Principal Investigator Kenneth Wexler
Project Website http://wexlerlab.mit.edu/
The Wexler ab/Normal Language Lab seeks to understand the nature of the computational system of human language in its many guises. We study most aspects of linguistic structure, including syntax, semantics, pragmatics and morphology. In pursuing these goals, we take as our primary linguistic data abnormal language, by which we mean nothing more than any system of language that seems to differ from standard adult language for biological reasons, including lack of maturation, difficulties in learning, and genetic variation.
Thus we study immature language in the child (the development of language), language learned at an unusually late age where there might be lack of plasticity (second language acquisition), aphasia (language loss) and unusual language due to genetic factors (Specific Language Impairment, Down & Williams syndrome, autism, etc). In addition, language learning (e.g. parameter-setting) is studied via precise computational models.
The lab integrates insightful ideas from current linguistic theory with intensive and extensive experimental investigations. At the same time, our theoretical and experimental results often feed back into and influence linguistic theory, for there is much evidence about the nature of the computational system of language that is unavailable to traditional linguistic inquiry, but is available under conditions of abnormal language.
The lab is responsible for many of the major results in the modern study of linguistic development.
The study of language is one of the most exciting and productive areas of modern cognitive science. The Wexler ab/Normal Language Lab is dedicated to the investigation of abnormal language, by which we mean nothing more than any system of language that seems to differ from standard adult language for biological reasons, including lack of maturation, difficulties in learning, and genetic variation. We stress both experimental and theoretical investigations. Our basic assumption is that these go hand in hand at the current stage of the study of linguistics. We study most aspects of linguistic structure, including syntax, semantics, pragmatics and morphology.
Many of the major advances in linguistic development have occurred in this laboratory over the years, including the demonstration of early knowledge of binding theory, the discovery of the exemption of bound variables from the Delay of Principle B Effect, the major properties of the Optional Infinitive stage of linguistic development and the demonstration that children at a young age don’t give the adult representation to unaccusative structures. The lab has recently been instrumental in establishing and explaining cross-linguistic variation in the omission of object clitics by children. At the same time theoretical and mathematical work that began with demonstrations of (non)learnability has expanded to computational work on parameter setting that includes a model of triggers. The formal testing of this model has set off a reawakening of computational learnability studies worldwide.
The style of the lab is to integrate advanced and insightful ideas from linguistic theory with an intensive and extensive experimental investigation, constantly testing hypotheses and expanding the range of empirical material covered. Thus our proposed explanation for the Optional Infinitive stage (the Unique Checking Constraint) has been recently extended to explain object clitic omission and its variation across languages. Clitics are omitted in French and Italian extensively, but much less so in Spanish and Greek.
At the same time, the lab has always been interested in discovering results in development and impairment that have important implications for linguistic theory. The result on the developmental difference between bound-variable and non-bound-variable pronouns in Principle B contexts is one early example. This is a result that any theory of binding has to be able to account for. Another example is the role of agreement in assigning nominative case during the Optional Infinitive stage (Schütze and Wexler), which lends strong credence to the hypothesis that agreement is crucial to nominative case. The current experimental investigation of the difference between wh and non-wh raised constructions (see below) has the potential to be as definitive as any investigation in the question of whether there is a ban on “improper movement”. The common denominator in these (but perhaps not in all) examples is that the child has a slightly different grammar than the adult, so the child allows us to perform an experiment (analogous to a gene-knockout experiment) on what happens when a certain aspect of grammar differs slightly from the adult form. Studying child grammar in this way has serious implications for the theory of Universal Grammar, the theory of how the brain specifies language.
One current major research project is pushing forward the investigation of the development of what have been traditionally called “Argument-Chains.” Ken Wexler has recently proposed the Universal Phase Hypothesis that is intended to cover the empirical range of the A-Chain Deficit Hypothesis (ACDH), but which solves an outstanding empirical problem of ACDH, and we are actively exploring the experimental implications of this theory. We have demonstrated that raising constructions are delayed in children through at least age 5, although there is no delay for the same verb in non-raising constructions. The theory seems to make a rather precise prediction that there will be a particular developmental difference between raising constructions, depending on whether or not they are involved in wh-movement, and we are currently carrying out an experiment addressing this issue. Other experiments are investigating other aspects of the theory, including the prediction that young children treat verbal passives in English as adjectival.
