Induction of gender-like linguistic categories using noun-marking and blocking of learning trials

Abstract

Traditional views propose that language is acquired and processed by specialized mechanisms and that language processing relies on well-defined symbolic representations that are manipulated according to rules of language. In contrast, previous research has shown that linguistic categories, like those associated with gender, can be readily induced through phonological or morphological cues or by blocking related cases (Taraban, 2004). The present experiment tested whether noun-marking and blocked learning trials would aid participants in inducing gender-like categories in an artificial language consisting of twenty-four locative phrases. Sixty English-speaking college students at a university in the United States learned eight nouns in locative phrases in an artificial language (e.g., to car = gartaik eef). Nouns were divided into two gender-like classes. Gender-marked nouns (using -aik and -oo endings) and unmarked nouns were tested in two conditions. In one condition the phrases associated with the eight nouns were presented in random order (No Blocking). In the second condition, phrases associated with the same noun were presented in random sequence, and the learner had to input the correct locative postpositions associated with those phrases before proceeding to the next noun (Blocking). The results showed that unmarked nouns with blocking required less time to reach the experiment learning criterion (90% correct) than marked nouns. Blocking resulted in significantly higher accuracy on generalization trials to new phrases, but noun marking did not. The strong blocking advantage and null effect of noun marking are discussed in terms of cognitive attention to grammatical markers.