안녕하세요, 서울대학교 언어학과 사무실입니다.
2018학년도 제3차 언어학 콜로퀴엄을 아래와 같이 개최하오니 많은 참석을 부탁드립니다.

[3차 언어학 콜로퀴엄]
- 일시: 2018. 6. 26.(화) 오후 4시
- 장소: 서울대학교 인문대학 신양인문관 국제회의실(4동 302호)
- 발표자: Bruce Hayes (UCLA)
- 제목: Phonological markedness effects in syntax: subtle but ubiquitous
- 초록: 아래 참고

As specialization spreads ever further in the field of linguistics it is of interest to see how “compartmentalizable” the human ability to create sentences really is. The traditional view (e.g., Chomsky 1965) posits a feed-forward model in which syntax is solely responsible for determining the set of well-formed sentences, and phonology then provides a phonetic interpretation for each sentence. This architecture has repeatedly been challenged, most recently by Shih and Zuraw (forthcoming), who demonstrate that when Tagalog speakers select from the two available Adjective + Noun constructions, i.e. Adj. na Noun and Noun na Adj., their choice is influenced by a tendency to avoid violations of phonological Markedness constraints, such as a ban on adjacent nasals.
This research, a collaboration with Canaan Breiss, adopts a broader line of investigation, checking multiple constraints against various large bodies of text (literature, speech corpora). We expressed in code a set of phonological constraints that are generally unviolated within English words (e.g. *GEMINATE, *SIBILANT CLASH). We machine-counted how many times these constraints are violated across word boundaries in our text corpora; for instance, the consecutive word pair success supposes [səkˈsɛs səˈpoʊzəz], from our Jane Austen corpus, would be counted as violating both constraints. We next constructed a statistical model intended to calculate the distribution of the violation counts that would be expected under the null hypothesis that sentence formation is not influenced by phonology. We then checked whether the number of observed violations falls significantly below the expected values. In general this turns out to be so, even after we imposed a variety of controls to be described in the talk.
Our tentative conclusion is that phonological markedness constraints are not confined to driving the derivations of the phonology; they work in tandem with syntactic constraints and lexical choice in guiding the output of the language system as a whole.