What are metaphonemes?
Metaphonemes are phoneme correspondences that exist across languages. If we know, for example, that words which are realised with an /{/ in English are usually realised with an /A/ in Dutch, and an /a/ in German (as in cat /k{t/ versus /kAt/ versus /kats@/), we might be able to generalise over these three language-specific phonemes and introduce a metaphoneme, e.g. |{Aa|, which captures this generalisation. |
Dutch: /kAt/
|
Depending on the language, this metaphoneme will then be realised as an /{/ in English, an /A/ in Dutch, and an /a/ in German.
What can you do with metaphonemes?
The aim is to incorporate these metaphonemes in a multilingual inheritance-based lexicon, which allows sharing of information across languages at all levels of linguistic description as described in PolyLex (Cahill and Gazdar 1999). The basic idea behind the PolyLex architecture is that abstraction across languages can be represented within a standard default inheritance hierarchy in virtually the same way as abstraction within a language. Thus, information which is common to several languages will be stated at higher points in the hierarchy than information which is language-specific. The PolyLex project defined a trilingual hierarchical lexicon for Dutch, English, and German sharing morphological, phonological and morphophonological similarities between these three languages. Here, we take the PolyLex framework as our basis. We focus on the phonological similarities between related languages and we extend the PolyLex approach by capturing cross-linguistic phoneme correspondences, such as the /{/ - /A/ - /a/ correspondence mentioned above. The assumption is that the introduction of metaphonemes increases the amount of phonological generalisations that can be captured between languages at the syllable level in a multilingual inheritance lexicon. Calculations on the syllable inventories of Dutch, English, and German in the CELEX (Baayen et al. 1995) database show that the amount of sharing increases by about 25%.How do you find metaphonemes?
The vowel phoneme inventories of the Germanic languages were taken as a testbed. We compiled a list of 800 (mono- and disyllabic) Germanic cognates, looked up their transcriptions, and then mapped words containing a particular vowel in one language onto its cognates in the other languages, to see how a particular vowel was realised in the other languages. This process was repeated for all vowels, for all languages of the test set. In a first pilot study, we defined vowel phoneme correspondences for three West Germanic languages - Dutch, English, and German. The phonemic transcriptions were taken from the CELEX database (Baayen et al. 1995). The resulting tables can be found hereIn a second pilot study, we extended our test set with a related but less-closely related language, i.e. Danish, a North Germanic language. The transcriptions for Danish were taken from (Hansen 1990). The resulting tables can be found here.