440 likes | 539 Views
Prehistory of the Papuan realm: the ASJP evidence. Søren Wichmann Max Planck Institute for Evolutionary Anthropology. Introducing the Automated Similarity Judgment Program. Database consisting of 40-item wordlists for 57% of the world’s languages
E N D
Prehistory of the Papuan realm:the ASJP evidence SørenWichmann Max Planck Institute for Evolutionary Anthropology
Introducing the Automated Similarity Judgment Program • Database consisting of 40-item wordlists for 57% of the world’s languages • Words are compared through a string dissimilarity measure, a version of the so-called Levenshtein or ‘edit’ distance • Pairwise distances among languages are calculated as averages over the 40 lexical items • Distances can be used for all sorts of things, including the generation of phylogenies and the calculation of diversity measures for the purpose of inferring homelands
Basic ASJP resources drawn upon for this presentation • A classification of all 780 Papuan doculects in the database, representing 538 ISO-639-3 codes and 7 languages not recognized by Ethnologue • Identification of hypothetical homelands for 68 Papuan language families in the classification of Hammarström (2010)
What’s useful about ASJP trees? • Known to accurately replicate uncontroversial mid-level subgroups • Help to distinguish more believable from less believable phylogenetic clusters • More consistent than the impressionistic way in which lexicostatistics was often carried out in the 60’s and 70’s and which still influences the way that historical relations among Papuan languages are thought about • Can be produced in a manner of minutes / hours for several hundred / thousand languages
3 random examples of ASJP classifications(which are from different areas and which fit a page)
Contributions in this talk • Contributions to the classification of Papuan languages • Tearing down: using the ASJP tree to distinguish more reliable from less reliable proposals • Building up: using the ASJP tree to find relations that are new or not yet fully established • Contributions to other aspect of Papuan prehistory: • Viewing and interpreting the distribution of hypothetical homelands of 68 families in the HaHa classification
One random page from the ASJP classification of Papuan languages
Strategy for tearing down • Using a semi-conservative classification scheme such as that of Dryer (2005) find all cases where language families are uninterrupted and put them in the category ‘unproblematical’ • Apply ASJP to the remaining languages and repeat • Classify the remaining according to the highest-order standardly accepted groupings that correspond to uninterrupted segments of the ASJP tree
Initially problematical families • Numbers in parentheses indicate the number of segments into which the families • are segmented in the ASJP Papuan classification • Indents are the broken up WALS genera
After removing unproblematical families • Numbers in parentheses indicate the number of segments into which the families • are segmented in the ASJP Papuan classification • Indents are the broken up WALS genera
Slightly reduced list of problematical families • Numbers in parentheses indicate the number of segments into which the families • are segmented in the ASJP Papuan classification • Indents are the broken up WALS genera
Building up again • Isolate cases where a family is split because of one • or two intruders which might actually be related
Sko Molmo One Bunak
Isaka / Molmo One comparisons ASJPcode: * = nasalization E = ε
Tor-Orya & Kwerba Kwerba NB: Already a recognized Ethnologue family, so no further details here Tor-Orya
Parts (!) of the Wapei-Palei Subgroup of Torricelli
Kukwo / Aiku comparisons ASJPcode: ~ = preceding two symbols are a unit N = ŋ 3 = schwa
Slightly more reduced list of problematical families • Numbers in parentheses indicate the number of segments into which the families • are segmented in the ASJP Papuan classification • Indents are the broken up WALS genera
Building up further • Go through the ASJP tree and find possible, not standardly recognized relations • Check for geography • Inspect word lists • If promising do more in-depth studies
Saberi (“Kwerba”) & Kapauri (isolate)
Kapauri / Saberi comparisons ASJPcode: 7 = glottal stop
Savosavo and Bilua NB: Both recognized as Central Solomons in Ethnologue, so no further detail here
Murkim (isol.) & Lepki (isol.) & Kimki (“Sepik, Biksi”)
Murkim (isol.) & Lepki (isol.) & Kimki(“Sepik, Biksi”) comparisons
Murkim (isol.) & Lepki (isol.) & Kimki(“Sepik, Biksi”) comparisons (cont.)
Dagan (“TNG, Dagan”) & Kolopom (“TNG, Kolopom)
Kamula & Pare (both “TNG” but different groups)
Kamula & Pare comparisons ASJPcode = L lateral other than l
Second part: homelands • Method (Wichmann et al. 2010): • Find the language in a family that has the highest diversity index and assign the homeland of the family to this language • Calculate the diversity indices by taking the average of the ratio L/G between the target language and all other language (L = linguistic distance, G = geographical distance)
NB: 66 slides showing homelands of various small families taken out here to make the file smaller
Quantifying territoriality The smallest distance from each Papuan homeland to all other Papuan homelands: Mean: 87 km (82 km without outlier) Standard deviation: 69 km (52 km without outlier)
References • Dryer, Matthew. 2005. Genealogical language list. In: Haspelmath, Martin, Matthew Dryer, David Gil, and Bernard Comrie (eds.), 584-644. The World Atlas of Language Structures. Oxford: Oxford University Press. • Hammarström, Harald. 2010. A full-scale test of the language farming hypothesis. Diachronica 27: 197-213. • Wichmann, Søren, André Müller, and VivekaVelupillai. 2010. Homelands of the world’s language families: A quantitative approach. Diachronica27: 247-276.