Academic journal article Memory & Cognition

Does Visual Speech Information Affect Word Segmentation?

Academic journal article Memory & Cognition

Does Visual Speech Information Affect Word Segmentation?

Article excerpt

We present an experiment in which we explored the extent to which visual speech information affects learners' ability to segment words from a fluent speech stream. Learners were presented with a set of sentences consisting of novel words, in which the only cues to the location of word boundaries were the transitional probabilities between syllables. They were exposed to this language through the auditory modality only, through the visual modality only (where the learners saw the speaker producing the sentences but did not hear anything), or through both the auditory and visual modalities. The learners were successful at segmenting words from the speech stream under all three training conditions. These data suggest that visual speech information has a positive effect on word segmentation performance, at least under some circumstances.

A key step in children's language acquisition is to find the word units of their language. Succeeding at this task can be quite difficult, because, most often, the words must be extracted from a fluent stream of speech that does not provide obvious cues (such as pauses) to the boundaries between words (e.g., Saffran, Aslin, & Newport, 1996). The statistical learning approach to language acquisition proposes that children are able to successfully segment words from a fluent stream of speech by exploiting statistical regularities in their linguistic input (e.g., Aslin, Saffran, & Newport, 1998; Saffran et al., 1996). In support of this approach, it has been shown that infants are able to use the transitional probabilities between syllables (i.e., the likelihood of one syllable following another) in order to find the boundaries between words in a fluent speech stream (Aslin et al., 1998; Saffran et al., 1996). The use of transitional probabilities between linguistic units has also been shown to be sufficient to begin solving other problems in language acquisition. For instance, children and adults can use the transitional probabilities between word classes to acquire the rudiments of syntax (e.g., Kaschak & Saffran, 2006; Saffran & Wilson, 2003; Thompson & Newport, 2007), and statistical word segmentation processes have been shown to facilitate the development of links between objects and their labels (e.g., Estes, Evans, Alibali, & Saffran, 2007; Mirman, Magnuson, Estes, & Dixon, 2008).

Although numerous researchers have demonstrated that the statistical properties of the linguistic input can be exploited in the service of language acquisition (e.g., Gomez & Gerken, 1999; Saffran et al., 1996), fewer have explored how other factors interact with statistical information in language learning. As one example, Thiessen and Saffran (2003, 2007) traced the developmental course of infants' use of statistical and prosodic cues to word boundaries in tasks in which both types of cues are present. Their results show that infants weight the importance of statistical and prosodic cues differently across time, initially giving more weight to statistical cues but later giving more weight to prosodic cues. Results such as these suggest that there are many layers of information that must be considered in a statistical approach to language learning. Transitional probabilities between syllables provide one layer of statistical information, and the interaction between these probabilities and the presence of prosodic cues (such as stress) in the linguistic input provides another layer of statistical information. Given the multiplicity of cues (and layers of statistical information) that are potentially available to the language learner, it is important to understand how and when (if at all) language learners exploit the information around them in acquiring their language (see Mattys, White, & Melhorn, 2005).

The goal of the present work is to examine the extent to which one type of linguistic information-the availability of visual speech information (i.e., the ability to lip read or speech read the speaker)-affects learners' ability to segment novel words from a fluent speech stream. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.