Benjamin K. Bergen

Research

Simulated Action and Perception in Language Understanding
Along with colleagues here and at UC Berkeley, I have been doing behavioral research on the idea that in order to produce or understand meaningful language, language users run a mental simulation of (that is, mentally imagine) perceptual and motor components of the content of the utterance. Our evidence suggests that running this simulation involves activating the same brain structures that are responsible for perceiving or performing the events described in the utterance. A major theoretical result of our findings is that they support a view in which linguistic units do not in and of themselves have meaning. Rather, they evoke meaning - they are cues that contribute to a blueprint, which the speaker provides to the hearer, so that the latter can run an internal simulation with the desired properties.
Radial categories of grammatical constructions
In collaboration with Madelaine Plauché, I have extensively documented deictic and existential constructions in French, in particular those making us of voila, voici, and il y a. We have proposed cognitive and functional explanations for the distributions of and relations between these constructions in french, for cross-linguistic differences between deictic and existential constructions, and for historical changes in these constuctions.
Embodied Construction Grammar
With colleagues in the Neural Theory of Language research group at the International Computer Science Institute, I have worked on developing a theory and implementation of Embodied Construction Grammar (ECG). ECG can be alternatively seen as either a more cognitively plausible and computationally tractable theory and implementation of Construction Grammar or a more formal and quantitative version of Langackerian/Lakoffian Cognitive Grammar. ECG is embedded in a larger natural language understanding device, which uses linguistic constructions along with contextual knowledge to parse natural language utterances and extract a deep semantic representations from them.
We're holding a theme session on ECG at the upcoming ICLC conference. A list of talks, their abstracts, and a session summary are available here.
Phonaesthemes
Phonaesthemes (sometimes spelled "phonesthemes") are sounds or sound sequences that correlate with some meaning in a language, but are not themselves morphemes. Examples are English gl- and sn-, which occur in a disproportionate number of words with meanings related to LIGHT/VISION or NOSE/MOUTH respectively. In my work on phonaesthemes, I have tested the psychological reality of phonaesthemes through priming experiments, and have found that disproportionate representation of sound-meaning pairings like gl- and sn- give rise to priming effects that are very similar to priming effects produced by classical productive morphemes. I conclude from this that productivity is not an essential factor for the psychological reality of form-meaning pairings.
Probabilistic Perception
In recent work, I have investigated linguistic and social factors that influence phonology, and which do so probabilistically. In a study of French liaison, I have shown that linguistic and social factors interact in their influence on a morphophonological rule. Moreover, I have shown through psycholinguistic experimentation that individual language perceivers demonstrate that they have internalized these probabilistic interactions through their unconscious use of them when listening to language. I have also shown how Belief Networks, neurally plausible, probabilistic computational models, can be deployed to capture this linguistic knowledge.
Native Esperanto
Esperanto is an artificial language, invented by L. L. Zamenhoff at the end of the 19th century. It is spoken not only as a second language by its proponents, but also by a number of their children. Some of these children even speak it as a native language. I have done fieldwork with these native Esperantophones and have uncovered a range of nativization effects. Some of these are very similar to nativization effects observed during creolization of pidgins.
Press on my native Esperanto research

Some recent papers pertaining to these areas of research are available for download. You can also consult my full academic C.V.