Wordhood and phonological boundness or independence - grambank/grambank GitHub Wiki
Many features in Grambank make a distinction between morphological marking (affixation, cliticization etc.) and marking by means of independent words (particles, auxiliaries, etc.).
The difficulty to define wordhood is well-known in linguistic typology (Haspelmath 2011). For the purpose of Grambank, we have decided to work with a phonological definition: 'bound' marking or 'morphology' means 'phonologically bound'; 'non-bound marking' (e.g. with inflecting or non-inflecting words) involves markers that are phonologically free.
A morpheme is phonologically free if it occurs as a phonological word. It is phonologically bound if it does not occur as its own phonological word. Whether a morpheme is bound or not can be established by looking at one or more of the following criteria.
-
stress: if a morpheme's stress depends on other morphemes it occurs with or if it falls within another morpheme's stress domain and cannot be stressed by itself, it is phonologically bound.
-
boundary phenomena or (phonological) allomorphy: if a morpheme's segmental form depends on other morphemes it occurs with, it is phonologically bound.
-
vowel harmony: if a morpheme is affected by or causes vowel harmony, it is phonologically bound.
Conversely, if a morpheme:
- has its own stress domain,
- is phonologically invariable, and
- does not show vowel harmony (in languages that have vowel harmony),
it should be treated as phonologically free.
In practice, there is rarely sufficient information in a grammar to judge the phonological boundness of a given morpheme according to these criteria. In those cases, we rely on the orthography an author uses. If a form is orthographically bound, it is assumed to be phonologically bound. If it stands on its own orthographically, it counts as a separate word. A small exploratory study of 134 morphemes across 23 random languages in Grambank showed that 90% of the orthographically bound morphemes were also phonological bound in the sense laid out in this wiki article (Lesage & Neshcheret 2017).
Wordhood is also relevant to the distinction between pronouns and indexes in Grambank: phonologically free person forms are considered as pronouns; phonologically bound person forms are considered as indexes.
References and further reading
Dixon, R. M. W. and Aikhenvald, Alexandra Y. eds. 2002. Word: a cross-linguistic typology. Cambridge University Press.
Haspelmath, Martin. 2011. The indeterminacy of word segmentation and the nature of morphology and syntax. Folia Linguistica, 45(1), 31–80.
Lesage, Jakob & Neshcheret, Nataliia. 2017. Wordhood issues while compiling a large-scale typological database. Zürich. (Talk presented at the workshop "What is a word?", Zürich.)