Emily's List

George Lakoff lakoff at COGSCI.BERKELEY.EDU
Wed Apr 26 06:51:39 UTC 2000

Emily Bender has been collecting answers to the question, "What is a
grammar a model of?" Here's mine:

A grammar of a language - including the lexicon - models the capacity of
native speakers to express ideas using the language and to understand ideas
expressed by others in discourse.

This entails:

(1) Since grammar is about speakers' capacities, it must take seriously
what is know about the mind and brain, including our knowledge of the
embodiment of mind. What is required is a cognitive grammar, consistent
with knowledge of how the brain works (e.g., consistent with what we know
about neural structure and neural computation)

(2) Serious cognitive semantics. Linguists are responsible for giving an
account of the ideas expressed - that is a theory of the conceptual systems
of speakers and of the concepts expressible in the language. This will
include what has been learned in cognitive semantics - Prototypes and
basic-level categories, radial categories, image-schemas, force-dynamics,
windowing of attention, frame semantics, conceptual metaphor and metonymy,
mental spaces, and conceptual blending.

(3) Dynamic simulation semantics: An account of dynamic mental simulation
of what sentences mean in context using background knowledge, with
inferences flowing from the simulation. This means that sentences must
contain enough information to characterize the parameters needed for such a

(4) Constructions: General mappings between surface phonological forms and
meanings in such a form that they can be used in real time. This requires a
construction grammar in which both form and meaning have a bodily
grounding. And it must work probabilistically to account for the facts
that:  (1) certain psycholinguistic effects (like garden-path sentences)
are dependent on frequency of lexical items, where frequency for a speaker
is presumably reflected in strength of neural connections, and (2) change
is in process,  with constructions having various degrees of entrenchment,
again presumably reflecting varying strengths of neural connections.

(5)  Grammars must be able to fit a theory of recruitment learning to mesh
with Chris Johnson's discoveries about the role of conflation in the
extension of constructions, polysemy, and metaphor.

(6) Whatever is universal in grammar should come out of what is universal
about the ideas expressed and the cognitive/neural capacities to express
and understand them online and to have them govern mental simulations in

                                                George Lakoff
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: text/enriched
Size: 2669 bytes
Desc: not available
URL: <http://listserv.linguistlist.org/pipermail/funknet/attachments/20000425/9d7664d5/attachment.bin>

More information about the Funknet mailing list