless stops (left column), voiced stops (middle column) and sonorants (right column). Utterance-initial tokens are provided xii

174

178

179

182

187

189

190

192

199

Figure 5.15.

Figure 6.1.

Figure 6.2.

Figure 6.3.

Figure 6.4.

Figure 6.5.

Figure 6.6.

in the upper panel, while utterance-medial tokens in the lower panel …………… Distribution of F0 ratios of the first and second syllables of disyllabic words beginning with singletons and geminates according to manner of articulation/voicing: voiceless stops (left column), voiced stops (middle column) and sonorants (right column). Utterance-initial tokens are provided in the upper panel, while utterance-medial tokens in the lower panel …………... Mean percentages of correct responses of the identification of natural stimuli of isolated tokens beginning with singletons (left) and geminates (right). Data are collapsed across manners of articulation/stop types (voiceless stops, voiced stops and sonorants) ………………………………………………………………………... Mean percentages of correct responses of the identification of natural stimuli of isolated tokens beginning with singletons and geminates according to manner of articulation/voicing: voiceless stops (left), voiced stops (middle) and sonorants (right) ………………………………………………………………………………….. Mean percentages of geminate responses to the manipulated stimuli (closure duration) for (a) the voiceless stop pair /kabo/-/kkabo/, (b) the voiced stop pair /bacɔ/-/bbacɔ/ and (c) the sonorant pair /lapu/-/llapu/. Geminates responses to different series of stimuli are indicated by line colour: blue lines (original singletons); red lines (original geminates). Original durations of manipulated stimuli are shown on the duration continua: first steps (lengthened singletons); final steps (shortened geminates). Horizontal lines show the crossover zones between singletons and geminates at 50%, while vertical lines indicate the 50% crossover points between the two series of stimuli ………………………………………………………………………………………………………... Mean percentages of geminate responses to the manipulated stimuli (amplitude at the onset of the following vowel) for the voiceless stop pair /pagi/-/ppagi/, showing six 2-dB steps of increased amplitude (dB) on the xaxis including the original singleton (leftmost step) and the original geminate (rightmost step). The blue line plots geminate responses to the stimuli created from the original singleton /p/ and the red line indicates geminate responses to the stimuli created from the original geminate /pp/. The horizontal line shows the crossover zone at 50%. The vertical lines indicate the 50% crossover points between the two types of stimuli ………………………….. Mean percentages of geminate responses to the manipulated stimuli (amplitude at the onset of the following vowel) for (a) the voiced stop pair /gaʝi/-/ggaʝi/ and (b) the sonorant pair /maɣi/-/mmaɣi/, showing six 2-dB steps of increased amplitude (dB) on the x-axes including original singletons (leftmost steps) and original geminates (rightmost steps). Blue lines plot geminate responses to the stimuli created from original singletons and red lines indicate geminate responses to the stimuli created from original geminates. Horizontal lines show the crossover zones at 50% ……………………… Mean percentages of geminate responses to the manipulated stimuli (F0 at the onset of the following vowel) for the voiceless stop pair /pitu/-/ppitu/, showing six half-semitone steps of increased F0 in semitones on the x-axis including the original singleton (leftmost step) and the original geminate (rightmost step). The blue line plots geminate responses to the stimuli xiii

201

202

218

220

228

236

239

Figure 6.7.

Figure 6.8.

Figure 6.9.

Figure 6.10.

