Page 397 - Encyclopedia of Philosophy of Language

pline itself. One such is the island constraint phenom- enon displayed in the interpretation of elliptical fragments.
Recall first (23) (repeated here):
Joan wants Sue to visit Bill in hospital. And Mary too.
(23)
We have in the interpretation of the first sentence some complex database leading to the conclusion:
want' (visit' (Bill') (Sue')) (Joan')): t.
With the fragment and Mary too, the hearer faces the goal-directed task of reaching some conclusion a : t but here he has only Mary provided as input. In order to arrive at some conclusion, he must create the necessary prepositional structure. To do this, he reuses the entire previous clause, creates a one-place predicate out of it, and reapplies this result to the new argument Mary. This step of building a lambda- abstract is licensed because it is a move from some premise of the form a : t to a premise of the form /lx(a): e -»• t, a perfectly licit step of conditional-intro- duction. Depending upon which position is abstracted over, we can create for (23) any of the interpretations (a)-(c) above. Relative clauses being two such local reasoning structures linked together through a com- mon variable, the relative-clause island phenomenon follows directly:
Joan visited the woman who likes Bill in (24) hospital. And Mary too.
We build for an interpretation of the first sentence of (24) the linked database:
Joan': e
visited': e -> (e -»/)
u:e
in-hospital' :(e-*t)-*(e-*t)
T
T
u:e likes':e-*•(e-»•/) Bill':*
To reconstruct the fragment we need to take one step of conditional introduction, but on which local struc- ture should we carry out this step? In order to create the complex reading substituting Mary in place of Bill in the relative clause, we would need to carry out conditional introduction on the first of these local structures (corresponding to the matrix clause). But the premise which we wish to withdraw is not there— it is only in some separate, albeit linked, structure. But conditional introduction is a local step of reason- ing—it cannot be vacuously carried out in one struc-
ture as a record of some such step in another structure. The logic itself precludes any such interpretation. Hence the island constraint phenomenona are a direct consequence of the logic discipline adopted. The apparent puzzle of interaction between grammar- internal constraints and the pragmatic process of utterance interpretation is resolved. Syntactic phenomena are explained not by properties defi- nitional of a discrete encapsulated language faculty but by the logic discipline in which the language fac- ulty is embedded. But to do so, we have had to aban- don Lewis's stricture and set up a syntactic concept of interpretation, manipulating semantic constructs (such as the type vocabulary) as expressions in a cal- culus for which inference is syntactically defined.
Within this new perspective, we are able to retain both the concept of a universal human capacity for constructing and manipulating structural con- figurations through the medium of language; and the concept of parametric variation between languages. Each language is a logic, with its own internal con- straints on how the building up of proof-theoretic structures is controled—with idiosyncratic locality restrictions (parametric variation in anaphor binding, for example), with idiosyncratic specification of how one structure is linked online to another (parametric variation in the value of wh- dependencies), and so on. But any one such logic falls within the general family of logics defined by the 'labeled deductive sys- tem': all such logics are natural deduction systems which model our ability to take elementary concepts, progressively build them up to form complex struc- tured concepts, and then reason with those complex structures as wholes. Our innate language capacity is, that is to say, firmly embedded in our capacity for reasoning.
Bibliography
Barwise J, Perry J 1984 Situations and Attitudes. MIT Press, Cambridge, MA
Chierchia G 1995 The Dynamics of Meaning. University of Chicago Press, Chicago, IL
Chierchia G, McConnell-Ginet S 1990 Meaning and Gram- mar: An Introduction to Semantics. MIT Press, Cambridge, MA
Chomsky N 1982 Lectures on Government and Binding. Foris, Dordrecht
Chomsky N 1986 Knowledge of Language. Praeger, New Y ork
Cooper R 1979 The interpretation of pronouns. In: Heny F, Schnelle H (eds.) Syntax and Semantics, vol. 10. Academic Press, London
Cresswell M J 1985 StructuredMeanings. MIT Press, Cam- bridge, MA
Dowty D R 1979 Word Meaning and Montague Grammar. Reidel, Dordrecht
Dowty D R, Wall R E, Peters S 1981 Introduction to Mon- tague Semantics.Reidel, Dordrecht
Evans G 1980 Pronouns. Linguistic Inquiry 11:337-62
u:e
woman': e -»t
Semantics versus Syntax: Perspectives on Natural Language Content
375