Developing Intelligence

Trueswell & Kim‘s paper in the Journal of Memory and Language describes a phenomenon known as “fast priming,” in which a reading task is momentarily interrupted by a brief presentation of a “prime” word, usually lasting around 30 to 40 ms. The reading task then continues, and although subjects are typically unaware of the presentation of this word (usually describing it merely as a “flicker”) the processing of subsequent words are influenced by many characteristics of the prime word, including its meaning as well as its sonic and orthographic characteristics.

Al Fin has an excellent overview of similar “fast priming” techniques.

Trueswell & Kim utilized this technique to show that part of speech determinations are carried out by a mechanism that is fundamentally similar to those which identify the meanings of words. In other words, word recognition appears to include automatic activation of information related to that word’s typical argument structure, and does so in a way that has a carry-over influence on the interpretation of subsequent words.

The authors demonstrated this surprising claim by presenting 32 subjects with 16 sentences of interest intermixed with 54 “distractor” sentences. Every sentence was interrupted by a fast prime – a word presented for 39 ms – occuring at some particular point during the sentence. Each of the 16 sentences of interest exemplified the Direct Object/Sentence Complement ambiguity, in which one word in each sentence is often first interpreted as a direct object but is subsequently reinterpreted as a sentence complement. For example, “the man accepted the prize was not going to him” is typically initially interpreted so that “the prize” is the direct object of “accepted,” until the word “was” requires that “the prize” is assigned a role as a sentence complement. Such sentences can be made unambiguous by, for example, inserting the word “that” after “accepted” and before “the prize.” 8 of the 16 sentences of interest were made unambiguous in this fashion, whereas the other 8 were left ambiguous. Four among each of these groups of 8 sentences included a fast prime word such as obtained, which usually takes a direct object, whereas the other four of each group of 8 sentences included a fast prime word such as realized, which usually takes a sentence complement. These prime words were always presented right before the word which required reassignment of the verb to accept a sentence complement as opposed to a direct object (either “that” or “was” in the example above). Finally, sentences were presented word-by-word such that subjects had to press the spacebar to see the next word.

The results essentially showed that subjects were slower to read the disambiguating verb (e.g., “was”) when it had been preceded by a direct-object fast prime (e.g., “obtained”) than when it had been preceded by a sentence-complement fast prime (e.g., “realized”). In other words, subjects seemed to show more difficulty in processing the disambiguating word when they had been primed with a word whose typical usage was consistent with their original interpretation of the sentence.

Importantly, these effects were found only for the ambiguous sentences. If one suspected that the effect was due only to a greater semantic overlap between sentence complement primes and the disambiguating verbs, it should have occurred for both ambiguous and unambiguous sentences. Because this was not observed, it suggests that the fast primes activated more complex and abstract grammatical concepts than would be expected from simple semantic priming.