created from the original singleton /p/ and the red line indicates geminate responses to the stimuli created from the original geminate /pp/. The horizontal line shows the crossover zone at 50% …………………………………………. Mean percentages of geminate responses to the manipulated stimuli (F0 at the onset of the following vowel) for (a) the voiced stop pair /gigi/-/ggigi/ and the sonorant pair /misa/-/mmisa/, showing six half-semitone steps of increased F0 in semitones on the x-axes including original singletons (leftmost steps) and original geminates (rightmost steps). Blue lines plot geminate responses to the stimuli created from original singletons and red lines indicate geminate responses to the stimuli created from original geminates. Horizontal lines show the crossover zones at 50% ……………………… Mean percentages of geminate responses to the manipulated stimuli (amplitude and F0 at the onset of the following vowel) for the voiceless stop pair /tido/-/ttido/, showing the stimuli created from the original /t/ (upper panel) and the original /tt/ (lower panel). The stimuli are rank-ordered on the x-axes as follows: lowest mean percentages (the beginning of the continua); highest mean percentages (the end of the continua). Horizontal lines show the crossover zones at 50% ………………………………………………………... Mean percentages of geminate responses to the manipulated stimuli (amplitude and F0 at the onset of the following vowel) for the voiced stop pair /dike/-/ddike/, showing the stimuli created from the original /d/ (upper panel) and the original /dd/ (lower panel). The stimuli are rank-ordered on the x-axes as follows: lowest mean percentages (the beginning of the continua); highest mean percentages (the end of the continua). Horizontal lines show the crossover zones at 50% ………………………………………………………... Mean percentages of geminate responses to the manipulated stimuli (amplitude and F0 at the onset of the following vowel) for the sonorant pair /nikɔh/-/nnikɔh/, showing the stimuli created from the original /n/ (upper panel) and the original /nn/ (lower panel). The stimuli are rank-ordered on the x-axes as follows: lowest mean percentages (the beginning of the continua); highest mean percentages (the end of the continua). Horizontal lines show the crossover zones at 50% ………………………………………………………...

xiv

244

247

254

258

261

List of tables Table 1.1. Table 1.2. Table 2.1.

Table 2.2.

Table 2.3.

Table 2.4.

Table 2.5.

Table 3.1. Table 3.2. Table 3.3. Table 3.4.

KM consonants (adapted from Che Kob, 1985; Karim, 1966) ………………………… KM vowels (after Che Kob, 1985; Mahmood, 2006) ………………………………………. A review of durational correlates associated with the singleton/geminate contrast in 40 languages with word-initial and/or word-medial consonant contrasts (expanded from Ridouane, 2010). The abbreviations are described as follows: ‘Wi’ word-initial position; ‘Wm’ word-medial position; ‘+’ the parameter varies significantly as a function of the consonant contrast (at least p CC’ singletons have greater values than geminates; ‘CC > C’ geminates have greater values than singletons) ………. Geminate-to-singleton duration ratios in 17 languages with word-initial and/or word-medial consonant contrasts ranked from the largest (top row) to the smallest ratio (bottom row). The duration ratio for Kelantan Malay is highlighted in grey (‘Wi’ word-initial position; ‘Wm’ word-medial position) ….. Rank of order [from the largest (top) to smallest (bottom) differences] between manners of articulation/stop types based on mean differences between singletons and geminates according to various acoustic parameters. All mean differences reach the p-values above .001. The mean differences for closure duration are expressed in terms of G/S ratios. All values are averaged across utterance positions, except vowel duration, i.e., /a/ in utterancemedial tokens only (‘VCL’ voiceless stops; ‘VCD’ voiced stops; ‘SON’ sonorants; ‘C > CC’ singletons have greater values than geminates; ‘CC > C’ geminates have greater values than singletons) ……………………………………………………………. The mean differences between singletons and geminates for utterance-initial voiceless stops within each place of articulation (i.e., bilabials, alveolars and velars) with the direction (Dir.) specified for each magnitude difference (‘C > CC’ singletons have greater values than geminates; ‘CC > C’ geminates have xxii

252

255

259

262 263

266

281

284

296

Table 7.5.

Table 7.6.

Table 7.7.

greater values than singletons). All the mean differences reach the p C’ geminates have greater values than singletons). All the mean differences reach the p CC’ singletons have greater values than geminates; ‘CC > C’ geminates have greater values than singletons) …………………………………………………………………... The trend of values from utterance-medial to utterance-initial positions for singletons and geminates and the degree of separation between the two consonant length categories in utterance-initial position (‘Enhanced’=the value is significantly enhanced (at least pF)

3.35 3.16 3.75 3.71 3.89 2.10 3.03 3.34 2.05 2.96 2.67 3.31 3.50 2.02 2.50 2.95

1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382 1,382

2252.7 1402.3 2824.7 2890.6 1797.5 469.87 1549.7 2059.1 982.26 1434.9 1257.2 2170.2 1535.1 1210.5 1361.5 1158.7