Repository logo

Syntactic and Referential Ambiguity Resolution Under Cue Competition in Silent Reading: Effects of Implicit Prosodic Cues, L1-L2 and Individual Differences



Change log


Tsoukala, Andromachi 


Research in the 80’s and 90’s had demonstrated that comprehenders consider multiple types of information, inclusive of syntactic, nonsyntactic as well as nonlinguistic, when parsing and interpreting ambiguous linguistic structures. Over the past two decades, more recent accounts of sentence processing argue that comprehenders do not just incrementally integrate bottom-up input; they use various types of information as predictive cues to pre-activate structural analyses. Within this growing body of research, it has become apparent that not all cues are created equal; structural biases have repeatedly been shown to be key determinants of syntactic and referential ambiguity resolution, whereas other cues may be less influential. Similarly, not all comprehenders exhibit equal cue sensitivity; research focusing on L1-L2 differences has typically reported absent, weaker or delayed effects in L2 populations. Finally, individual variability within L1 and L2 groups in terms of working memory, language experience and L2 proficiency, inter alia, has also been shown to modulate these effects.

The present PhD thesis extends the work on L1 and L2 ambiguity resolution under conditions of cue competition in silent reading; this is achieved by investigating the influence of formal cues commonly found in poetry, namely rhyme, meter and lineation. Whereas previous studies have examined competition between structural preferences (e.g. syntax-based parsing principles, established biases in reference interpretation) and nonsyntactic information (e.g. lexical semantics, verb subcategorisation frequencies, discourse context), little is known about how formal features of text that introduce implicit prosodic regularities influence decisions about syntactic-prosodic analysis and reference interpretation. These features were manipulated in globally and locally ambiguous poem-like texts which were silently read by L1 comprehenders (self-paced reading and eye-tracking) as well as L2 participants (self-paced reading). To examine effects of rhyme and meter on global anaphoric ambiguity, competition was introduced by having a nonsubject pronoun antecedent be a rhymed element in regularly metered and rhyming versions of the stimuli. To examine effects of line breaks on local (direct object/subject) ambiguity, competition was introduced by altering the transitivity status of an optionally transitive line-final verb (line break and clause boundary match or mismatch), along with the structural completeness of lines preceding this region (complete or enjambed lines to prime match or mismatch, respectively).

The findings of these studies bring attention to L1-L2 dissimilarities and within-group variability. The global ambiguity results suggest that L1 comprehenders who did not activate alternative referential interpretations (attributable to lower working memory) and L2ers were not as sensitive to the implicit prosodic cues (absent or delayed effects of rhyme and meter on anaphora resolution). Additionally, the evidence from the local ambiguity studies suggests that both L1 and L2 readers normally expect and benefit from having line breaks coincide with syntactic boundaries. In fact, this expectation for coterminality rather than line-break and clause boundary mismatch may be too strong to subvert. To the extent that some evidence of adaptation to mismatch was detected, this was more clearly observed in L1 rather than L2 readers. When preceding context signaled that this expectation may be falsified, L2ers were not as efficient as L1 readers at using these cues predictively, they took additional time to activate alternative parses and struggled more with reanalysis.





Tsimpli, Ianthi


anaphora resolution, implicit prosody, individual differences, L2 processing, prediction, priming, referential processing, structural ambiguity resolution, syntactic processing, working memory


Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge
Economic and Social Research Council (2275541)