Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Competing Cues in Ambiguous Speech Comprehension: The Role of Visual Competition and Linguistic Context
Poster Session E, Sunday, September 14, 11:00 am - 12:30 pm, Field House
Lauren McBay1, Kirsten Lo1, Jennifer Rodd2, Ingrid Johnsrude1; 1University of Western Ontario, 2University College London
Successful spoken language comprehension requires listeners to quickly resolve ambiguity, often relying on contextual cues and word-meaning frequency information to correctly comprehend words with multiple meanings (homonyms; e.g., bat as an animal or baseball equipment). Several psycholinguistic models offer different accounts of how homonyms are processed. Some propose that all potential meanings of a homonym are always accessed, regardless of contextual information (Dopkins et al., 1992). Others suggest that strong contextual constraints can prevent activation of irrelevant meanings, especially when encountered prior to the homonym. The frequency of a particular meaning (derived from speech corpora) also appears to predict meaning activation (Gardner et al., 1987). Although more frequent (dominant) meanings are generally accessed more easily, it remains unclear how these two cues to the appropriate meaning – context and meaning frequency – interact, particularly when they are in tension (i.e., context supports less frequent meaning). We used a modified (online) visual-world paradigm (Huettig et al., 2011), in which 202 participants listened to sentences containing an ambiguous target word and were instructed to select the item representing the 'correct’ meaning from an array of four images presented in four quadrants of the screen. Images were presented after participants heard the target word. We measured how quickly and how accurately listeners selected the image representing the correct meaning. We manipulated whether the correct meaning was dominant or subordinate, and whether sentential context disambiguating the homonym was available before, or after, it was heard. In this initial study, we also manipulated visual competition – whether items representing both the subordinate and dominant meanings were present in the array of four items. We asked whether the mere presence of the ‘incorrect’ meaning in the array slows selection times, or whether slowing is only evident when the subordinate meaning is correct, and whether reaction times depend on the timing of disambiguating context relative to the homonym in the sentence. This design allowed us to assess whether visual competition interferes with meaning selection, and whether these effects are further influenced by context and meaning frequency. Both context and word-meaning frequency strongly influenced target word interpretation. Prior context and dominant target word meanings resulted in faster and more accurate target item selection. Furthermore, the benefit to accuracy when the dominant meaning was correct was smaller in the prior context conditions, likely due to a ceiling effect. Accuracy was significantly reduced when the competitive distractor was present, particularly when the target word meaning was subordinate. Notably, although competitive distractor presence influenced accuracy, it did not significantly affect reaction time on correct trials. Accuracy at selecting the target meaning of a homonym was higher, and reaction time faster, when disambiguating context was heard before (vs after) the homonym, and when the dominant (vs subordinate) meaning was correct. Intriguingly, the presence of distractors compromised accuracy (particularly when the subordinate meaning was correct), but not reaction time on correct trials. This “all or nothing” effect – distractor presence either drove interpretation entirely, or did not affect it – requires further investigation.
Topic Areas: Meaning: Lexical Semantics, Speech Perception