Research/Clinical Interests:
Language acquisition
Psycholinguistics
Development of semantics-pragmatics
interface
Courses Taught in the Past Five Years
HESP 818: Topics in Language Development
HESP 400: Speech and Language Development in Children
HESP 300: Introduction to
Psycholinguistics
Recent Research/Clinical Activities:
Language development involves more than just learning words or syntactic
rules. In order to comprehend and produce utterances, young learners must
also acquire the ability to coordinate linguistic representations in
real-time. During this process, do children consistently generate the same
kinds of interpretations as adults do? When they fail to do so, what do
these differences reveal about nature of development? Over the past ten
years, my fascination with these questions has led to a research program
that explores how the moment-to-moment changes that occur during language
processing influence the year-to-year changes that emerge over the course of
language development. This work relies on the use of eye-tracking, an
innovative method that yields implicit and fine-grained measures of
children’s interpretation, and focuses on changes in development that occur
during middle childhood. By five years of age, children have learned the
grammatical structures of their language. But many of their cognitive
capacities are not fully mature: Unlike adults, they have smaller memory
spans, slower processing speeds, and poorer inhibitory control. My research
explores how limitations in these domain-general capacities interact with
language development.
The following examples illustrate the scope of this work:
1. Semantics-pragmatics interface. While communication involves both
understanding the literal meanings of a speaker’s utterance and generating
inferences to capture his/her intent, sometimes the boundary between the two
is unclear and counterintuitive. For example, in the case of quantifiers,
theories of language posit that the meaning of an utterance like “some of
the cookies” encompasses the meaning of all of them. Yet in everyday
conversations, there is a strong intuition to interpret some as implying
not all. How do these levels of representation interact during language
comprehension and acquisition? A priori, the sheer frequency and ubiquity
of the inference made it unlikely that adults (and possibly children) would
ever entertain the less common and less precise meaning. Nevertheless, our
research has found that both adults and children initially interpreted some
with respect to its semantics. Critically unlike adults, children never
invoke the late-emerging inferential process. These findings suggest that
inferences that require the coordination of multiple representations during
real-time comprehension may be less reliable in populations with limited
cognitive resources. Current studies examine other aspects of the
semantics-pragmatics interface, including how children predict the meaning
of utterances based on referential cues in the scene.
2. Cross-linguistic comparisons. English-speaking children have been
shown to have problems with the comprehension of passive sentences like “the
seal is eaten by the shark.” While previous theories have attributed these
difficulties to an immature grammatical system, we recently explored the
degree to which children’s failures reflect the challenges that passives
pose for incremental language comprehension. Passive and active sentences
cannot be distinguished until after the verb, creating a temporary ambiguity
in the grammatical role of the subject (agent vs. theme). Because passives
occur less frequently, they may be initially misanalyzed as actives and
require later revision following the onset of the verb. This ambiguity is
less pronounced in languages like Mandarin where morphosyntactic markers can
identify the roles of arguments early in an utterance. We found that
Mandarin-speaking children efficiently use these markers to make on-line
predictions about role assignments. However, children’s off-line actions
revealed that they consistently misinterpreted passives when it required
revision of an agent-first bias. This suggests that developmental
difficulties with passives stem from children’s incremental grammatical
processing and their failure to revise initial interpretations. Current
studies examine how cross-linguistic differences in word order and
linguistic cues can offer unique insights into the role of processing during
language acquisition.
3. Word recognition. Recognizing words in spoken utterances requires the
coordination of multiple procedures, including the mapping of speech onto
phonemes, the activation of lexical entries/concepts, and the interpretation
of the utterance with respect to the visual scene. In adults, these
processes occur rapidly and efficiently but how does this cascaded
processing develop? To address this question, we presented 5-year-olds with
pictures of logs (target) and a key (competitor) and asked them to “pick up
the logs.” If cascaded processing is an architectural feature that is
present early in development, then speech input from the target (“lo-“)
should lead to access of an absent phonological cohort (lock) which in turn
should activate a semantically-related competitor (key). In contrast, if
this ability depends on extensive linguistic experience, then children’s
looks to the competitor may not differ from an unrelated control item. We
found that both children and adults showed increased looks to the competitor
following target onset. However unlike adults, children also had a small
but reliable tendency to erroneously select this competitor in their
actions. These findings demonstrate cascaded processing in children’s word
recognition and suggest a developmental difference in the inhibitory
mechanisms that rule out competing lexical entries.
4. Reading development. Research on reading development has traditionally
focused on the acquisition of domain-specific knowledge such as letter
recognition or phonological awareness. However reading for comprehension
also requires the on-line coordination of several cognitive and linguistic
processes including visual encoding, phonological/orthographic analysis, and
lexical retrieval. The current research aims to understand the role of
processing in reading development by bringing together a traditional
paradigm from early literacy research (the Rapid Automatized Naming task)
with methods from psycholinguistic research (eye-tracking while
reading/speaking). This innovative approach will offer a fine-grained
measure that is sensitive to changes in encoding for individual items in the
array, providing more detailed descriptions of how factors influence the
execution of various component processes. It will also offer a continuous
measure that is sensitive to changes in performance over adjacent items in
the array, providing more detailed descriptions of how the processing of a
current item affects the processing of a subsequent item. Thus rather than
expressing variability along a single final response time, variability can
be captured along multiple dimensions. This opens the door for richer
descriptions of how processing strategies may differ across children and
over time.
back to top of page
