The evolution of word-order universals: Some word-order correlation are lineage specific - others might be universal Gerhard Jäger, Gwendolyn Berger, Isabella Boga, Thora Daneyko & Luana Vaduva Tübingen University Association for Linguistic Typology, Canberra December 15, 2017 Jäger et al. (Tübingen) Word-order Universals ALT2017 1 / 26 Introduction Introduction Jäger et al. (Tübingen) Word-order Universals ALT2017 2 / 26 Introduction Word order correlations Greenberg, Keenan, Lehmann etc.: general tendency for languages to be either consistently head-initial or consistently head-final alternative account (Dryer, Hawkins): phrases are consistently left- or consistently right-branching can be formalized as collection of implicative universals, such as With overwhelmingly greater than chance frequency, languages with normal SOV order are postpositional. (Greenberg’s Universal 4) both generativist and functional/historical explanations in the literature Jäger et al. (Tübingen) Word-order Universals ALT2017 3 / 26 Introduction Phylogenetic non-independence languages are phylogenetically structured if two closely related languages display the same pattern, these are not two independent data points ) we need to control for phylogenetic dependencies (from Dunn et al., 2011) Jäger et al. (Tübingen) Word-order Universals ALT2017 4 / 26 Introduction Phylogenetic non-independence Maslova (2000): “If the A-distribution for a given typology cannot be as- sumed to be stationary, a distributional universal cannot be discovered on the basis of purely synchronic statistical data.” “In this case, the only way to discover a distributional universal is to estimate transition probabilities and as it were to ‘predict’ the stationary distribution on the basis of the equations in (1).” Jäger et al. (Tübingen) Word-order Universals ALT2017 5 / 26 The phylogenetic comparative method The phylogenetic comparative method Jäger et al. (Tübingen) Word-order Universals ALT2017 6 / 26 The phylogenetic comparative method Modeling language change Markov process Jäger et al. (Tübingen) Word-order Universals ALT2017 7 / 26 The phylogenetic comparative method Modeling language change Markov process Phylogeny Jäger et al. (Tübingen) Word-order Universals ALT2017 7 / 26 The phylogenetic comparative method Modeling language change Markov process Phylogeny Branching process Jäger et al. (Tübingen) Word-order Universals ALT2017 7 / 26 ... transition rates, stationary probabilities and ancestral states can be estimated based on Markov model The phylogenetic comparative method Estimating rates of change if phylogeny and states of extant languages are known... Jäger et al. (Tübingen) Word-order Universals ALT2017 8 / 26 The phylogenetic comparative method Estimating rates of change if phylogeny and states of extant languages are known... ... transition rates, stationary probabilities and ancestral states can be estimated based on Markov model Jäger et al. (Tübingen) Word-order Universals ALT2017 8 / 26 The phylogenetic comparative method Correlation between features Pagel and Meade (2006) construct two types of Markov processes: independent: the two features evolve according to independend Markov