CRLMB welcomes postdoctoral fellow Hyekyung Hwang

Hyekyung Hwang is the recipient of our 2007/2008 CRLMB postdoctoral supplement

Pictured, left to right: L. White, H. Hwang and H. Goad

The Centre for Research on Language, Mind and Brain extends a warm welcome to Hyekyung Hwang, who joined us as a postdoctoral fellow in September 2007. Currently working with Drs. Heather Goad (Linguistics, McGill) and Lydia White (Linguistics, McGill), Hyekyung comes to the Centre after completing her Ph.D. in Linguistics at the University of Hawaii with Dr. Amy J. Schafer.

Hyekyung’s doctoral thesis investigated how prosody is perceived and used in the processing of spoken English sentences, with a particular focus on the disambiguation effects of prosodic phrasing in English native speakers (L1ers) and Korean second language learners of English (L2ers). The prosodic structures of English and Korean demonstrate similar phonetic and phonological properties for intonation phrases (IPhs), the highest level of prosodic phrases. However, differences between the two languages reside in the lower level of prosodic phrases: English intermediate phrases (ips) have more variable tonal patterns and stronger durational cues than Korean accentual phrases (APs), the Korean counterpart. Three on- and off-line sentence comprehension experiments tested ambiguous early/late closure utterances (e.g., Because Mike phoned, his mother is relieved vs. Because Mike phoned his mother, she’s relieved). The prosodic conditions varied the size of the prosodic boundaries surrounding the ambiguously attached noun phrase (e.g., his mother). The results indicated significant differences in the effects of ips between the two language groups despite a generally consistent role of prosodic structure in sentence processing.

Intermediate prosodic boundaries provided markedly weaker evidence for syntactic boundaries for Korean L2ers than for English L1ers, but the L2ers performed similarly to the L1ers with IPh boundaries. A subsequent discrimination experiment demonstrated that the differential effects of ips were not due to perceptual differences between L1ers and L2ers. Based on these results, Hyekyung argued that universal underlying processing principles exist that govern the role of prosodic phrasing. The different prosodic effects seem likely to stem from categorizing a phonetic pattern as an ip or forming prosody-syntax mappings, which sheds light on the Korean L2ers’ developmental stage in perceiving and processing English prosody. Taken together, Hyekyung’s dissertation contributes to the L2 literature, in which little is known with respect to the acquisition of prosody and the role of prosody in processing. It also provides further insights for L1 processing by discriminating among hypotheses for the use of prosody in sentence comprehension.

Hyekyung’s most recent work has focused on the effects in Korean sentences of subject phrase length on the on-line processing of a following dative NP in order to test the Implicit Prosody Hypothesis in Korean. The Implicit Prosody Hypothesis (Fodor 1998, 2002) claims that attachment preferences in silent reading are sensitive to a language's prosodic patterns and that the projected implicit prosody resembles explicit prosody, which is directly affected by phrase length. Although a number of studies have supported this hypothesis across languages—especially studies that have explored the effects of relative clause (RC) length on its preferred attachment—the length in the critical region could have non-prosodic effects that influence attachment. Furthermore, the evidence for the hypothesis to date has been weak in Korean. One production study and one self-paced reading study were designed to put the hypothesis to a more substantial test, using a case in which the length of the ambiguously parsed phrase did not vary and was unlikely to cause unintended effects on discourse complexity, non-phonological memory, or other factors; the results of the two experiments consistently showed on-line effects of phrase length in prosodic phrasing and parsing decisions, providing strong cross-linguistic support for the Implicit Prosody Hypothesis.

Hyekyung’s postdoctoral research will extend this previous work by investigating how variation in edge tones affects the processing of ambiguous structures in English with Korean L2ers and English L1ers. Many studies have demonstrated the effects of prosodic boundaries in the resolution of different types of syntactic ambiguities while tonal features of the prosodic boundaries in the stimuli were carefully controlled across prosodic conditions. Yet tonal information is also one component of natural sentence prosody; Schafer et al. (2000) demonstrated that variation in edge tones, pitch accents, or pitch range can ameliorate native syntactic parsing. Since English ips have more varied intonational tonal patterns than Korean APs, the choice of phrase tones might affect Korean L2ers’ processing of ips and their disambiguation of sentences. This investigation will help determine which prosodic components are conducive to native and non-native listeners’ syntactic parsing of utterances and how types and placement of prosodic boundaries interact with tonal information to lead the parser in the recovery of the proper syntactic analysis. Further studies will address the role of intonation phrases in L2ers’ semantic/pragmatic interpretation, another area that has received little attention. Hyekyung will also be conducting a research project with Dr. Goad and Dr. White, investigating whether second language learners produce distinctive prosodic patterns in ambiguous RC-attachment construction and whether the prosodic patterns are reflected in their attachment preferences during silent reading. In order to determine more precisely when implicit and explicit prosodic information is used in parsing and how strong its effects are to determine one interpretation over another as compared with syntactic, semantic, or discourse information, the questions will be investigated by using more online techniques, such as eye-tracking and ERP. These studies will help specify a cross-linguistic architecture of natural language processing and provide further evidence for a universal processing mechanism.