2018 Sat Poster 6737

Saturday, November 3, 2018 | Poster Session II, Metcalf Small | 3:15pm

L2 Adaptation to Unreliable Prosody During Structural Analysis: A Visual World Study
C. Nakamura, J. Harris, S. Jun, Y. Hirose

Recent work on sentence processing indicates that comprehenders rapidly adapt their expectations according to exposure [1,2] and that they can weigh different kinds of information according to their reliability [3]. Our study investigates whether L2ers modulate the degree to which they use prosodic cues in structural analysis based on the reliability of prosody they experience in the experiment. In two visual world eye-tracking experiments, we examined (1) whether L2ers use prosodic boundary information in the interpretation of globally ambiguous sentences in English, and (2) whether structural analysis in L2 is affected by the overall reliability of prosodic cues.

Experiments. L1-Japanese speakers of L2-English (intermediate-advanced) participated in two visual-world eye-tracking experiments, each with 24 target and 48 filler items. In Experiment 1 (N=33), filler items had reliable prosody (1a). In Experiment 2 (N=28), filler items had ‘unreliable’ prosody (1b), in which the boundary was inappropriately located in the middle of the final NP, and was realized as a L-L% (falling tone) instead of L-H% (continuation rise). In target sentences, location of a prosodic boundary was manipulated to appear either before the NP preceding the ambiguous PP, encouraging Modifier analysis (2a, the tiger has the binoculars), or after it, encouraging Instrument analysis (2b, see with the binoculars). We also manipulated the semantic fit of the final noun in the PP to be either appropriate for the verb action (2ab; binoculars) or not (3ab; popcorn). As a control, thirty-two native English (L1ers) were tested separately in the same paradigm.

Results. Looks made to each object in the picture for the duration of the final noun (binoculars/popcorn) were analyzed. The results showed that L2ers were sensitive to the location of the prosodic boundary, looking more at Modifier object (tiger  with binoculars/popcorn) with Modifier prosody and at the Instrument object with Instrument prosody. However, unlike the L1ers whose effects were observed before the onset of the final noun, no anticipatory looks to these objects were observed with L2ers. The results also showed that L2ers looked more at the Instrument object in (3b) than in (3a), suggesting that they experienced a garden-path effect when prosodic structure and noun mismatched. These results indicate that although L2ers made structural expectations from prosodic information, these expectations were not as strong as L1ers. In addition to the effects observed in Experiment 1, Experiment 2 showed a significant effect of Item Order for the looks to Modifier object, indicating L2ers adopted the Modifier interpretation more as they experienced trials in the experiment. Combined analysis between the two experiments confirmed that the Item Order effect was significant only in Experiment 2, demonstrating L2ers relied on information from the visual array more when filler items had unreliable prosody and adopted the visually salient Modifier interpretation. The results together revealed that L2ers use prosodic boundary information in structural analysis, although their predictions are delayed compared to L1ers. The results also suggest that L2ers track how reliable prosodic cues are, and adjust the extent to which they use prosody in making judgments during structural analysis.

References

  1. Chang et al. (2006). Becoming syntactic. Psychological Review,113
  2. Fine et al. (2013). Rapid expectation adaptation during syntactic comprehension. PloS one,8.
  3. Tanenhaus et (2015). Prosody and intention recognition. In: Explicit and implicit prosody in sentence processing.