Recent work in the psychology of language helps us appreciate the complexity of spoken language comprehension in the face of varied pronunciation, writes Victoria University’s Paul Warren
No doubt many of us are familiar with tales of visitors to New Zealand being flummoxed by hearing from our national airline that they have made “online chicken” available or that they provide “fears to suit every traveller”. (I’ll have some arachnophobia, please; but hold off on the fear of heights until we have landed.) We have probably also encountered complaints of how it now seems that cars are universally fitted with “ear bags”, and have heard the rising intonation pattern of youthful uptalk described as an “irritating verbal tic”.
Yet neither mishearings nor the apparent grating effect of youngsters’ speech appear to have any enduring effect on comprehension. The context in which such utterances are heard will generally provide enough information for the speaker’s intention to be clear. But it is also clear that as listeners we continually adapt our expectations based on our prior experience. Such experience allows us to make connections between patterns of speech sounds and characteristics of the speakers who produce those patterns. Indeed, it has been argued that memories for speech events are stored as exemplars that include not just the sounds that were spoken but also traces of the contexts in which they were uttered, including identifying features of the speakers and the situation.
New Zealand English (NZE) has provided a rich vein for the exploration of such effects, because it is a variety of English that is undergoing many changes in its pronunciation. This particularly affects its vowel sounds, but innovative forms of intonation have also been noted, such as uptalk, a rising intonation on statements used to check that the listener is following and to encourage them to engage in the interaction. Recent research shows that uptalk has started to show subtle differences from the intonation patterns used to ask questions.
Just as discrimination between sounds can be affected by the actual or implied characteristics of the speaker, so too the nature of the speech sounds invokes certain beliefs about the speaker.
Research on the perception of NZE vowel sounds has shown that listeners are sensitive to the distribution of variants of a vowel across age, gender and social groupings. For instance, hearing the word ‘cheer’ from a younger speaker activates young NZE listeners’ mental representations for both ‘cheer’ and ‘chair’, reflecting the merger of the vowels in these two words in younger NZE speech. However, when ‘cheer’ is heard from an older-sounding voice, the meaning of ‘cheer’ is activated far more strongly than that of ‘chair’. This is unsurprising, since older NZE speakers still maintain a distinction between the two vowels involved. Such findings demonstrate how listeners dynamically adjust their perception to match speaker characteristics.
Intriguingly, this effect can also result from implied features of the speaker. For example, a photograph purported to be of the speaker’s face suggested that the speaker was older or younger. Both age conditions used the same voice recording. For the older face, listeners were better able to hear the difference between words like ‘cheer’ and ‘chair’. A resetting of the perceptual system was also triggered by the mere presence in the room of a stuffed toy – vowels were more likely to be categorised in a way that matches the Australian English vowel system if the toy was a kangaroo, and in a way that matches the NZE system if it was a kiwi.
The relationship between speech sounds and speaker characteristics in an exemplar system is two-way. Just as discrimination between sounds can be affected by the actual or implied characteristics of the speaker, so too the nature of the speech sounds invokes certain beliefs about the speaker. In one study, this has been shown to impact on subsequent interpretation of intonation. This study used spoken sentences like “John’s mother cared for stray animals”. The sentences finished with a rising intonation matching one of two patterns, either that of a question or that of the uptalk intonation used by younger NZE speakers. When the vowel in ‘cared’ was manipulated to sound like a younger speaker’s vowel (something like ‘keered’), then young NZE listeners could more reliably tell apart the uptalk and question intonation patterns than when the vowel had a more conservative pronunciation. The vowel indicated a particular speaker demographic, which in turn affected the likely interpretation of the intonation pattern.
Our linguistic repertoire is being continually updated by our experiences. Even potentially negative experiences such as mishearings have the potential to alert us to the connections we need to forge between features of pronunciation and the characteristics of speakers. As listeners, we dynamically adapt to variation in speech features, and exploit this variation to make sense of what and who we are listening to.