Level definitions define the hierarchical relationships between the levels of labels for utterances in the database. They define the names of the different levels of label which will be used and the domination relations between them. The basic syntax is:
Where levelname is an arbitrary name for the level and parent is the optional name of the level which dominates this level. All but one level in the hierarchy must have a parent. (Note: here and elsewhere in this document we will use the following notation for describing the syntax of template file statements. A bold word indicates that the word is typed as it appears in the definition, an emphasised word is a placeholder (for example for a level name above). Question marks around a word or set of words mean that they are optional and an asterisk means that the word or bracketed group of words may be repeated).level levelname ?parent? ?many-to-many?
An optional many-to-many tag at the end of the line denotes two levels where segments at the lower level can have more than one parent as well as segments at the higher level having more than one child. For example, the Phonetic level can be defined to have many-to-many links to the Phoneme level allowing a Phoneme /p/ to dominate a stop closure [p] and aspiration [H] at the Phonetic level, and to allow a phonetic segment such as a lengthened [s:] to map onto two successive phonemic /s/ segments that span a word boundary. Many-to-many links should normally only occur at the bottom level of a hierarchy since they could cause ambiguities in segment times if they occur anywhere else.
Example:
level Utterance level Intonational Utterance level Intermediate Intonational level Word Intermediate level Syllable Word level Phoneme Syllable level Phonetic Phoneme many-to-many |