Contextual Variations in Trisyllabic Sequences of Thai Tones

Abstract

It is well known that tones assimilate in much the same way as consonants and vowels do. In tonal assimilation, the height and shape of a given tone is altered by adjacent tones. Earlier studies on tonal assimilation in Thai have used two-tone sequences with an intervening obstruent between the two tones of interest. In the present study, three-tone sequences were used with continuously voiced syllables throughout the utterance. Assimilatory effects were expected to be greater between F₀ contours of two successive syllables that are continuous in voicing across the syllable boundary. Stimuli consisted of 125 possible three-tone sequences of the five Thai tones, superimposed on monosyllabic words in a carrier sentence. All syllables were stressed in order to eliminate potentially confounding interactions between stress and tonal assimilation. Acoustic analysis revealed that perseverative effects were assimilatory in nature; anticipatory effects, however, were dissimilatory in nature. Perseverative effects appear to be independent of tonal categories, and restricted to contiguous tones. Perseverative effects extend through more than half the duration of the following tone. Both the height and shape of the tones are affected. Anticipatory effects, on the other hand, are limited to the high and rising tones, and extend through less than half the duration of the preceding tone.

Toward invariant functional representations of variable surface fundamental frequency contours: Synthesizing speech melody via model-based stochastic learning

“…Second, the current version of PENTAtrainer2 simulates only F 0 variations due to the normal target approximation process. It has not incorporated algorithms for simulating additional articulatory mechanisms, including, in particular, anticipatory raising (Gandour et al, 1994;Potisuk et al, 1997;Xu, 1999), post-low bouncing (Chen and Xu, 2006), consonantal perturbation (Silverman, 1986) and vowel intrinsic pitch (Whalen and Levitt, 1995). Of these, post-low bouncing has already been simulated in a separate study by adding an extra component added to qTA (Prom-on et al, 2012).…”

“…For example, particularly in short-short vowel combinations, when H tone was followed by tones that approach a relatively low F 0 , such as M, L or R, the synthesized contours are lower than the original. Since the same pitch targets can simulate H tone in other cases, this error could be attributed to the wellestablished phenomenon of anticipatory raising (Gandour et al, 1994;Potisuk et al, 1997). Also, consistent mismatches in the H-H sequence in both long-long and short-short vowel combinations, but not in other H-tone related cases, suggest that speakers may have slightly changed the pitch target for a second H tone by increasing either slope or strength.…”

Effects of Speaking Rate on Thai Tones

“…In each utterance, the first and third syllables were stressed, the latter being the reduplicated adverbial itself. For each reduplicative adverb, the tonal environment is constant, thus controlling for effects of tonal coarticulation [Gandour et al, 1994;Potisuk et al, 1997]. All five target words began with a sonorant.…”

Synthesis of Polysyllabic Sequences of Thai Tones Using a Generative Model of Fundamental Frequency Contours

“…The 125 sample words of 3 tone sequences were combinations of syllables in the first set: {/niaN/, /nòoN/, /nŴaN/, /nÓON/, /nǑON/}, together with syllable in the second set: {/yOO/, /yÈE/, /wâa/, /lÓO/, /lǐi/} and the third set: {/naaN/, /lÒOn/, /lâam/, /náam/, /lǎan/} (16) . All sample words begin with nasal, lateral or semivowel consonants, and the vowels for all 3 syllables were long vowels.…”

“…In Thai tones, Potisuk et al (16) showed that coarticulation effect restricted to continuous tones. In this experiment, we assume the interaction between adjacent tones in continuous speech should be the same patterns.…”