Speech Errors as a Window on Language and Thought: A Cognitive Science Perspective

We are so used to speaking in our native language that we take this ability for granted. We think that speaking is easy and thinking is hard. From the perspective of cognitive science, this view is wrong. Utterances are complex things, and generating them is an act of linguistic creativity, in the face of the computational complexity of the task. On occasion, utterance generation goes awry and the speaker’s output is different from the planned utterance, such as a speaker who says “Fancy getting your model renosed!” when “fancy getting your nose remodeled” was intended. With some notable exceptions (e.g. Fromkin 1971) linguists have not taken speech error data to be informative about speakers’ linguistic knowledge or mental grammars. The paper strives to put language production errors back onto the linguistic data map. If errors involve units such as phonemes, syllables, morphemes and phrases, which may be exchanged, moved around or stranded during spoken production, this shows that they are both representational and processing units. If similar units are converged upon via multiple methods (e.g. native speaker judgments, language corpora, speech error corpora, psycholinguistic experiments) those units have stronger empirical support. All other things being equal, theories of language that can account for both representation and processing are to be preferred.