Katie Drager

Factors influencing speech perception in the context of a merger-in-progress

Abstract In New Zealand English there is a merger-in-progress of the near and square diphthongs. This paper investigates the consequences of this merger for speech perception. We report on an experiment involving the speech of four New Zealanders—two male, and two female. All four speakers make a distinction between near and square . Participants took part in a binary forced-choice identification task which included 20 near / square items produced by each of the four speakers. All participants were presented with identical auditory stimuli. However the visual presentation differed. Across four conditions, we paired each voice with a series of photos—an “older” looking photo, a “younger” looking photo, a “middle class” photo and a “working class” photo. The middle and working class photos were, in fact, photos of the same people, in different attire. In a fifth condition, participants completed the task with no associated photos. At the end of the identification task, each participant was recorded reading a near/square wordlist, containing the same items as appeared in the perception task. The results show that a wide range of factors influence accuracy in the perception task. These include participant-specific characteristics, word-specific characteristics, context-specific characteristics, and perceived speaker characteristics. We argue that, taken together, the results provide strong support for exemplar-based models of speech perception, in which exemplars are socially indexed.

From fush to feesh: Exemplar priming in speech perception

Abstract Niedzielski (1999) reports on an experiment which demonstrates that individuals in Detroit ‘hear’ more Canadian Raising in the speech of a speaker when they think that speaker is Canadian. We describe an experiment designed to follow up on this result in a New Zealand context. Participants listened to a New Zealand English (NZE) speaker reading a list of sentences. Each sentence appeared on the answer-sheet, with a target word underlined. For each sentence, participants were asked to select from a synthesized vowel continuum the token that best matched the target vowel produced by the speaker. Half the participants had an answer-sheet with the word ‘Australian’ written on it, and half had an answer-sheet with ‘New Zealander’ written on it. Participants in the two conditions behaved significantly differently from one another. For example, they were more likely to hear a higher fronter /i/ vowel when ‘Australian’ appeared on the answer sheet, and more likely to hear a centralized version when ‘New Zealander’ appeared – a trend which reflects production differences between the two dialects. This is despite the fact that nearly all participants reported that they knew they were listening to a New Zealander. We discuss the implication of these results, and argue that they support exemplar models of speech perception.

Stuffed toys and speech perception

Abstract Previous research has shown that speech perception can be influenced by a speakers social characteristics, including the expected dialect area of the speaker (Niedzielski, Journal of Language and Social Psychology 18: 62–85, 1999; Hay et al. The Linguistic Review 23: 351–379, 2006a). This article reports on an experiment designed to test to degree to which exposure to the concept of a region can also influence perception. In order to invoke the concept, we exposed participants, who were all speakers of New Zealand English, to either stuffed toy kangaroos and koalas (associated with Australia) or stuffed toy kiwis (associated with New Zealand). Participants then completed a perception task in which they matched natural vowels produced by a male New Zealander to vowels from a synthesized continuum which ranged from raised and fronted Australian-like tokens to lowered and centralized New Zealand-like tokens. Our results indicate that perception of the vowels shifted depending on which set of toys the participants had seen. This supports models of speech perception in which linguistic and nonlinguistic information are intricately entwined.

Speaker Age and Vowel Perception

Recent research provides evidence that individuals shift in their perception of variants depending on social characteristics attributed to the speaker. This paper reports on a speech perception experiment designed to test the degree to which the age attributed to a speaker influences the perception of vowels undergoing a chain shift. As a result of the shift, speakers from different generations produce different variants from one another. Results from the experiment indicate that a speaker’s perceived age can influence vowel categorization in the expected direction. However, only older participants are influenced by perceived speaker age. This suggests that social characteristics attributed to a speaker affect speech perception differently depending on the salience of the relationship between the variant and the characteristic. The results also provide evidence of an unexpected interaction between the sex of the participant and the sex of the stimulus. The interaction is interpreted as an effect of the participants’ previous exposure with male and female speakers.The results are analyzed under an exemplar model of speech production and perception where social information is indexed to acoustic information and the weight of the connection varies depending on the perceived salience of sociophonetic trends.

Exploiting random intercepts: Two case studies in sociophonetics

An increasing number of sociolinguists are using mixed effects models, models which allow for the inclusion of both fixed and random predicting variables. In most analyses, random effect intercepts are treated as a by-product of the model; they are viewed simply as a way to fit a more accurate model. This paper presents additional uses for random effect intercepts within the context of two case studies. Specifically, this paper demonstrates how random intercepts can be exploited to assist studies of speaker style and identity and to normalize for vocal tract size within certain linguistic environments. We argue that, in addition to adopting mixed effect modeling more generally, sociolinguists should view random intercepts as a potential tool during analysis.

