Graduate Student, Morgan Robertson awarded Doctoral Dissertation Research Improvement Grant from the National Science Foundation (NSF)

Title:

Doctoral Dissertation Research: Clear Speech Productions and Intelligibility Benefit for Native and Non-native Speakers and Listeners

 Abstract:

Clear or hyperarticulated speech is a strategy that speakers use to overcome potential communication problems that arise in situations like communicating with someone who does not share the same first language. However, not all clear speech is the same nor does it always accomplish its goal of aiding understanding, ultimately improving intelligibility. This doctoral dissertation examines the acoustic properties of clearly spoken language use produced by native and non-native speakers. Additionally, it examines if and how much the clear speech improves intelligibility for native and non-native listeners. Previous research has shown that the acoustic properties speakers enhance in clear speech tend to be the primary cues to the category. However, primary cues can differ from language to language. There is reason to believe that non-native clear speech, which is likely influenced by native cues to categories, will differ from native clear speech in acoustic modifications and therefore may not be as effective at improving intelligibility. Understanding exactly how non-native clear speech is produced and to what extent it enhances intelligibility can inform linguistic theories of second-language speech perception by shedding light on how phonetic categories are organized and interact within the mind of a second-language learner. Additionally, knowing specifically what speech modifications are beneficial (and which are not) can lead to focused instruction and ultimate removal of communication barriers.

 The doctoral dissertation aims to understand the characteristics of second-language (L2) clear speech, probe the potential differences between contrast-specific clear-speech productions (Did you say X?) and general clear-speech productions (What did you say?), and determine the intelligibility benefit received by both native and non-native speakers as a direct consequence of clear-speech enhancements. The Revised Speech Learning Model (SLM-r) posits that L2 phonetic categories are position-sensitive input distributions situated in an n-dimensional phonetic space shared with the native language (L1). Crucially, this model proposes that the more similar an L2 phone is to an L1 category, the more difficult it will be to unlink the L2 phone (and its primary cue) from the L1 category. Taken together with the claim that clear-speech modifications tend to be made along the primary cue dimension, these claims predict differences in clear-speech strategies for phones that differ in their L2-to-L1 mapping. The present dissertation utilizes a simulated interactive computer program methodology to elicit clear speech for three types of mappings: similar phones (oral stops), new phones (front rounded vowels), and position-specific allophones (vowel nasalization). Contrast-specific clear-speech productions are clear speech productions in response to a specific competitor. This distinction from general clear speech is important, as it determines whether the presence of a contrast influences the cues used in clear speech. Lastly, the present dissertation determines the intelligibility benefit received by the clear speech productions using identification- and discrimination-in-noise tasks. In sum, the doctoral dissertation seeks to understand how native and non-native speakers? phonetic and phonological knowledge is utilized to maintain contrasts and increase intelligibility.