Another present focus is on semantics, in particular on the development of determiners, especially the word “the”. We are testing whether some well-known phenomena in this area are due to pragmatic considerations (the traditional “egocentric” theory) or to semantic ones (difficulties with Maximality). We hope to apply these lessons to some recent results under discussion in the field about differences in child and adult processing abilities (garden paths and so on).
Over the last few years, the lab investigated in some depth the properties of determiners in second language acquisition (see the thesis of Tania Ionin 2003, and papers by Ionin, Ko and Wexler). Results demonstrated that second language (L2) learners have problems deciding whether definiteness or specificity define the marked determiner. These results for L2 learners look different from what has been claimed for first language (L1) learners. But many of the relevant experiments have not been done on L1 learners, and we are just beginning to plan these. So an exciting area of research that spans both L1 and L2 is the development of determiners (and, of course, related semantic development). It looks as if this field is opening up again.
The theoretical focus of the lab involves the ideas of linguistic theory and learning theory, where our hope is to explain development in terms of brain-based abilities, so the idea of linguistic “growth”, “development” or “maturation” is fundamental, as it is in linguistic theory. Thus the interplay between learning (e.g. setting parameters, learning a lexicon) and biologically based (genetically conditioned) growth provides an important tension for explanation in linguistic development. One of the consequences of the Optional Infinitive stage is that children are brilliant learners of language-particular inflectional material. This was not known until our lab demonstrated it - the traditional assumption of all approaches to language development had been that children are very bad at learning language particular material like inflection. It turns out that language is probably the best of all fundamental cognitive areas in which to study learning. This is because language is an area where there clearly is learning (i.e. cross-linguistic variation). In other areas of the cognitive sciences, it is usually less clear whether fundamental cognitive processes are learned or arise from genetics. In language, though, we know that at least the parameters must be learned.
The question of linguistic growth leads naturally to the question of what happens when a human does not have full linguistic capacity. For many years we have been concerned with Specific Language Impairment (SLI), and in collaboration with speech experts in this field (Mable Rice) we have defined and tested a central hypothesis of SLI - that SLI children are in an Extended Optional Infinitive stage. There is now a standardized test that attempts to pick out SLI children on the basis of scientific research on this theory. The general results of this field lend further credence to the theory that the Optional Infinitive stage exists and ends due to biological (genetically-based) growth.
We are now expanding the range of our investigations in the study of language impairment. The lab is currently studying the status of the computational system of language in Broca’s aphasia, attempting to apply the results of developmental syntax to this area. A current focus is on whether the difficulties with A-chains/phases found in young children are replicated in agrammatic aphasics. We are experimentally studying the difference between passives of actional and non-actional verbs, a telltale sign of difficulties with these structures in young children, where young children appear to treat verbal passives as adjectival passives. Tentative results suggest that aphasics have the same difficulty as children but are less consistent in their use of the adjectival strategy. Similarly we are repeating our developmental raising experiments in aphasics.
As part of the Brain Development and Disorders collaboration between the Department of Brain and Cognitive Science at MIT and the Developmental Medicine Center at Children’s Hospital, Boston, we are beginning to investigate the development of language in autism and in Williams Syndrome. The plan is to study the abilities of children who have these disorders with respect to the computational properties of language (like raising, passive, binding theory, determiners) and also pragmatic aspects. We hope to understand the nature of the linguistic deficit in each syndrome, to compare it to normal language, SLI, Down syndrome, and to see what can be learned about the relationship between the impairment and linguistic theory. The range of current experimental data in this area is limited, so we see this as attempting to fill in the gaps, and to see if current hypotheses (e.g. autistic children mostly have cognition problems, nothing particularly delayed in computational aspects of language) can be maintained.
The lab considers all methodologies as potentially useful, so there have been times that we have used ERP and MEG, in collaboration with investigators that have the requisite expertise and technologies. We are currently considering branching out to new methodologies, including eye-tracking studies, fMRI, and TMS. The lab defines itself not by a particular methodology (although it has strong expertise in experimental and other empirical studies of language) but by its commitment to the fundamental question of how language develops in the human and the contributions of the brain and learning to this development.