#speech error

Speech errors – we’ve all made them, we’ve all heard them. Slips of the tongue we call them. But it seems our tongues slip in particular ways and these can give us a valuable insight into linguistic structures.

The majority of speech errors can be classified by type, for example, additions and deletions, exchanges, blends, substitutions, mis-selections, perseverations, anticipations etc. and these can affect phonemes, bound morphemes and free morphemes. Interestingly there are some errors which hardly ever seem to occur, for example, scrambling of the phonemes within a word, exchange errors between different parts of a syllable etc.

The fact that speech errors can target phonemes supports (among other evidence) the idea that phonemes are psychologically real. Also the fact that phonemic exchange errors affect corresponding parts of syllables suggests that phonemes are structured into syllables during speech production, thus supporting the idea that syllables are also psychologically real.

When speech errors affect bound morphemes (i.e. morphemes which must be attached to other morphemes, e.g. the past tense <-ed> in English), it is possible to infer some of the ordering of the processes involved in speech. For example, if the past tense morpheme in English is, for whatever reason, attached to the wrong word, it will be realised as /d/, /t/ or /ɪd/ depending on the ending of the erroneous word, regardless of what it would have been if attached to the target word. This suggests that the error was already made before the phonological/phonetic rules applied. This in turn suggests that there is a sort of ‘mapping’ stage where abstract morphemes are mapped onto syntactic structures before the structure reaches the phonological part of producing a sentence.

Speech errors can also affect free morphemes (i.e. morphemes which can exist independently). These errors can give us an insight into how these morphemes (or ‘words’) are stored in our mental lexicons. Humans are able to learn thousands of words but it is very hard to see directly (though new technology might change this) how these thousands of words are stored. It might be that words are stored in a sort of phonological dictionary. This would be a very efficient filing system but, on its own, it does not account for many of the exchange errors we find. Have you ever said a word as part of a sentence but it was not the word you meant to say? Odds are that although the word was ‘wrong’ so-to-speak, it was at least the right category of word (e.g. verb or noun etc.). Also, have you ever said one thing but meant to say something different but obviously related? Like saying ‘cow’ when you meant ‘sheep’, or even ‘left’ when you meant ‘right’? I certainly have, and many others have too. This suggests that words are also stored by syntactic category, semantic similarities, pragmatic similarities, by their various relationships with other words etc. etc. like a vast dictionary with multiple links between entries criss-crossing the lexicon. It is not random, but highly and intricately structured. If it wasn’t, if I meant to say ‘right’, I would be just as likely to say ‘between’ or ‘thirty’ or ‘explode’ as I would be to say ‘left’ and yet only this last possibility occurs with any significant frequency in this case.