Spoken typology: Integrating phonetics and grammar
Laura Becker
In this talk, I will present the approach of “spoken typology” and show how it is a useful addition to other typological approaches using (large) language samples. The idea of this approach is to use crosslinguistic spontaneous speech data for the analysis of a given linguistic phenomenon. Using such data can still allow for typological insights, as by now, collections of spontaneous speech reflect linguistic diversity rather well. At the same time, spontaneous speech data can be analyzed in a more fine-grained and meaningful way than aggregated data from large-scale typological samples.
The grammatical phenomenon that I will discuss to illustrate this approach is the effect of efficiency in verbal person markers. Efficiency is one of the functional factors that have been used to account for observed crosslinguistic patterns, i.e. different lengths of comparable expressions that differ in their frequencies in language use. By now, there is ample evidence for communicative efficiency affecting human language in the form of probabilistic variation in language use, e.g. more or less reduction depending on how frequent (or expected) an expression is (in a given context). However, it is still much less clear to what extent grammatical structures are systematically shaped by efficiency. Greenberg (1966) observed early on that for many grammatical categories, more frequent values (e.g. 3SG) have shorter markers than less frequent values (e.g. 2PL). How strong and systematic such effects really are is still an empirical question. Also, there are two main shortcomings of previous typological approaches examining efficiency effects in grammar: (i) the length of grammatical markers is often approximated by counting the number of some type of segments, which may ignore relevant variation in speech, (ii) the forms of grammatical markers are aggregated during sampling, which may ignore variation in the markers’ realization.
Using data from 8 languages from the DoReCo collection, I will show how we can test for efficiency effects in verbal person marking, how those issues can be addressed, and how integrating phonetic information and usage distributions can lead to new important perspectives in typology and grammar.