The citation-phonemic representation will not show such phenomena of running speech as place assimilation , consonant deletion , French liaison , RP linking-r and vowel reduction, etc. Hence there is a need for this more detailed level, which may (at least initially) be generated entirely by phonological rules from the citation-phonemic representation. It uses only symbols that have the status of phonemes , marking the output of connected speech processes that either insert or delete phonemes , or transform one phoneme into another.
This is the level referred to as ``phonotypic'' for French labelling by [Autesserre et al. (1989)]. It may be derived initially from the citation-phonemic labelling by phonological rules which, for example, reduce or delete unstressed vowels in English, or which delete some cases of word-final schwa in French. At this stage, it does not require reference to the sounds actually produced by the speaker (though this may come later, in manual post-editing). The ``broad phonetic'' level of [Barry & Fourcin (1992)] is intended as a phonemic-level representation of the speaker's tokens, i.e. the transcriber makes reference to the speech signal. The ``phonotypic'' level of [Autesserre et al. (1989)] is intended to be derived purely by rule. Both approaches are possible: however, there are certain advantages in deriving this level of representation purely by rule, as this is relatively quick and easy. This level uses a limited inventory of symbols (and so it is still a practical proposition for very large databases), while also offering more phonetic detail than the citation forms. So there is a balance between accuracy of representation and ease of derivation of the representation. However there will still be discrepancies between completely rule-derived labelling of an automatically segmented corpus) and the utterance, and hence some loss of accuracy; this may be offset by a large volume of data.