Harim Kwon’s research while affiliated with George Mason University and other places

Non-native consonant clusters are modified to conform with native phonotactics. This study investigates how onset clusters that are phonotactically licit, but have non-native phonetic patterns, are adapted to match the native patterns. We tested Georgian speakers in a sentence completion task using CCV/CVCV sequences produced by a French talker. Georgian differs from French in having (1) longer inter-consonant timing lag for CCVs, and (2) default initial prominence for CVCVs. The long inter-consonantal lag often results in a transitional vowel in Georgian. Georgian participants (n = 11) first saw the target CCV/CVCV sequences embedded in a Georgian carrier phrase and read the phrase aloud (baseline). Then they heard the target sequences produced by a French talker while seeing the Georgian carrier phrase with an empty slot, and produced the phrase completed with the heard targets (test). Participants' test productions reflected modifications of French targets towards their native phonetic patterns, not only by producing occasional transitional vowels that are absent from French CCV targets, but by deleting the unstressed first vowel in French CVCVs (/pøta/ produced as /pta/). We claim that the effects of native language on adaptation of non-native sequences are not limited to their segmental composition, but also involve their phonetic implementation.

This study examines the role of native inter-consonant timing patterns in perception and production of word-initial consonant clusters, asking how phonotactically native clusters with non-native timing patterns are perceived and produced by speakers of a cluster-heavy language. We tested Georgian speakers in two experiments using CCV/CVCV stimuli produced by a French talker. Georgian is a cluster-heavy language with a relatively long inter-consonant lag, often resulting in transitional vowels between two consonants within a cluster. French onset clusters have shorter inter-consonant lag than Georgian ones. In Experiment 1 (shadowing), Georgian participants (n = 14) were exposed to French CCV/CVCV stimuli and asked to produce what they heard. In Experiment 2 (discrimination), the same Georgian participants heard the same French stimuli in AX pairs and determined whether the two sequences (A and X) were the same or different. The results from the two experiments suggest that Georgian participants often confused French CCV (without any vocalic transition) with French CVCV. The confusion occurred almost exclusively when the first vowel of CVCV was /ø/, which is acoustically similar to the transitional vowel in Georgian, but did not seem to stem from Georgian speakers' incapability of producing French short-lag CCV. We claim that native inter-consonant timing patterns influence perception and production of consonant clusters even when the clusters are phonotactically licit in one’s native language.

Cross-language research on consonant cluster production has shown that consonant clusters in different languages are produced with different degrees of articulatory timing lags. The present study examines perceptual sensitivity to these cross-linguistic timing differences in consonant clusters. Native German listeners were tested on an AXB similarity judgment test using stimuli including consonant clusters produced by German and Georgian speakers. (German consonant clusters are produced with relatively shorter lag between two consonants than Georgian ones.) Stimuli were /bla, gla, gna/ syllables recorded along with articulatory (EMA) data. Short lag German tokens and long lag Georgian tokens were selected as A and B, with Xs of varying degrees of lag chosen from either Georgian or German recordings. Results showed that German listeners are sensitive to the cross-linguistic differences in articulatory timing lag: when the timing lag of X was closer to A, participants were more likely to choose A. Moreover, listeners’ sensitivity was influenced by the types of clusters: listeners were more sensitive in /bla/ than they were in /gla/ and in /gna/. The effects on the similarity judgment of different measures of articulatory lag, of vocalic releases produced within clusters, and of other sub-phonemic details were investigated. Overall, the results show that the lag differences are salient to German listeners while German and Georgian clusters can differ in a number of respects.