|
A Parallel Architecture for the Syntax-Semantics Interface| old_uid | 4112 |
|---|
| title | A Parallel Architecture for the Syntax-Semantics Interface |
|---|
| start_date | 2008/02/18 |
|---|
| schedule | 10h-11h30 |
|---|
| online | no |
|---|
| details | Suite à 11h30 |
|---|
| summary | In the `syntactocentric' EST/GB model of the 1970's and 1980's, it was assumed that SS `feeds' LF; in somewhat more current parlance, overt syntax precedes Spell-Out and the covert component follows it. By contrast, in constraint-based architectures, such as LFG (Bresnan and and Kaplan 1982), HPSG (Pollard and Sag 1987, 1994), and Representational Modularity (Jackendoff 1997), there is no notion of `feeding'; instead syntax and semantics (inter alia) are parallel, mutually constrained structures.
The two approaches are hard to compare because the constraint-based architectures lack any notion of derivation. However, what seems to have been overlooked by almost everyone (except Lecomte and Retor'e 2002), is that it is entirely possible to embrace parallelism without rejecting derivationality. In this talk I will give a sketch of CONVERGENT GRAMMAR, a grammar framework in which syntactic derivations (roughly, overt syntax) and semantic derivations (roughly, covert computations) proceed in parallel `all the way down', i.e. even at the level of the lexical entries. From a present-day perspective, this may look like phase theory carried to its logical extreme, but in reality it is an updated version of the `rule-to-rule' approach employed in the Extended Montague Grammar paradigm (e.g. Bach and Partee 1980). |
|---|
| responsibles | Aroui |
|---|
| |
|