Welcome to the University of Victoria’s Speech Research Lab!
The University of Victoria Speech Research Lab was founded in 2007 through a Canadian Foundation for Innovation (CFI) Leaders Opportunity grant awarded to Dr. Sonya Bird. The SRL infrastructure includes a range of equipment for the collection and analysis of data on:
1) the articulation of speech sounds by speakers (articulatory phonetics),
2) the acoustic signal associated with these speech sounds (acoustic phonetics),
3) the perception of this signal by listeners (auditory phonetics).
The SRL supports various research projects related to phonetic variability, i.e. the variability that is intrinsic to the speech signal. On the speech production side, studying phonetic variability involves acquiring and analyzing acoustic and articulatory data on how speech sounds vary when they are uttered by speakers. On the speech perception side, studying phonetic variability involves acquiring auditory perception data on how listeners deal with the variability they encounter when listening to speech: what aspects of variability do they use to understand speech and its context, and what aspects do they treat as ‘random variation’ and ignore? Given the wealth of indigenous languages spoken in the areas surrounding UVic, this research program focuses on phonetic variability in these languages, which have so far received very little attention from phoneticians.
Research conducted in the lab and using lab equipment in the field has been supported by grants from SSHRC, the Jacobs Fund, The Philips fund, and the University of Victoria.
Project overview: Canadian Foundation for Innovation (CFI) infrastructure grant (2007)
At first glance, speech processing, or the production and perception of sounds in human communication, appears to be a straightforward process: the speaker produces a series of articulatory gestures to convey meaning; and the resulting sounds are perceived by the listener who extracts the intended meaning. Upon closer examination however, this seemingly simple process is fraught with complexity. This is because the way we produce speech is a bit like the way jazz musicians improvise: while there are rules and preferred patterns, no two performances are ever the same. In speech, each sound or sequence of sounds has its own unique acoustic “print”. This phenomenon, known as phonetic variability, is a fascinating puzzle with vast implications for our understanding of the human capacity for language.
Dr. Bird is a new faculty member at the University of Victoria (UVic) whose pioneering work on some of the most endangered languages in British Columbia has already contributed significantly to our knowledge of phonetic variability. The requested infrastructure – the UVic Speech Research Laboratory (SRL) – will support her groundbreaking research on the languages of the Pacific Northwest with the long term goal of understanding the role of phonetic variability in speech processing.
Phonetic variability is puzzling from the perspective of speech production and perception alike. On the production side, WHY is there so much variability in the way we speak? Previous work has shown that variability is to a certain extent predictable. Dialect is one factor that influences the way we pronounce sounds, e.g. the word ‘tomato’ is pronounced differently by American and British speakers. Linguistic context is another factor that accounts for variability, e.g. across many dialects of English, the ‘t’ in ‘top’ is pronounced differently from the one in ‘stop’. Although much of the natural variability in speech production can be traced to factors such as dialect or linguistic context, some variability is simply random: humans are not machines programmed to repeat the same sequences of movements (e.g. of the tongue) over and over again. As a result, both the movements themselves and the way in which they are timed with respect to one another differ from one utterance to the next, leading to variability in the resulting speech signal. Recent research on variability has come a long way towards clarifying the factors contributing to phonetic variability. However, the exact way in which these factors interact to create the speech signal that we eventually hear is a mystery that remains to be solved.
Variability in speech production leads to an equally captivating puzzle on the perception side: HOW can we make sense of the speech signal we hear, given that it is never exactly the same as anything we’ve heard before? Presumably, listeners learn to ignore unimportant differences in the signal, recognizing it as “close enough” to something familiar. The exact mechanism that underlies our ability to understand the speech signal we hear, despite extensive variability, is not yet understood. Unraveling of the role of phonetic variability in speech processing is essential to understanding the human capacity for language. While much important work has already been done in this area, it has been hampered by two methodological limitations. The first limitation is access to speakers of a broad range of languages. Studies on phonetic variability have tended to focus on widely spoken languages such as English, for which data is relatively easy to collect. If we are to fully understand the role of phonetic variability in speech processing, it is crucial to extend this research to the full range of languages spoken around the world. The second limitation is technological in nature. Until recently, many of the articulations relevant to speech have been impossible to observe directly, in particular outside of a laboratory setting, because the tools used to study articulation have been relatively invasive and non-portable. As a result, much remains to be learned about variability in articulation and how it correlates with variability in the resulting acoustic speech signal. The proposed SRL will include the complete range of equipment necessary to study the articulatory, acoustic, aerodynamic, and perceptual properties of speech sounds in an integrated manner. Located at UVic, it will be easily accessible to speakers of the Pacific Northwest languages spoken on Vancouver Island and the surrounding areas, allowing Dr. Bird to overcome both limitations discussed above.
On Vancouver Island alone, there are two language families, Wakashan and Salish, encompassing six distinct languages and approximately twenty regional dialects. These languages are known for their complex sound inventories, including many sounds that are very rare in other languages of the world. Despite the wealth of information on phonetic variability that these languages could provide, only one of them, Nuu-chah-nulth (Wakashan), has ever been studied phonetically in any detail. The SRL will create a unique facility on Vancouver Island, allowing Dr. Bird and her research team to study these languages which have so far received very little attention.
In combination with more commonly-used tools, Dr. Bird will employ ultrasound, a highly innovative tool for studying speech articulation. UVic is already at the leading edge of articulatory research: Dr. John Esling is one of only two researchers in North America to use laryngoscopy to study articulations in the lower regions of the vocal tract. His work has led to the discovery of many articulations that were previously unknown or poorly understood. Ultrasound, still very new as a research tool in linguistics, is ideal for observing articulations in the higher regions of the vocal tract: the transducer is placed under the chin, providing a clear image of tongue movement in real time. Using ultrasound, Dr. Bird will be able to obtain previously inaccessible information on the pronunciation of complex sounds. In collaboration with Dr. Esling, she will be able to provide – for the first time – a complete picture of speech articulation in Salish and Wakashan languages, from the vocal folds all the way up to the lips.
In summary, the proposed SRL will provide vital infrastructure for Dr. Bird’s study of speech sounds in the Pacific Northwest languages of Vancouver Island, which will advance our understanding of the role of phonetic variability in human communication. Given that these languages are critically endangered, it is essential to conduct this research immediately, before the opportunity is lost.