Three kinds of generalization
An absolute universal says that a property holds in every language under a stated definition. A statistical tendency says that it is more frequent than alternatives in a specified sample. A conditional generalization says that one feature is associated with another: if a language has A, B is more likely, obligatory, or otherwise constrained. These are different claims. A single secure counterexample can refute an absolute universal, while it need not erase a probabilistic tendency.
The definition must be narrow enough to test. “Languages put important words first” has no stable measurement; “in transitive declaratives with two lexical noun phrases, the object normally precedes the verb” does. One must also decide what “normally” means and how ties or constructional alternations are coded. WALS explicitly has a value for languages lacking a dominant order.
A robust association and a weak one
WALS chapter 95 compares object–verb order with adpositions. In its sample, OV languages commonly use postpositions and VO languages commonly use prepositions, though the cross-combinations are attested. The relationship is therefore a strong tendency, not an unbreakable law. Its counts apply to the author’s coding and sample, not to all stages of all languages.
WALS chapter 97 makes a useful contrast: the order of adjective and noun does not correlate with object–verb order nearly as strongly. A regional cluster can tempt a reader to infer a worldwide rule that disappears in a broader sample. Comparing both chapters shows why typology is empirical rather than a set of neat charts to memorize.
Explanation after description
A recurring pattern may invite explanations in processing, acquisition, communicative use, historical inheritance or areal diffusion. The frequency pattern itself does not choose among them. Ask whether the proposed explanation predicts independent facts, whether related languages were counted as separate observations, and whether exceptions are expected. If a claim concerns ancient change, assemble dated stages rather than drawing a historical arrow from a modern distribution map.
Exercise: write the OV–postposition relationship twice, once incorrectly as “all OV languages have postpositions” and once as a testable tendency. Then find a documented cross-combination in the WALS chapter and explain why it matters differently to each formulation.
Modern-language examples on this page clarify a linguistic pattern. They do not establish unattested sounds or forms for an ancient language. Follow the cited sources below and keep editorial readings separate from the surviving trace.