Word Order Typology

Joseph Greenberg, in "Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements" (in Universals of Language, ed. Greenberg, MIT Press 1963), examined a sample of thirty languages and proposed 45 universals — most implicational rather than absolute. Examples: "If a language has dominant order VSO, it has prepositions" (Universal 3); "Languages with dominant SOV order are postpositional with overwhelmingly greater than chance frequency" (Universal 4). The paper inaugurated a research tradition of typological generalizations and motivated theories (Greenberg's, later Hawkins's processing accounts) explaining the observed correlations.

Head-directionality is the parameter governing where the head of a phrase appears relative to its complement. Head-initial languages place heads first — VSO and SVO orders, prepositions (in the room), noun-genitive (book of John), determiner-noun. Head-final languages place heads last — SOV order, postpositions (Japanese kyoto-ni, "Kyoto-LOC"), genitive-noun, complement-final relative clauses. Mark Baker's The Atoms of Language (2001) and Mark Baker's earlier theoretical work formalized head-directionality as a global parameter. The correlation between verb-final and postpositional is one of the most robust typological findings.

Several explanations have been proposed. Russell Tomlin's Basic Word Order: Functional Principles (1986) suggested an evolutionary or animacy-driven preference for placing the subject (typically animate) before the object (often inanimate) and the verb. Tom Givón and others propose a processing account — animate-first, given-before-new pragmatic structure. The Givón-Tomlin frequency, the existence of "default SOV" in early human language reconstruction (Christopher Heine, 2008), and the strong tendency of pidgins to start as SOV before innovating SVO all point to SOV as a default. The reasons remain debated.

Latin, Russian, Czech, Sanskrit, and Warlpiri allow word order to vary considerably. The grammatical relations (subject, object) are signaled by case markings rather than position. Latin "puella puerum amat" (girl-NOM boy-ACC loves) and "puerum puella amat" both mean "the girl loves the boy" — the order shifts focus or topic. Free-word-order languages typically have rich morphological case systems. Word order in such languages encodes pragmatic structure (topic, focus, contrast) rather than grammatical relations. Lila Gleitman and Aravind Joshi formalized "configurational" vs. "non-configurational" languages.

Three frameworks compete. Generative parameters (Chomsky, Mark Baker, Hubert Haider) treat word order as a setting on parameters of UG, learnable from minimal exposure. Functionalist accounts (John Hawkins's Efficiency and Complexity in Grammars 2004) explain word-order patterns by parsing efficiency — head-initial vs. head-final orders minimize structural complexity differently. Construction grammar (Goldberg, Croft) treats word-order patterns as conventionalized constructions, not derived from deeper parameters. Each framework captures different aspects of the data; the parameters approach handles acquisition, the functionalist approach handles processing, and constructionist approaches handle semi-productive idiomaticity.

Yes. Latin (predominantly SOV) gave rise to Romance languages (predominantly SVO); the change occurred during the Vulgar Latin period (3rd-7th centuries CE). English shifted from SOV in Old English to SVO in Middle English under contact with French and as case morphology eroded. Indo-Aryan languages (Sanskrit, Hindi) retain SOV. Niger-Congo languages have varying orders. Mandarin Chinese has been SVO since at least Old Chinese, with some SOV-like patterns under Altaic contact in northern dialects. Word-order change is rarely abrupt; intermediate stages with mixed orders are typical (Edith Moravcsik 2013, Introducing Language Typology).

Object-subject-verb is the rarest basic order. Languages with documented OSV include Apurinã (Arawakan, Brazil), Warao (Venezuela), some Carib languages, and Yoda's English (humorously). The rarity reflects strong cross-linguistic preferences for subject-first order — animate, topical subjects tend to come early. The 1980s typological literature debated whether OSV was even a valid basic order or merely a marked variant. Languages with rich case marking can produce OSV in pragmatically marked contexts (as topicalization), but few use it as the basic, unmarked pattern. The frequencies are: SOV 45%, SVO 42%, VSO 9%, VOS 3%, OVS 1%, OSV under 1%.