Ferreira & Patson (2007): The 'Good Enough' Approach to Language Comprehension

Ferreira and Patson (FP) give an overview of their "Good Enough" view of language processing, which holds that "... the language comprehension system creates syntactic and semantic representations that are merely ‘good enough’ (GE) given the task that the comprehender needs to perform. GE representations contrast with ones that are detailed, complete, and accurate with respect to the input." This is a very sympathetic idea, esp. when considering FP's argument against 'unbounded rationality'. A system that arrives quickly at the correct interpretation most of the time using low-level heuristics might has an evolutionary advantage over a system that arrives at the correct interpretation all of the time, but does so very slowly.

When we accept that the existence of time pressure on the processing side might lead to a local-heuristics system, there are profound consequences for the cooperative speaker. Assume that the cooperative speaker knows the hearer is like himself in that, depending on the task, she will use certain heuristics to infer the most likely speaker intention. If this is the case, the speaker will adjust his verbalization in such a way that the heuristics can lead to the hearer inferring the speaker's intention. This can be done by using the usual suspects: minimize code length while maximizing the desambiguation potential (MDL, basically), use as much shared code as possible (convention). At the same time, the multiplicity of construals of a situation leading to the same, simple code allows the speaker to manoeuvre the message for his own benefit.

(Aside: doesn't the speaker perspective help explain language's dislike of pleonastic verbalization: the speaker is unduly burdening the hearer's memory.)

What would this mean for language acquisition? If the developing child understands himself to be a communicative agent, he will (at a certain age) also understand others to have the same properties. Therefore, assuming an utterance U = (w1...wn) uttered given a (hypothesized) set of intentions I, the child will expect
1) that the elements of intentions that signify it (cf. Verhagen 2009) to be the minimal set of smaller meaning constructs that maximally distinguishes the intention i from all other intentions i' ∈ I
2) that this set of verbalizable semantic elements at the same time maximizes the intention's potential to be expressed: that is, the elements will be the most conventional and entrenched ones.

(connections with Chafe's 1970 linearization & deletion ideas, Chafe's observed asymmetry between production and perception, metonymy-as-grammar in Verhagen 2009)