Hilmi Hamzah, Janet Fletcher, John Hajek School of Languages & Linguistics, The University of Melbourne, Australia mhhamzah@student.unimelb.edu.au, janetf@unimelb.edu.au, johnth@unimelb.edu.au Abstract This study investigates whether the amplitude of release bursts is a potential cue to the word-initial singleton/geminate contrast in Kelantan Malay (KM). The voiceless stop series (/p-pp/, /t-tt/, /k-kk/) were examined in two different utterance conditions. Results show that release bursts for geminate stops were produced with significantly greater energy than those for singleton stops, suggesting greater articulatory efforts to mark gemination. Although the effects of prosodic context and stop place of articulation were noticeable, significant amplitude differences were preserved across utterance contexts. The findings support the role of spectral parameters, in particular release burst amplitude, in enhancing the short/long consonant contrast in this Malay variety. Index Terms : geminate, voiceless stops, release burst amplitude, Malay 1. Introduction The singleton/geminate consonant contrast is usually defined in terms of length, i.e., geminate stops exhibit longer closure durations than their singleton counterparts (e.g., [1]). Studies on short versus long consonants, at least intervocalically, have shown, however, that rather than being restricted to timing differences alone, there are additional non-durational correlates to the stop length contrast which include release burst amplitude. For example, in Finnish [2] and Malayalam [3], which have a singleton/geminate contrast like KM, burst amplitude for geminate stops is significantly greater than that for singleton stops. One of the unusual typological features of KM is that geminates can only occur word-initially, and potentially utterance-initially. To this point, no work on burst amplitude has previously been conducted on KM or any variety of Malay with a short/long consonant contrast. In fact, the acoustic characteristics of word-initial geminates have not been subject to as much investigation as in medial position, especially in the case of burst amplitude in isolated tokens – one of the utterance contexts tested in this study. Previous investigation of burst characteristics in Pattani Malay [4], the Malay variety with which KM shares many phonological features, mainly focused on the amplitude of the first syllable of disyllabic words relative to the second syllable. The mean amplitude ratio of initial syllables beginning with geminate stops was found to be higher than that of initial syllables with singleton stop onsets. Given that stop closure for voiceless geminate stops cannot be a potential cue to the singleton/geminate contrast in utterance-initial tokens, we hypothesize that the amplitude of release burst may also serve as an important auxiliary cue to gemination in the same environment in KM. In a number of related studies, initial strengthening has been observed in higher-level prosodic domains. In Cheju Korean [5], for example, it has been found that the intensity level of release burst is higher in prosodically strong positions, i.e., utterance-initial position, than in medial contexts. We therefore consider the possibility that the singleton/geminate consonant contrast in KM may also be affected by similar conditioning factors. In other words, the singleton/geminate contrast is also heightened in utterance-initial position due to potential domain-initial strengthening. Previous findings for KM [6, 7, 8] have confirmed the roles of closure duration and VOT as sufficient acoustic markers for the short/long consonant contrast in utterance- medial contexts in KM, while post-consonantal vowel duration also shows some significant, albeit small, effects. By examining the spectral properties of release burst in greater detail, we extend these earlier results by determining whether the utterance-initial geminate articulation is (a) manifested in greater burst energy and (b) potentially associated with the singleton/geminate contrast in KM. These results will further increase our understanding of the word-initial stop length contrast in KM and possibly have important cross-linguistic implications for acoustic cues to consonant gemination. 2. Method 2.1. Materials We conducted an acoustic phonetic experiment to examine the possible role of release burst amplitude in the singleton/geminate stop contrast in KM. The corpus is presented in Table 1. Singleton Geminate Word Gloss Word Gloss /p/ /pitu/ door /ppitu/ at the door /pagi/ morning /ppagi/ early morning /t/ /tido/ sleep /ttido/ sleep by chance /tanɔh/ land /ttanɔh/ outside /k/ /kiɣi/ left /kkiɣi/ to the left /kabo/ blurry /kkabo/ a beetle Table 1. List of stimuli and their glosses. The KM data in this study contained twelve tokens composed of six minimal pairs opposing singletons to geminates in two vowel contexts: high front vowel /i/ and low central vowel /a/. The singletons and geminates represent three types of voiceless stops: /p, t, k/. All tokens were disyllabic words with either C(C)VCV or C(C)VCVC structures presented in two utterance conditions: utterance-initial and utterance-medial positions (see § 2.2 below). An Acoustic Analysis of Release Burst Amplitude in the Kelantan Malay Singleton/Geminate Stop Contrast ISSN 1039-0227  2012 ASSTA 3 -6 December 2012, Macquarie University, Sydney, Australia SST 2012 85