Semantic Valence Tendencies in
Monolingual and Bilingual Sentence Comprehension
Luca Onnis - Principal Investigator
Abstract
A series of five studies is proposed to examine sentence comprehension in three groups of participants:
adult native monolinguals, early and late bilinguals. It is hypothesized that fluent sentence comprehension
can be achieved because proficient native speakers possess knowledge of generalized units of meaning
larger than the word. For instance, in English, the verb 'provide' typically precedes positive words (e.g.
'provide work') whereas 'cause' typically precedes negative items (e.g. 'cause trouble'). These statistical
patterns form units of meaning that imbue lexical items, and their argument structures, with semantic valence
tendencies (positive/neutral/negative connotations).
This proposal examines the impact of semantic valence tendencies on fluent sentence comprehension. To
do so, first an accurate algorithm will be developed to assess the pervasiveness of valence tendencies in
large corpora of English and Italian. Second, the proposal examines whether fluent reading is affected in
sentences containing a violation of a valence tendency (e.g. 'cause optimism'). Third, the proposal examines
fluency in bilinguals. Even very proficient second-language speakers seem to lag behind native speakers
specifically in the degree of knowledge of language-specific selectional restrictions. It is proposed that part of
the difference between early and late bilinguals' fluency in sentence comprehension may be in the
processing of extended units of meaning such as 'cause + negative valence tendency'. The broader scope of
this work is thus to contribute to better teaching practices that develop native-like fluency in second language
learners. Fourth, the proposal examines whether both mental lexicons in bilinguals are activated concurrently
when valence tendencies are processed.
The research has direct implications for instructional methods used in second-language training, as being
aware of statistical tendencies toward differential semantic valences for certain cognates may greatly
improve second-language fluency. The research will also be informative for the further development of
automated translation systems. And it will expose undergraduate research assistants, including ethnic
minorities, to laboratory training in the language sciences.
Please email suggestions/errors to lo35@cornell.edu |