Research

Specific Aims

Our research group is interested in understanding how children acquire the vocabulary- and sentence-processing abilities that are necessary for speaking and understanding a first language.  At the most general level, our research explores how these “two” learning procedures – the one for the meanings of words, the other for how these meanings combine to yield the meanings of sentences – interact and mutually support each other in a bootstrapping process whose outcome is the mature language user.

Methods

Much of our research uses methods that are designed to study the moment-by-moment processes that comprise real-time language processing. Most notably, our lab has developed a set of child-friendly methods for recording the eye movements of children as they hear spoken descriptions of their surroundings.  This ‘visual world’ approach to studying spoken language comprehension (and production) allows us to obtain a moment-by-moment record of what a child considers are the referents for the utterance as the speech unfolds over time.  By providing children with utterances that are temporarily ambiguous (referentially, syntactically or semantically), we can gain insight into the sources of information children use to make interpretive commitments of linguistic material. Similarly, by providing children with utterances containing novel words (Oh, she’s blicking the ball) we can explore the evidential sources used by the child to learn the meaning of this new word.

Example from a Study

In the video below, the speaker on the screen uses a novel verb ("biffing") to describe the scene, which can be taken from different perspectives, either the dog's ("The dog is chasing the man") or the man's ("The man is chasing the dog"). This experiment was designed to see how the eyegaze of the speaker influences the listener's interpretation of the meaning of new words. The blue dot and its trail are the path of the watcher's/listener's eyegaze. As you can see, her gaze starts at the speaker and moves to examine the different actors in the scene.

Lexicalist Theory

Our research findings have supported a specific version of a process-oriented approach at both these levels: a constraint-satisfaction, lexicalist theory of both learning and language use.  By lexicalist, we mean a picture of the architecture of language knowledge in which structural and semantic representations exploited in speech and comprehension reside primarily at the site of lexical entries.  Thus our learning studies support and enhance independent evidence favoring lexicalized grammars now coming from linguistic, computational-linguistic, and adult psycholinguistic research.

Eyetracking Equipment

Tobii Remote Eyetracking System

The Tobii system has all of its components fully integrated into a flat panel display.  Both eyes are tracked by this system, and no eye tracking visor is used.  This system is surprisingly robust and works well under a range of viewing conditions and remains accurate even with fairly large head movements.  This system can be used with infants, though we are currently using ours with children 3 years of age and older.

ISCAN Head-Mounted Eye Tracking System

The ISCAN eyetracker uses miniature cameras and is worn as a light-weight visor on the head. Two cameras are used, one trained on the eye and one trained on the scene, as seen by the subject. The scene camera records the general direction that the subject is looking. The image from the eye camera is analyzed in realtime by a computer and computes the direction of gaze within the scene image.  This system is especially useful for situations in which we wish to record eye movements as participants carry out actions on co-present objects.

Video-based Eye Tracking System (the “poorman’s eye tracker”)

In this system, a videotape is made of the child’s eyes and face as he/she carries out spoken instructions to move objects about on a platform. The videotape is later analyzed by hand for gaze direction and word onsets.