Cagliari, Italy  15-18 September 2003

ALT V Conference

Bickel, Balthasar & Johanna Nichols

 Typological Enclaves
bickel@uni-leipzig.de

Large areal patterns of typological variables are sometimes punctuated by what we call enclaves: geographical zones whose typological profile deviates from the surrounding area,  but which are not themselves necessarily homogenous nor necessarily contact-induced. In this paper we provide evidence that the Himalayas and the Caucasus constitute typological enclaves in large-scale arealities that characterize Eurasia (comprising trends in both the Inner Eurasian and the Southeast Asian areas) and discuss possible historical scenarios that explain these distributions.

The methods used for establishing arealities and enclavehood are classical statistical tests run on genetically-balenced samples of 120-250 languages (depending on the variable sureveyed). However, since enclaves contain many fewer genetically independent datapoints than the surrounding area, individual sampling choices can have large effects. To assess the probability of such effects, we also submit the obtained test statistics to permutation tests that randomly re-assign data values and estimate how extreme the observed distribution is in the set of possible distributions (Manly 1991, Janssen et al 2003). Establishing enclavehood (or areality) requires statistical significance on both test series.

The most robust evidence obtained this way for Himalayan and Caucasian enclavehood comes (so far) from (i) the synthesis degree of verbal inflection (possible maximum of inflectional categories marked by a verb form), (ii) the presence of obligatory polypersonal agreement (i.e. transitive clauses obligatorily showing both subject and object agreement), (iii) the presence of conjunct/disjunct marking (marking of information relayer instead of current speaker), (iv) the presence of multiple possessive classes (more than two inflectional classes based on possessum types like body parts, kin terms etc.), (v) the presence of bipartite stems (inflectional affixes occur in the middle of stems but unlike infixes, their position is not defined phonologically), and (vi) the presence of obligatorily possessed nouns. Other variables will be added.

The key to explaining enclavehood lies in the fact that the Caucasus and the Himalayas are accretion zones (in the sense of Nichols 1997) and have by and large been shielded from the great spreads that led to the relatively uniform profile of Eurasia. Relative isolation gives rise to three types of enclave development scenarios:

1.  De-Skewing Enclave: typologies stabilize (over a long period) on the universal default, unhampered by areal skewing pressure. Example: multiple possessive classes. Evidence: the enclaves show the same frequencies as the universal average (30%); the rest of Eurasia (0%) deviates from the world-wide norm; and there is no significant large-scale (hemispheric, continent- or subcontinent-sized) areality outside Eurasia.

2.  Rarity Enclave: rare features develop undisturbed by areal homogeneization effects. Example: conjunct/disjunt marking. Evidence: found elsewhere only once; no known arealities.

3.  Preservation Enclave: the enclaves preserve what used to be the Eurasian standard profile before the great spreads changed that profile. Example: all other variables mentioned, and other diachronically stable variables. Evidence: the enclaves show the same (p > .1) statistical profile as the populations that orginally colonized the Americas and Australasia, as can be inferred from modern distributions on the Pacific Rim and the adjacent interior. The profiles are unlikely to be independent developments (as under the Rarity Enclave scenario) because they all show significant areality on many levels of geographical granularity.

References:

Janssen, Dirk, Fernando Zúñiga, and Balthasar Bickel. 2003. Randomization tests for areal typology. Routines for R and ms., University of Leipzig

Manly, Bryan (1991). Randomization and Monte Carlo methods in biology. London: Chapman and Hall

Nichols, Johanna, 1997. Modeling ancient population structures and population movement in linguistics and archeology. Annual Review of Anthropology 26, 359-84.