The Generation of Continuous Semantic Constraints from Semantic Propositions

. THE GENERATION OF CONTINUOUS SEMANTIC CONSTRAINTS FROM SEMANTIC PROPOSITIONS

Gregg C. Oden University of Wisconsin Madison, Wisconsin

Language comprehension is an exceedingly complex process which requires the extensive use of many different kinds of information in order to be successfully accomplished. One potentially very important type of information which has to date been largely ignored is the ~b~ree to which possible interpretations are sens1 e. While the sensibleness of candidate interpretations has long been recognized to be important, sensible­ ness has usually been treated as if it were an all-or-none property. However, it is clear that many things are more-or-less sensible and, there­ fore, the relative sensibleness of alternative interpretations may well be extremely useful information. For example, Oden (1977) has argued that degree of sensibleness information is re­ quired in order to disambiguate sentences to obtain the meaning that people normally do and has proposed language processing mechanisms which would use this information.

The degree of sensibleness of an interpreta­ tion depends on the degree to which the semantic constraints of that interpretation are satisfied. Therefore, to account for the continuous nature of sensibleness, semantic constraints must be fuzzy restrictions (Zadeh, 1975). A semantic constraint will be defined to be a function asso­ ciated with a particular semantic relation which specifies, for every combination of semantic elements which may enter into that relation, the degree of sensibleness of the resultant semantic structure. The present paper outlines how such semantic constraints may be generated from the kinds of knowledge already represented in current semantic memory models (e.g., Norman & Rumelhart, 1975), plus the fuzzy predicates and operations which will be necessary in order to handle other problems like the continuousness of subjective

class membership.

Defining semantic constraints to be func­ tions makes it natural to think of complex seman­ tic constraints as being compositions of simpler constraints. Furthermore, since semantic con­ straints are considered to be bound to particular semantic relations, the decomposition of a con­ straint may be expected to parallel the decompo­ sition of its associated semantic relation. This appears to be what happens in most cases, but there are certain "configural constraints" which do not seem to be derivable from component con­ straints corresponding to primitive semantic relations (see Oden, 1977, for details). However, such configural constraints seem to be relatively exceptional and, consequently, semantic con­ straints will still be "cognitively economical." More importantly, it will be argued below that elementary (non-composed) semantic constraints, whether directly associated with primitive seman­ tic relations or configural, are based upon specific semantic propositions which would be in semantic memory anyway.

In a fundamental sense, all knowledge is constraining. For example, knowing that it is -15° today affects the sensibleness of the ·state­ ment "Maxine went swimming in Lake Mendota this morning." More generally useful knowledge speci­ fies information about the normal and/or necessary properties of things which may enter into parti­ cular case relations with particular verbs. The most elementary knowledge of this sort (selec­ tional restrictions) is often considered to be part of the basic meaning of the verb. However, the more interesting semantic constraints are those which are based on much less elementary · knowledge, such as that only people normally drive trucks, which we might represent as:

The obvious interpretation of the constraint exerted by this knowledge is that it only makes sense for someone to drive a truck if that some­ one is human. However, I propose that in fact the constraint is that the degree to which it makes sense for someone to drive a truck is equal (assuming that all other constraints are satis­ fied) to the degree to which it is true that t.he person is human:

In general, it is proposed that knowledge of the

form:

leads to corresponding semantic constraints of

the form:

that is, that the sensibleness of the proposition is equal to the degree to which the predicate is true for the specified arguments, where the pre­ dicate may be arbitrarily complex and may itself be decomposable.

In summary, it has been proposed (1) that semantic constraints are functions which specify the degree of sensibleness of semantic .structures, (2) that complex semantic constraints are composi­ tions of elementary semantic constraints and {3) that elementary semantic constraints are derived directly from knowledge about the expected pro­ perties of case nouns for particular verbs.