The database may also be labelled or ``tagged'' at linguistic levels above the phoneme , such as the morphological, syntactic , semantic and pragmatic levels. This is of particular interest to those wishing to study, for example, the relationship between prosody and syntax . On the syntactic level , the researcher will label such items as clause and phrase boundaries, compound word internal boundaries, and possibly affix boundaries also. These units may be derived from the orthographic transcription , and so there is no need to examine the speech waveform for this level of representation. Automatic tagging software developped for large written text corpora may be used here, although adaptations are necessary for spontaneous speech.