When Intuitions Fail

The title of this article looks in two different directions. It asks, "When do intuitions fail?" hoping to locate the conditions under which intuitions are likely to beunreliable.But it can also be the subordinate clause, of the question, "When intuitions fail, what can we do?": Both approaches, the remedial and the prophylactic, will be considered here. In 1951, Carl Voegelin and Zellig Harris outlined the basic opposition between two methods of gathering data. They characterized the method of Boas and Sapir as simply "Ask the informant," and pointed out the many biases introduced into the data by this method. They acknowledged that the second method, to record the speech of the informant directly, also involved many biases, particularly in the artifacts introduced by having the speaker wait while the investigator transcribed the data. They were hopeful that the recent invention of the magnetic tape recorder would resolve these difficulties, and allow linguistics to move forward on a more objective and scientific basis. My own bias is on the side of Voegelin and Harris, towards the study of speech behavior. However, it must be acknowledged that linguistic analysis will always rely to a large extent on elicited judgments, the intuitions of the native speakers. Though studies of the speech community have focused primarily on spontaneous speech, they have not neglected the subjective dimension. Investigations of regional grammatical forms and crossethnic differences have searched for answers to six types of questions to assess the linguistic knowledge of a given rule or construction (adapted from Labov 1972) X = the form or construction in question WXY = a sentence employing X. (1) Recognition of the form as a proper part of the language in question. ("Is WXY English?") (2) The ability to identify the regional, social or stylistic level ("What kind of English is WXY? Who would say that?") (3) The ability to interpret the meaning of the form in a sentential context. ("What does the sentence WXY mean?") (4) The ability to interpret the meaning of the form out of context. ("What does X mean in WXY?") (5) The ability to predict the acceptability of X in other contexts. ("If you can say WXY, can you say WXZ? can you say VXY?") (6) The use of X in spontaneous speech with native frequency and native pattern of categorical and variable constraints.