Perception of intrusive /r/ in English by native, cross-language and cross-dialect listeners

Abstract

In sequences such as law and order, speakers of British English often insert /r/ between law and and. Acoustic analyses revealed such "intrusive" /r/ to be significantly shorter than canonical /r/. In a 2AFC experiment, native listeners heard British English sentences in which /r/ duration was manipulated across a word boundary [e.g., saw (r)ice], and orthographic and semantic factors were varied. These listeners responded categorically on the basis of acoustic evidence for /r/ alone, reporting ice after short /r/s, rice after long /r/s; orthographic and semantic factors had no effect. Dutch listeners proficient in English who heard the same materials relied less on durational cues than the native listeners, and were affected by both orthography and semantic bias. American English listeners produced intermediate responses to the same materials, being sensitive to duration (less so than native, more so than Dutch listeners), and to orthography (less so than the Dutch), but insensitive to the semantic manipulation. Listeners from language communities without common use of intrusive /r/ may thus interpret intrusive /r/ as canonical /r/, with a language difference increasing this propensity more than a dialect difference. Native listeners, however, efficiently distinguish intrusive from canonical /r/ by exploiting the relevant acoustic variation.

Resolving ambiguity in familiar and unfamiliar casual speech

“…2). This is as predicted from the results of our phonetic categorization study (Tuinman et al, 2011), in which British English listeners based their interpretation of /r/ as onset or intrusive exclusively on the acoustic-phonetic evidence.…”

“…In our phonetic categorization study (Tuinman et al, 2011), where an artificial durational manipulation was built into our materials, we also found that L1 and L2 listeners alike could perceive this manipulation; the native listeners based their ice-rice categorizations exclusively on the durational factor, and ignored all else, but the L2 listeners allowed semantic and orthographic factors to outweigh it. The present cross-modal priming results have reinforced the conclusion that native listeners can act upon their processing of the precise realization of /r/; they showed fully appropriate priming patterns for the intended word only, whichever it was.…”

“…We measured the duration of the /r/ and the amplitude difference between the center of the /r/ and the surrounding vowels. In the sentences with an onset /r/, the /r/ was as predicted (Cruttenden & Gimson, 1994;Tuinman et al, 2011) longer and had a larger intensity decrease than in similar sentences with intrusive /r/s. The full set of measurements may be found in Tuinman et al (2011).…”

“…In a previous study (Tuinman, Mitterer, & Cutler, 2011), we examined the sensitivity of listeners to onset versus intrusive /r/ realisations using a phonetic categorization task. First, we found that, as previously reported in the phonetic literature, intrusive /r/ is acoustically weak compared with onset /r/ -onset tokens are longer in duration and have a larger intensity decrement than intrusive tokens.…”

French Speech Segmentation in Liaison Contexts by L1 and L2 Listeners

“…Compensation failure could lead to perception mistakes (e.g., listeners may identify the second word as aces or as races). Tuinman, Mitterer, and Cutler (2011) found that listeners utilize duration differences between intrusive /r/ and word-initial /r/ to compensate for this phenomenon and identify the second word with a high level of accuracy.…”