Skip navigation

You are in:  Home » Postgraduate research » Postgraduate research in speech and language

Postgraduate research: Speech, Language and Hearing

The overall challenge for the development of a cognitive neuroscience of language is to relate functional accounts of the language system, behaviourally based and cognitively specified, to emerging accounts of the underlying neural system, based on neuropsychological and neurobiological inputs. This forms the background for the CBU Speech and Language group's current research programme, weaving together a variety of cross-disciplinary strands. These include (i) research aimed at the cortical localisation of language function, ranging from neuropsychological work with brain-damaged patients to modern neuro-imaging research (ii) cognitive and psycholinguistic behavioural research into the functional structure of the language system, incorporating influences from linguistics and from computational modelling (iii) psycho-acoustic and acoustic-phonetic research into the basic properties of human speech analysis systems, incorporating both behavioral and neuro-imaging techniques (iv) insights and constraints derived from the neurobiology of the relevant brain systems, and (v) research from a cross-linguistic perspective, essential to sort out the general from the particular in the functional and neural realisation of any one language system.
Specific project areas in the Speech and Language Group are listed below (typically these involve collaborative research between several Unit scientists):

  1. Auditory scene analysis, speech perception, and attention, focused on the question of how the listener extracts the speech of a given talker from a background of other sounds - known as auditory scene analysis (ASA) - and investigates the neural basis for ASA and its relation to higher-level cognitive processes. This research also provides a basis for the study of ASA in a clinical population - users of cochlear implants.
  2. Neural and functional foundations of speech and language: From speech to meaning, is concerned with the higher-order functional properties of the language system, focussing on speech perception, lexical access and representation, and on the deployment of lexical and semantic information during on-line interpretation of spoken utterances. We approach this work using multiple methods from cognitive neuroscience, including event-related fMRI and EEG/MEG studies that investigate the distributed cortical processes engaged during spoken language comprehension, computational modelling of word recognition, and through a range of behavioural and neuropsychological in different patient populations and languages.
  3. Brain mechanisms of words, meanings, and syntax, uses EEG and MEG to address a range of issues in the neurophysiology of language, focusing on the spatio-temporal patterns of brain activity elicited by words and sentences, studied by the use of multimodal neuroimaging methods.
  4. Perceptual learning and plasticity in spoken communication, explores the formidable capabilities of the human speech perception system for perceptual learning and adaptation. We combine behavioural and neuroimaging studies to explore the mechanisms by which the perceptual system adjusts to phonetically unfamiliar or degraded speech (with a focus on processes likely engaged for new users of a cochlear implant), and the neural systems involved in initial learning and consolidation of newly-learnt spoken words (with application to vocabulary acquisition in first and second language learning).
  5. Short-term memory systems and language, works towards developing an integrated account of the storage and processing of serial order in short-term memory and in spoken word-recognition.

Details of our current supervisors and their topics follow below:

Bob Carlyon

Cochlear implant users and people with sensory hearing losses often experience great difficulty in listening to one person when others are talking at the same time. In contrast, most normally hearing people are capable of attending to one voice in the presence of competing speech, even when all the voices present come from the same location in space (eg: a radio set). In many respects, the surprising thing is not that impaired listeners have problems with this task, but that normal listeners do so well: it requires the listener not only to group together the frequency components of the desired voice, but also to perceptually separate them from other speech sounds with similar frequency spectra and temporal characteristics. Fortunately, speech, along with other sounds, contains useful cues which listeners can exploit in order to perform the appropriate grouping. A possible PhD project would investigate either how normal listeners process such cues, and/or how one could encode these cues in a cochlear implant. This latter part of the project would benefit from the specialist hardware and software for testing implant patients that are available at the CBU, and from our close collaboration with Addenbrooke's hospital. In addition, recent research conducted jointly with Dr. Friedemann Pulvermuller's group has used EEG techniques to study the neurophysiological basis of auditory grouping. Hence the possibilty exists for a student to undertake a Ph.D. using both auditory and electrophysiological techniques. For more information please see my personal pages or our group page.

Matt Davis

My research is concerned with the cognitive and neural processes involved in perceiving and understanding spoken and written language. The approach taken is multi-disciplinary, combining methods from experimental psycholinguistics, functional neuroimaging (fMRI, MEG/EEG) and computational (connectionist) modelling. My recent research has sought to characterise processes of learning and adaptation that alter the way in which spoken stimuli are processed. For instance, repetition priming experiments have shown rapid and long-lasting changes in neural responses to primed words that predict the magnitude of behavioural facilitation. Work on the perception of speech that has been artificially distorted points to processes of perceptual learning that allow the comprehension of distorted speech to improve with certain forms of exposure. Finally, neuroimaging studies of vocabulary acquisition seek to determine the neural correlates of consolidation processes that have been suggested for newly acquired words. In all this work, a combination of behavioural and neuroimaging investigations allow a more detailed understanding of underlying processing mechanisms. This knowledge aims towards functionally and neuroanatomically-realistic models of processes that are fundamental to powerful and flexible spoken communication. For more information please see my personal pages or our group page.

