Studying the Neural Bases of Real-time Language Understanding
Our research investigates the cognitive and neural mechanisms that allow humans to understand language. Communicating with human language is a signature capacity of our species and debilitating when disordered during development or following brain damage. Although neurologically normal humans understand the words and sentences of their native language with an automatic ease, this ease belies complex processes that allow understanding to occur. We use neuroimaging and behavioral experimental techniques to study these processes, with the aims of advancing basic science and guiding treatments for disordered language and reading.
Electrophysiological investigation of language comprehension
Our primary research tool is the analysis of the brain’s electrical activity, recorded from a network of electrodes placed on the scalp (the electroencephalogram, or EEG). We record EEG activity while experimental participants read or listen to words and sentences. By averaging across trials of EEG activity that occur during a given type of cognitive event (e.g., reading a word), we extract a multi-dimensional and time-varying signal that reflects the brain’s response during the event of interest, while “averaging out” aspects of the recording that vary randomly from trial to trial (the noise). This signal-averaged measure is called the Event-Related Potential (ERP) and provides millisecond-level sensitivity to neural events. The excellent temporal resolution of ERP is a crucial aspect for our research program, since we study the transient neural events involved in recognizing and integrating words in sentences, which is something humans do at the rate of 3-4 times per second. We are also working to apply algorithms that estimate the anatomical generators of scalp-recorded EEG data, allowing a clearer spatial-temporal understanding of the brain responses that underly language processing.
Understanding Words and sentences
Current research in the lab is focused on two main issues. First we investigate the critical events that occur in the recognition of a written word (reading), focusing on the initial ~250 ms after the eye has encountered the stimulus. While early thinking has characterized this initial period of word recognition as concerned with extracting sensory representations of the word, we find that early-stage word recognition is influenced by higher-level (grammatical and semantic) representations, and by predictive influences. This is especially true when we study word recognition in the context of a sentence, as opposed to words in isolation. The second strand of our research is the question of how representations of words are combined to form larger representations like sentences. We investigate the processes associated with grammatical and semantic levels of psycholinguistic analysis and how those processes are coordinated in real time to allow language interpretation.