CNL Lunch

Michael Israel

Four Kinds of Creativity in Child Language

Thursday April 11th, 12:30pm, 3416 Marie Mount Hall

     Creativity poses a problem for any theory of language acquisition (Braine 1971, Baker 1979). The problem is that children not only learn how to form novel grammatical utterances, but also, crucially, that they learn to avoid novel combinations which might seem plausible but are in fact ungrammatical. My thesis for this talk is that children manage to do this by adhering to two basic PRINCIPLES OF CONSISTENCY as constraints on analogical learning in a usage-based grammar (Langacker 1988, Tomasello 2000). Local consistency applies to utterance tokens, and requires that novel tokens closely match formal and semantic features of at least one entrenched utterance type. Global consistency applies to the repertoire of constructions as a whole, and requires that inheritance links between constructions should closely match other links established in the grammar. Together the principles allow flexibility in individual utterances but constrain the overall shape of a child's emerging grammar.

I will begin with a brief overview of Cognitive Grammar (Langacker 1987; 2000) and the idea that linguistic competence is grounded in speakers‚ experience of actual linguistic usage˜that grammar is, in an important sense "usage-based." I will then examine the emergence of a family of related constructions involving non-finite clausal complements (NFCCs) in the spontaneous speech of seven English speaking children between the ages of 1.6 and 5 years. As the examples in (1-3) show, NFCCs appear early with a variety of matrix verbs and different predicative complements, all of which conform to the basic schema in (4).

1. that [make [it dark outside]]. Eve 2;0
2. [put [man on the shelf]]. Nina 2;1
3. I‚m gonna [get [the cow to drink some milk]]. Peter 2;8
4. (NP1) [Vmatrix [NP2 X-Comp]nfcc]

As a family, NFCCs are united by important surface similarities and divided by
a host of formal and lexical idiosyncracies. As such they offer abundant
opportunities for false analogies and overextensions, and so form an ideal
domain to examine children‚s creative use of syntax.

Four basic types of creative utterances are found across subjects: groping constructions, flexible routines, interference effects and blended constructions. Groping constructions (cf. Braine 1976) occur early in acquisition and tend to feature the most anomalous combinations (e.g. I sock put on, Eve 1;7): this follows from global consistency, as constraints on constructions are weakest early on when there are fewer constructions in a child‚s repertoire. Flexible routines emerge as specific uses of particular matrix predicates become entrenched and begin to combine with progressively more complex complement types and embedded structures. This follows from local consistency, with novel structures built on well-entrenched utterance types. Interference effects begin to appear as children‚s repertoire of constructions expands and related construction types compete to license usage events. Such mixed uses, which might be thought of as simple performance errrors, show a growing awareness of relations among NFCC constructions, in accord with global consistency.

Finally, I consider a variety of child-invented constructions, including two which achieve unit status and persist for over a year. Both of these˜Nina‚s wear X on construction and Sarah‚s erase X off construction˜involve novel blends of constructions which are both locally consistent with entrenched utterance types and globally consistent with the repertoire of constructions as a whole.

The results show that early child grammar involves both rote-learning and complex creativity, and that these are complementary rather than antithetical processes. Children need not follow an essentially conservative learning strategy; rather, children do over-extend their grammars, but only when the principles of consistency give them good reason to.