SLIPS OF THE EAR
Errors of misperception by a listener, which provide insights into how the speech signal is processed and how words are recognised in connected speech.
The data can be analysed at phoneme level, with the proviso that it is difficult to determine to what extent top-down lexical effects (the knowledge of whole words that nearly fit the signal) may have led to a particular interpretation. With consonants, three types of error occur: deletions where no consonant is heard, additions where a consonant is inserted for which there are no cues in the signal and substitution where the reported consonant resembles the target one. Most consonant errors are word-initial. Since words are processed online, a mistake in this position is more likely to lead to a wrong match at word-level. Plosives are the most liable to be misinterpreted– supporting the findings of confusability studies. So far as vowels are concerned, those in stressed syllables are much less prone to misinterpretation than those in unstressed.
Slips of the Ear are useful in providing insights into how listeners determine where word boundaries lie in connected speech. When listeners misplace boundaries, they tend to insert them between a weak syllable and a strong– suggesting that segmentation is influenced by the predominant SW (strong-weak) pattern which characterises English rhythm. This finding from naturalistic Slips of the Ear is supported by similar evidence from slips induced by the faint speech method, which involves playing anomalous sentences at a level just above the subject’s hearing threshold. What both sources of data show is the vulnerability of weakly stressed function words, which may be misheard or attached to preceding strong ones.
Although there is a well-established corpus of Slips of the Ear, there are some problems in relying upon it as data. The slips are not usually audio-recorded, which means that the written record we have is dependent upon the observer’s analysis of the situation and limited in terms of contextual information. It is difficult to determine if any part was played by ambient noise, regional accent or by context (the absence of disambiguating information or the presence of misleading information). A major question is how representative the slips are, and how many others may have occurred without the listener heeding them. It is also possible that some apparent Slips of the Ear may in fact have originated in Slips of the Tongue. Finally, it is not always easy to determine with certainty the cause of a slip. These difficulties are admitted by researchers, but the corpus is large enough for some general conclusions to be reached.
See also: Lexical segmentation, Speech perception, SW (strong-weak) pattern
Further reading: Bond (1999)