processes dependent: rates of change in one feature depends on state of the other feature fit both models to the data apply statistical model comparison Independent model Dependent model VO PN VO/PN OV/PN VO/NP OV NP OV/NP Jäger et al. (Tübingen) Word-order Universals ALT2017 9 / 26 Dunn et al. (2011) Dunn et al. (2011) Jäger et al. (Tübingen) Word-order Universals ALT2017 10 / 26 Dunn et al. (2011) Dunn et al. (2011) all 28 pairs of 8 word-order features considered 4 language families: Austronesian, Bantu, Indo-European, and Uto-Aztecan main finding: wildly different results between families conclusion: word-order correlations are lineage-specific Jäger et al. (Tübingen) Word-order Universals ALT2017 11 / 26 Universal and lineage-specific models Universal and lineage-specific models Jäger et al. (Tübingen) Word-order Universals ALT2017 12 / 26 Universal and lineage-specific models This study Experiments 1 replication of Dunn et al. (2011) with different data 2 model comparison: universal vs. lineage-specific correlations 3 word-order correlations across a world-tree of languages 4 automatically identifying lineage-specificity Jäger et al. (Tübingen) Word-order Universals ALT2017 13 / 26 Universal and lineage-specific models Data word-order data: WALS phylogeny: ASJP word lists (Wichmann et al., 2016) feature extraction (automatic cognate detection, inter alia) ; character matrix Maximum-Likelihood phylogenetic inference with Glottolog (Hammarström et al., 2016) tree as backbone advantages over hand-coded Swadesh lists applicable across language familes covers more languages than those for which expert cognate judgments are available 1004 languages in total Austronesian: 123; Bantu: 41; Indo-European: 53; Uto-Aztecan: 13 Jäger et al. (Tübingen) Word-order Universals ALT2017 14 / 26 Universal and lineage-specific models Replication of Dunn et al. Jäger et al. (Tübingen) Word-order Universals ALT2017 15 / 26 Universal and lineage-specific models Comparing universal and lineage-specific models so far: fitting a separate model for each language family advantage: good fit of the lineage-specific data disadvantage: many parameters (8 per family for a dependent model) statistical model comparison: quantifying to what degree the data support the excess parameters of lineage-specific models models to be compared: universal: one set of rates (8 parameters), applying to all 4 families lineage specific: a separate set of rates for each family comparison via Bayes Factor (implementation with RevBayes; Höhna et al. 2016) Jäger et al. (Tübingen) Word-order Universals ALT2017 16 / 26 Universal and lineage-specific models Results very strong evidence for universality: noun-adjective $ noun-numeral feature pair Bayes Factor NA-NNum 16.24 adposition-noun $ verb-object PN-VO 15.22 PN-NG 9.45 VO-NRc 9.21 strong evidence for universality: PN-NRc 8.69 NRc-VS 8.18 universal NG-VO 7.92 adposition-noun $ verb-object $ noun-genitive $ noun-relative NG-NRc 6.55 NA-NRc 6.49 clause PN-ND 5.42 ND-NRc 4.32 VO-VS 3.15 strong or very strong evidence for lineage specificity: PN-VS 1.71 NA-ND 0.54 behavior of noun-adjective and noun-numeral ND-VO 0.37 NA-VO -2.07 ND-NG -3.17 NNUM NG