The perception of word boundaries in a second language

Abstract

Adult Spanish second language (L2) learners of English and native speakers of English participated in an English perception task designed to investigate their ability to use L2 acoustic-phonetic cues, e.g., aspiration, to segment the stream of speech into words. Subjects listened to a phrase and indicated whether they heard, e.g., keep sparking or keeps parking. The results indicate that learners are significantly worse than native speakers at using acousticphonetic cues, and that some types of stimuli are easier for learners to segment than others. The findings suggest that various factors, including transfer and markedness, may be relevant to success in L2 segmentation. I IntroductionIn order to comprehend language, part of what the listener needs to do is to segment the continuous stream of incoming speech into meaningful units. A number of researchers (e.g., Quené, 1992;Gow and Gordon, 1995) suggest that speakers of a language need to use acousticphonetic information to segment speech and retrieve words, probably in conjunction with other information, for example, syntactic information. The majority of research in second language (L2) speech has explored speech production (Leather, 1999); those studies dealing with L2 speech perception have focused largely on the perceptual categorization of L2 speech sounds. (For reviews, see, for example, Leather, 1999; Major, 2001.) Thus, our knowledge of second language speech perception in general is limited, our knowledge of how well second language Address for correspondence: Evelyn P. Altenberg, Department of Speech-Language-Hearing Sciences, Davison 106, Hofstra University, Hempstead, NY 11549, USA; email: sphepa@hofstra.edu learners are able to use acoustic-phonetic information in L2 speech perception even more so. The goal of the study reported here is to explore the abilities of adult learners of English to use acoustic-phonetic cues to segment the second language stream of speech into words. Learners listened to English stimuli such as keep sparking and indicated whether they heard, for example, keep sparking or keeps parking. Such an investigation adds to our knowledge of what learners do and do not know in terms of L2 speech perception and provides an avenue for the exploration of relevant issues in second language acquisition.The study addresses the following two questions:• How do second language learners compare to native speakers in their ability to use acoustic-phonetic cues to segment natural speech into words? In particular, is there evidence that learners are able to acquire and use second language acoustic-phonetic cues?• Are some stimuli easier than others for second language learners to segment and, if so, what factors can account for these differences? In particular, is there evidence of transfer at the allophonic and /or syllabic levels, and/or of universal factors, such as markedness, in L2 speech segmentation?Two issues that have been central to the study of L2 phonology, transfer and markedness, are relevant to the investigation here. Research...

Perception of Allophonic Cues to English Word Boundaries by Polish Learners: Approximant Devoicing in English

“…This result for Polish learners is very similar to the one obtained for Spanish learners (58.5%) by Altenberg (2005) and slightly lower than for Japanese learners (Ito and Strange 2009). It adds to the previous results that non-native speakers are outperformed by native speakers, because non-native speakers transfer segmentation strategies from their L1 into L2 (Otake et al 1993;Cutler et al 1986;Mehler et al 1981;Otake et al 1993).…”

“…These cues are accessed by infants in the early acquisition of the ambient language, even before they are tuned to the language-specific properties. Although, in their study, adult English speakers effectively segmented words in unfamiliar languages such as French, Turkish and Hungarian when only prosodic cues to word boundaries were given, it is quite contrary to many reports that non-native segmentation is significantly compromised compared to the segmentation in the native language (e.g., Altenberg 2005;Cutler, Mehler, Norris and Segui 1992). It appears that, while there may be some language-independent word-boundary cues available, most segmentation relies on languages-specific cues.…”

“…Nevertheless, as emphasised earlier, any definitive conclusions are precluded by the similarly low accuracy scores in both groups. The RTs obtained in the current study cannot be compared to the studies by Altenberg (2005) and Ito and Strange (2009), because the RTs were not collected in these studies.…”

“…Krakow (1993) found that vowel nasalization induced by a following nasal consonant was significantly reduced when the two sounds were separated by a word boundary. Cole, Kim, Choi and Hasegawa-Johnson (2007) reported longer VOTs for word-initial than word-medial stops and this cue has been found to be attended to by native speakers of English in segmenting words (Altenberg 2005;Nakatani and Dukes 1977). All these studies point to the fact that, despite the frequent lack of observable points in the acoustic signal that would indicate the division of sound sequences into words, the onset and offset of words is signaled by fine-grained spectral and temporal properties of sounds that listeners are sensitive to in speech segmentation.…”

“…Previous research on non-native word segmentation from voicing cues in English concentrated on voiceless-stop aspiration as a boundary cue. Altenberg (2005) tested Spanish listeners' segmentation of sequences such as keeps parking vs. keep sparking, where the word-boundary cue is the aspiration of word-initial /p/ in parking and the absence of aspiration in sparking. The results indicated that Spanish learners performed significantly worse than native speakers of English in exploiting aspiration as a boundary marker.…”