Phonological processing and encoding of tones in Mandarin Chinese

Abstract

The present study examines the perception and production of lexical tone in Mandarin Chinese. Lexical tone, which adds pitch contours to the sonorant part of a syllable, is a lexically contrastive phoneme similar to how consonants and vowels (also known as “segments”) distinguish between words in Indo-European languages. This study uses behavioral methods to test two aspects of tone in spoken language in an attempt to answer the question: are tones perceived and produced in a qualitatively different way from segments? The first part of the study looks at the status of tones in phonological processing. The first series of 4 experiments uses a lexical selection task to examine how willing Mandarin speakers were to change tones as opposed to segments. The results support previous research which claims that tones are less lexically binding than onsets and vowels; in the present study, participants were more likely to change a tone than to change an onset or a vowel. However, segments include not only onsets and vowels, but also codas, of which Mandarin has two: /n/ and /ŋ/. Given that there are only two items in the Mandarin coda category, they provide significantly less lexical information than onsets and vowels. Depending on the model, onsets and vowels are counted and categorized differently, with onsets being anywhere from 19 to 23 items, and vowels classically analyzed as anywhere from 5 to 7 distinct items, but up to 24 items when counting complex vowels, such as diphthongs and triphthongs. More importantly, the coda category also has fewer items than the tone category, which has 4 items. Experiments 2-3 showed that speakers were more willing to change codas than to change tones, which lends support to the idea that tones do not behave intrinsically differently from segments as a whole. Instead, tones behave in a way similarly to segments - ­they constrain word access in the same way as segments do, and the amount of that constraint is determined by how much information they provide. In the second part of the study, seeing as how tones don’t behave qualitatively differently than segments in perception, I turn to see whether tones behave like segments in production, using a tongue twister paradigm to examine the role of tones in speech production. Speech production involves a process called phonological encoding, the process by which a speaker builds the articulatory plan for an intended utterance. Certain phonemes can be viewed as having early or active phonological encoding, which shows that they play an active role in the speech preparation process. Speech errors have often been examined as a way to shed light on the role of different phonemes in speech production, and the general consensus is that if a type of phoneme is actively encoded, it would incur a substantial amount of errors in speech, due to the active mapping of a phoneme to its position in a lexical sequence. Previously, there has been conflicting research supporting both frequent tone error and infrequent tone error in natural speech. Frequent tone error lends support to the theory that tone is actively encoded in speech, encoded similarly to segments. Infrequent tone error supports the theory that tone is inactively processed, or processed later, as compared to segments. Overall, the present study provides a great deal of new data showing how tone behaves in the perception and production of Mandarin Chinese. Although tone is a suprasegmental phoneme, the results suggest that tones and segmental phonemes are perceived similarly and that tones and segments are encoded similarly once different classes of segmental phonemes are considered separately.