The emergence of sub-syllabic representations q Yongeun Lee * , Matthew Goldrick Department of Linguistics, Northwestern University, USA article info Article history: Received 5 February 2007 Revision received 6 March 2008 Available online 23 May 2008 Keywords: Sub-syllabic constituents Short-term memory tests Phonotactic probability abstract In a variety of experimental paradigms speakers do not treat all sub-syllabic sequences equally. In languages like English, participants tend to group vowels and codas together to the exclusion of onsets (i.e., /bet/=/b/-/et/). Three possible accounts of these patterns are examined. A hierarchical account attributes these results to the presence of categorical sub-syllabic divisions (e.g., onset /b/ vs. rime /et/). In contrast, a non-hierarchical account attributes these findings to the strength with which particular consonants and vowels co-occur (e.g., the relative frequency of /be/ vs. /et/). A third emergentist alternative is articulated, where speakers’ knowledge is sensitive not just to the frequency with which particular segments co-occur but also to the general patterns of association within a language. These accounts are contrasted by examining sub-syllabic patterns in two languages: Korean and English. A statistical study shows distinctions in the general patterns of sub-syllabic associations in Korean and English. Consistent with the emergen- tist perspective, results of short-term memory experiments reveal that Korean and English speakers are sensitive not only to these general patterns but also to the strength with which particular segment sequences are associated. The implications of these results for theories of sub-syllabic patterns in languages are discussed. Ó 2008 Elsevier Inc. All rights reserved. 1. Introduction Researchers in the field of phonology generally agree that there are certain linguistically significant patterns that involve segment sequences occurring inside the syllable. For example, within the syllables of several Indo-European languages there are differences in phonological restrictions governing onset–vowel vs. vowel–coda sequences (e.g., Dutch: Martensen, Maris, & Dijkstra, 2000; English: Kessler & Treiman, 1997; French: Perruchet & Peereman, 2004). For instance, Kessler and Treiman’s (1997) statistical anal- yses of English monomorphemic monosyllables show that many phonologically legal combinations of vowel and coda never occur (e.g., /ib/), or occur less than expected by chance (e.g., /l/), while there are relatively few restric- tions on the combinations of onset and vowel. As reviewed below, behavioral results indicate that these patterns of associations exert an influence in many domains of lan- guage processing. Despite the general agreement among researchers on the existence and significance of these sub-syllabic pat- terns, the exact mechanism that accounts for them re- mains an active issue of debate (Gupta & Dell, 1999; Pierrehumbert & Nair, 1995; Treiman & Kessler, 1995; Yip, 2003). One approach (referred to as the hierarchical 0749-596X/$ - see front matter Ó 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.jml.2008.03.002 q This research was supported in part by a Graduate Research Grant to the first author from the Northwestern University Graduate School. Portions of this work were presented at the 34th annual meeting of the Linguistic Association of the Southwest (October, 2005), the 42nd annual meeting of the Chicago Linguistic Society (April, 2006), and the 15th annual meeting of the European Society of Cognitive Psychology (August, 2007) and submitted as part of the author’s Doctoral dissertation (Northwestern University, 2006). The authors thank the members of the first author’s dissertation committee (Ann Bradlow and Janet Pierrehum- bert) for their invaluable advice and support during this project as well as Kristin Van Engen and Midam Kim for assistance in transcription of the production data. * Corresponding author. Present address: Division of English Language and Literature, College of Humanities, Cheongju University, 36 Naedok- Dong, Sangdang-Gu, Cheongju, Chungbuk 360-764, Republic of Korea. E-mail address: ylee@cju.ac.kr (Y. Lee). Journal of Memory and Language 59 (2008) 155–168 Contents lists available at ScienceDirect Journal of Memory and Language journal homepage: www.elsevier.com/locate/jml