NA-PN -3.40 NNum-VS -8.13 NNum-NRc -8.40 c NA-VS -9.66 fi VS PN NG-VS -9.84 NA-NG -10.94 ND-NNum -12.12 ND-VS -15.01 PN-NNum -16.37 NNum-VO -17.57 NA VO NG-NNum -28.63 lineage-speci ND NRc Jäger et al. (Tübingen) Word-order Universals ALT2017 17 / 26 Universal and lineage-specific models Results universal (PN/VO) lineage-specific (NG/NNum) Austronesian PN NG NNum 1.18 VO 9.15 4.91 1.6 4.06 0.51 NG GN NumN 11.35 NNum 1.24 0.22 0.92 9.1 GN 3.98 NumN PN NP Bantu OV 1.07 VO NG NNum 4.85 8.87 0.38 0.37 13.22 4.8 NG GN NumN 4.08 NNum 3.92 2.74 3.86 4.2 GN NP NumN OV Indo-European NG NNum 4.09 4.86 3.87 0.56 NG GN NumN NNum 2.5 4.5 3.23 0.7 GN NumN Uto-Aztecan NG NNum 3.46 4.41 4.61 2.63 NG GN NumN NNum 2.02 5.8 3.76 2.14 GN NumN Jäger et al. (Tübingen) Word-order Universals ALT2017 18 / 26 Universal and lineage-specific models Using the world tree Glottolog family Macro-Area Atlantic-Congo.D27 BANGUBANGU MUTINGWA Atlantic-Congo.KOUNDOUMOU BAMBOMBA BOMITABA Atlantic-Congo.BOMA NORD PLATEAUX CONGO Atlantic-Congo Atlantic-Congo.BANGUBANGU KABAMBARE Africa Atlantic-Congo.HEMBA BWINYANYEMBA Atlantic-Congo.KONGO SAN SALVADOR 2 Atlantic-Congo.H166 KONGO MANYANGA Atlantic-Congo.B78 IWUM WUUMBU SUD TEKE Atlantic-Congo.K15 MBUNDA GANGELA Atlantic-Congo.KONGO SAN SALVADOR Atlantic-Congo.IBOLO BAMBOMBA BOMITABA Atlantic-Congo.SONGOLA KASENGA Atlantic-Congo.B44 LUMBU YI BANDA Atlantic-Congo.LEGA SHABUNDA 2 Atlantic-Congo.SONDE GISOONDI Atlantic-Congo.SONGOLA ULINDI Atlantic-Congo.LEGA SHABUNDA Atlantic-Congo.L31a LUBA KASAYI Atlantic-Congo.HEMBA MAMBWE Atlantic-Congo.NGWI MATEKO KINGOLI Atlantic-Congo.L21 KETE KATAMB Atlantic-Congo.H12 VILI MAYUMBA Atlantic-Congo.SONDE KYAANZA Atlantic-Congo.TOUKOULAKA BOMITABA Atlantic-Congo.BOMA NORD EKEMWA Atlantic-Congo.B63 NDUUMO KUYA Atlantic-Congo.OYUOMI TCHERRE Atlantic-Congo.D28 HOLOHOLO Atlantic-Congo.L51 SALAMPASU Atlantic-Congo.ENYA KIBOMBO Atlantic-Congo.ENYA Atlantic-Congo.LEGA MWENGA Atlantic-Congo.B87 MBUUN LABAEMPI Atlantic-Congo.B70 TEKE OMVULA Atlantic-Congo.LUMBU YI TANDU Atlantic-Congo.H42Atlantic-Congo.H42Atlantic-Congo.H42Atlantic-Congo.H42 HUNGANA HUNGANA HUNGANA HUNGANAAtlantic-Congo.KONGO 3 2 4 1 MBOMA Atlantic-Congo.MASHI ZAMBIA Atlantic-Congo.BANGUBANGU Atlantic-Congo.L21 KETE IPILA Atlantic-Congo.RUUMBU KIMWAANSA Atlantic-Congo.MBAMBA LIWEME Atlantic-Congo.K15Atlantic-Congo.K15 MBUNDA MBUNDA 1 2 Atlantic-Congo.L221Atlantic-Congo.L221 LWALWA LWALWA 2 1 Atlantic-Congo.OYUOMI MBAMA Atlantic-Congo.YAKA PELENDE Atlantic-Congo.MOKENGUI BOMITABA Atlantic-Congo.L32Atlantic-Congo.L32 KANYOK KANYOK 1 2 MANDA Atlantic-Congo.ENYA Atlantic-Congo.BUJA BUMBA YAMOLOTO Atlantic-Congo.KWANYAMA 2 Atlantic-Congo.LIOUESSO BOMITABA Atlantic-Congo.SONDE FESHI Atlantic-Congo.BONDEKO BOMITABA Atlantic-Congo.TEE TEKE KALE Atlantic-Congo.L22 MBAGANI Atlantic-Congo.K14Atlantic-Congo.K14 LWENA LWENA 2 1 Atlantic-Congo.NKOMO KELLE Atlantic-Congo.KWANGALI 2 Atlantic-Congo.B414 VARAMA Atlantic-Congo.H41 MBALA 1 Atlantic-Congo.MULONGA 1 Atlantic-Congo.MULONGA 2 Atlantic-Congo.MBAMBA SIBITI Atlantic-Congo.UMBUNDU 2 Atlantic-Congo.BOMA NORD SAIO Atlantic-Congo.K11 CIOKWE Atlantic-Congo.MBUKUSHU Atlantic-Congo.HOLOHOLO Atlantic-Congo.BUJA MONOGO