Page 17 - Fall 2011

IMITATION IN SPEECH Molly Babel Department of Linguistics University of British Columbia Vancouver, British Columbia V6T 1Z4, Canada Why do we sound the way we do? As people learn to speak, they acquire the language and dialect spoken around them. Sentence structure, word choice, and pronuncia- tion are all determined by the patterns used in the ambient language to which we are exposed. Having grown up in Minnesota, I did not learn how to speak with a British accent, but with a Minnesotan one. As a language-learning child, the language input I received determined the general shape of my language output. This article focuses on the spontaneous or natural imitation of speech acoustics. In this article I use the terms imitation, convergence, and accommodation interchangeably. I use all of these terms to describe the unintentional process by which exposure to a speech stimulus causes an observer to display characteris- tics of the stimulus in their own productions. This phenomenon of imitating the input is not limited to the acoustic signal that we use to transmit language. Let’s begin with a straightforward example from the literature on syntactic priming and word order of how recent linguistic exposure modifies our subsequent speech behavior. Imagine a picture of a man holding a cake and facing a woman. The orientation of the image suggests the man intends to pass the cake to the woman. Participants who have been exposed to the sentence The boy gave the toy to the teacher prior to view- ing this image are more likely to describe the cake picture as The man gave the cake to the woman as opposed to The man gave the woman the cake. The second description is a com- pletely grammatical utterance that accurately conveys what is going on in the image, but having been previously exposed to the construction give X to Y biases the future use of that con- struction over give Y X (Bock, 1986). Bock’s seminal finding reveals quite convincingly that what we say is highly influ- enced by what we have just heard. The notion that children learn the speech variety to which they are exposed seems intuitive, but the situation becomes a little more complicated when we shift our atten- tion to the acquisition patterns of adults. What happens when adults who have already acquired a particular speech variety move to a new dialect area? As a young adult, I moved to California and, with time, my speech lost many of its orig- inal Minnesotan features. To native Californian ears, I might never have sounded truly Californian, but I eventually sounded much less Minnesotan. Several recent studies have documented this personal anecdote on a larger scale (Evans and Iverson, 2006; Munro et al., 1999). Interestingly, the Minnesotan features of my speech return when I am inter- acting with old friends and family who have retained our native dialect. This indicates that the acquisition of a new “Social preferences and liking modulate the process of spontaneous phonetic imitation.” dialect or the adoption of new speech features serve as an update to and expansion upon my linguistic system, as opposed to wholly replacing a previous system. The fact that we do imitate the ambient language tells us that our lin- guistic categories are malleable and eas- ily influenced by new information. Inherent variability in speech production When one considers how speech is produced, it becomes apparent that physiological and anatomical variation across talkers will inevitably be reflected in the spectral characteris- tics of speech sounds. Let’s start with the sex of the talker. Men’s voices typically pattern together in having lower pitch and lower resonant frequencies than women’s voices. These differences are due in part to sexual dimorphism: men’s vocal tracts and vocal folds are generally larger than women’s. Age is another key factor in cross-talker variability. Aging is accompanied by various physiological changes. For example, the extrinsic muscles that support the larynx become slack with age, and the mucosal tissue covering the vocal folds loses its elasticity. These changes lead to alterations in voice quality, along with lower pitch and lower resonant frequen- cies for a talker’s voice. These factors highlight the fact that some of the variation in speech is due to anatomical and physiological age- and sex-based variation. That being said, the extent to which physiological factors determine speech characteristics is frequently oversold. A closer inspection of large datasets reveals that physiological variation does not account for all of the observed differences between groups divided by gender or age. While it is true that classic studies describing vocalic resonant frequencies of men and women show that women produce higher resonant fre- quencies than men (Peterson and Barney, 1952), a recent examination of gender differences across languages illustrat- ed that gender-based differences in vowel production vary across languages, even when population height is controlled (Johnson, 2006). Such a finding indicates that some of the gender-based differences in resonant frequencies are the result of learned social norms. In addition, despite the fact that many significant anatomical differences do not emerge between males and females until puberty, children acquire gender-specific speech patterns starting in toddlerhood (Sachs et al., 1973; Perry et al., 2001). This suggests a strong socio-cultural component to language production. Armed with this information, we can hedge a response to the question of why we sound the way we do by acknowl- edging that a portion of one’s speech acoustics is determined by the size and shape of the vocal tract. There is clearly more at issue, however. As mentioned above, we acquire the speech 16 Acoustics Today, October 2011