Olaf Hauk

I use a multi-modal imaging approach to problems in the neuroscience of language. Although the term "neuroimaging" is usually associated with spatial "maps" or "tomographies", timing information is crucial in order to characterise the processing stages at which effects occur. For example, if motor cortex is activated when a subject sees or hears the word "kick", does this mean that this brain area is required in order to understand this word, or is it just activated because the subject is actively imagining the last time she played football? I address questions like these using electro- and magnetoencephalography (EEG/MEG), functional magnetic resonance imaging (fMRI) and transcranial magnetic stimulation (TMS). Specific projects address the timing of different processing stages in visual word recognition, in particular the neuronal representation of words referring to actions, and the role of mirror neurons in action word comprehension. I am also interested in developing and evaluating new analysis methods. For example, localising brain activity from the surface recordings obtained by EEG/MEG requires well-justified modeling assumptions (the "inverse problem"). These methods have to be understood theoretically, and tested in simulations and systematic experiments. For more information please see my personal pages or our group page.

Dennis Norris

A major theme running through much of the research on spoken word recognition conducted at the Unit in recent years has been the problem of recognising words in continuous speech. In contrast to written language, where there are white spaces between the words, spoken language contains few reliable cues to the location of word boundaries. How do listeners solve the problem of recognising spoken words without first knowing where the words actually are in the input? We have been tackling this, and related problems in spoken word recognition by a mixture of conventional experimental work, both on English and on other languages with contrasting phonological properties, and by the development of a large scale connectionist model of spoken word recognition. Other work concentrates on understanding how spoken words are represented in the mental lexicon and on the issue of whether lexical information can influence the processing of sub-lexical units such as phonemes. PhD work in this area would involve experimental work on spoken word recognition but could also involve computational modelling. We are currently developing a new connectionist model of word recognition and vocabulary acquisition.

A second programme studies short term memory and the relation between short- and long-term memory. Most of this work focusses on memory for serial order in short-term memory, and on how short-term memory contributes to long-term learning. Much of the experimental work is driven by computational models of short-term memory (Page and Norris, 1999). The computational work continues to generate a range of experimental predictions which could form the starting point for a PhD. The PhD could itself include computational work to extend the scope of the model.

Finally we are interested in building on the work in both speech and memory to develop a better understanding of both the memory systems underpinning language use, and the way in which the requirements of language have shaped the memory systems themselves. For more information please see my personal pages or our group page.

Yury Shtyrov

The main research focus is on the processing of spoken language in the brain using modern neuroimaging tools. This includes neural mechanisms and time course of language processing in the brain, covering levels from phonology to syntax and semantics.

Particular stress is made on automaticity and attention control in language comprehension. A lot of research we have done concentrated on uncovering automatic aspects of the cerebral language processing, i.e. those taking place outside the focus of voluntary attention. This helps us in defining distinct temporal stages in speech perception by the brain, and to see how language, as a function, interacts with other cognitive systems.

Currently, we are putting special efforts into investigating linguistic processes in various clinical conditions. The choice of topics here ranges from healthy ageing to various brain disorders (including dementia and psychosis), where we aim at looking for biomarkers of underlying functional neural deficits.

PhD supervision can be offered in a number of areas such as:

-Brain mechanisms of language processing (incl. laterality)
-Automatic cerebral language processing outside the focus of attention
-Neuroimaging of syntactic and semantic processes using EEG/MEG/fMRI;
-Neural memory traces for individual words
-Interaction of language processing and attention
-Neuroimaging of language function in brain disorders

Most of this research is done in the Cognition and Brain Sciences Unit's MEG, EEG and fMRI laboratories. The main CBSU collaborators are Dr. Olaf Hauk, Prof. Bob Carlyon, Prof. Karalyn Patterson, as well as other CBU scientists. Other collaborations in Cambridge include University of Cambridge, Addenbrookes Hospital and GlaxoSmithKline. The latter two are particularly involved in more clinically-oriented projects. Further MEG and EEG work is carried out in collaboration with the University of Helsinki, Finland. For more information please see my personal pages or our group page.