Short-term Exposure to One Dialect Affects Processing of Another

It is well established that speakers accommodate in speech production. Recent work has shown a similar effect in perception—speech perception is affected by a listener’s beliefs about the speaker. In this paper, we explore the consequences of such perceptual accommodation for experiments in speech perception and lexical access. Our interest is whether perceptual accommodation to one speaker—the experimenter who meets participants, for example— might have carry-over effects on participants’ behavior in subsequent tasks that do not directly involve the experimenter’s voice. We explore this possibility by exposing groups of participants to different varieties of English before they participate in experiments involving speech perception and/or lexical access. Our results reveal that the nature of this prior exposure considerably influences participants’ behavior in the tasks. This suggests that the phonetic detail of encountered speech is stored in the lexicon, together with information about the speaker’s regional origin. Subparts of phonetically detailed lexical distributions can then be effectively ‘primed’ by exposure to speakers or lexical items associated with particular dialects. We argue for an exemplar model of lexical representation with both word-level and phoneme-level representations. The consequences of cross-dialectal priming vary, depending on whether tasks involve primarily word-level or phoneme-level access.

Using nonsense words to investigate vowel merger

In previous work, we have found that New Zealand listeners who produce merged tokens of near and square can accurately distinguish between the vowels in perception even though they report that they are guessing. The ability to distinguish the vowels is affected by a variety of factors for these listeners, including the likelihood that the speaker and experimenter maintain the distinction (Hay et al . 2006b; Hay et al . 2010). In this article, we report on experiments that examine the production and perception of real and nonsense words in the context of two mergers: the Ellen/Allan merger in New Zealand English and the lot/thought merger found in American English. The results demonstrate that speakers’ degree of merger depends at least partially on whether the word is a real or nonsense word. Additionally, the results indicate that a tokens real word status affects the merger differently in production and perception. We argue that these results provide evidence in favour of a hybrid model of speech production and perception, one with both abstract phoneme-level representations and acoustically detailed episodic representations.

Car-talk: Location-specific speech production and perception

Abstract Some locations are probabilistically associated with certain types of speech. Most speech that is encountered in a car, for example, will have Lombard-like characteristics as a result of having been produced in the context of car noise. We examine the hypothesis that the association between cars and Lombard speech will trigger Lombard-like speaking and listening behaviour when a person is physically present in a car, even in the absence of noise. Production and perception tasks were conducted, in noise and in quiet, in both a lab and a parked car. The results show that speech produced in a quiet car resembles speech produced in the context of car noise. Additionally, we find tentative evidence indicating that listeners in a quiet car adjust their vowel boundaries in a manner that suggests that they interpreted the speech as though it were Lombard speech.

Careful Who You Talk to: An Effect of Experimenter Identity on the Production of the NEAR/SQUARE Merger in New Zealand English

This paper reports on results from a production experiment conducted in New Zealand, where the diphthongs and are involved in a merger-in-progress. The results suggest that the cognitive status of and word pairs as being the ‘same’ or ‘different’ is not entirely related to their production. This status could be affected by functional load: words that could lead to contextual ambiguity may seem more different than those for which the distinction is not functionally important. The results also provide evidence that an experimenters dialect area can influence participants’ production of these diphthongs. Participants in this study were met by two different experimenters from different dialect regions: New Zealand and the United States. Post hoc analysis reveals that participants who met with the US experimenter were more likely to maintain a distinction between the diphthongs than those who met with the NZ experimenter. Additionally, merged participants were more likely to report that and word pairs were distinct in their own speech if met by the US experimenter. Individuals from the US are more likely to maintain a distinction than speakers of NZE, so these results suggest that participants accommodate to the speech of an experimenter. We compare this finding with related results from a perception experiment completed by the same participants, presented in Hay, Warren, and Drager, (2006). While there appears to be a mismatch between results from production and perception, we interpret this as evidence that the observed effects are due to exposure to the experimenter and not to the distribution of participants across the different groups.

Divergence in speech perception

Abstract This paper presents results from an experiment designed to test whether New Zealand listeners’ perceptual adaptation towards Australian English is mediated by their attitudes toward Australia, which we attempted to manipulate experimentally. Participants were put into one of three conditions, where they either read good facts about Australia, bad facts about Australia, or no facts about Australia (the control). Participants performed the same listening task – matching the vowel in a sentence to a vowel in a synthesized continuum – before and after reading the facts. The results indicate that participants who read the bad facts shifted their perception of kit to more Australian-like tokens relative to the control group, while the participants who read good facts shifted their perception of kit to more NZ-like tokens relative to the control group. This result shows that perceptual adaptation towards a dialect can occur in the absence of a speaker of that dialect and that these adaptations are subject to a listener’s (manipulated) affect towards the primed dialect region.