On feature-based vocabulary selection mechanisms in late insertion: third person number agreement in the French future tense

Abstract

We propose that feature bundles derived in syntactic computations activate congruent vocabulary entries inducing feature-based conceptual-structure processes in retrieval. Thus, for the French future tense, an inflectional node baring Number: Plural activates the forms mangera (EAT-FUT.3PS.SG) and mangeront (EAT-FUT.3PS.PL), which compete for insertion following the Subset Principle of Distributed Morphology. Indeed, the affix -a (3PS.SG) encodes Number with no further specification (notated Number: Ø), whereas -ont (3PS.PL) encodes Number: Plural, where Number: Ø is a subset of Number: Plural. This feature structure defines an information scale where plural-marked -ont is stronger. On this scale, informationally weaker -a (3PS.SG) is interpreted as [-Plural] in contrast with -ont via a scalar inference, becoming unsuitable for insertion. Thus, -ont (3PS.PL) is selected when -a (3PS.SG) is eliminated. We present evidence of conceptual-structure processing linked to underspecified morphology. In forced-pace reading and listening tasks, 19 native speaker subjects per task classified picture probes accompanying matching and mismatching subject-verb future tense agreement. Classification times for pictures semantically linked to the verb probed for an interaction between the processing of agreement morphology and the ongoing conceptual processing of the sentence. Classification times were modulated by the type of morphological mismatch. Singular verb form mangera (EAT-FUT.3PS.SG) slowed down picture classifications in plural contexts, whereas plural verb form mangeront (EAT-FUT.3PS.PL) in singular contexts did not. This interaction between purely formal agreement and conceptual-structure processing is unexplained by interface relations, frequency, information load, and phonological cohort activation. It suggests that domain-general principles of inference enrich domain-specific feature-based computations.