Does verb order predict prepositions and postpositions?

By Chase Morgan · Published 27 October 2026 · Data: WALS Online 2013 edition (CLDF release v2020.4)

Yes, and very strongly. Among the languages WALS codes for both features, 472 of the 516 that put the object before the verb use postpositions (91%), and 454 of the 543 that put the object after the verb use prepositions (84%). Japanese and Turkish show the first pattern, English and Swahili the second.

The link between the order of object and verb and the type of adposition is the clearest of the word-order correlations in the atlas. It is not perfect: 14 object-before-verb languages have prepositions and 41 object-after-verb languages have postpositions. But exceptions run to a few per cent, which is rare for a pattern that crosses families and continents.

Why the two orders travel together

In both pairings, one word governs another. The verb takes an object; the adposition takes a noun phrase. A language that puts the object before the verb tends to put the noun phrase before the adposition as well (“the house in”), so the governing word comes last in both. A language with the verb before the object tends to say “in the house”, with the governing word first. The WALS chapter on this relationship, by Matthew S. Dryer, notes that it is only one of many correlations with the order of object and verb, discussed in a large literature that includes Greenberg (1963), Hawkins (1983, 1994) and Dryer (1992). One of the hypotheses it lists is a consistent ordering of grammatical heads with respect to their dependents; another links the correlations to consistently left-branching or right-branching structures.

Which other pairings follow the verb

Not every part of the noun phrase follows the same rule. The table compares languages with the object before the verb and languages with the object after it, and asks how often each puts a dependent before its noun.

Genitives follow the verb almost as faithfully as adpositions do: 90% of object-before-verb languages put the possessor first (“the man’s house”), while 69% of object-after-verb languages put it after (“the house of the man”). Relative clauses show a lopsided pattern. Among object-after-verb languages, 415 of 433 place the relative clause after the noun and only 5 place it before. Object-before-verb languages are split: 132 put it before the noun, as Japanese and Turkish do, but 113 put it after. A relative clause before the noun is close to a one-way signal: if a language has it, the object almost certainly precedes the verb.

Adjectives and numerals barely care

The order of adjective and noun is a different story. Adjectives come before the noun in 36% of object-before-verb languages and in 19% of object-after-verb languages, a gap of 17 points. Both groups mostly place the adjective after the noun. English, with the adjective before the noun and the object after the verb, is the less common combination among its type. Numerals are weaker still: 40% against 42%. Word order in the noun phrase, it turns out, has its own regional habits that cut across verb order. Our note on adjective order looks at those habits area by area.

The exceptions with a page here

The exceptions are worth a look because they show that the pattern is a tendency, not a rule. Among the languages profiled on this site, 7 put the object before the verb but use prepositions: Iraqw, Kuku-Yalanji, Kurdish (Central), Mangarrayi, Persian, Tigré and Tuvaluan. Another 12 put the object after the verb and use postpositions, among them Finnish, Estonian and Hungarian, whose postpositions sit alongside rich case systems, and Ewe and Dagbani in West Africa. Each of these pages shows the mismatch in the order diagram at the top: the adposition row leans the opposite way from the verb row.

How to see it in one language

Each language page on this site draws these pairings as a column of short rows, with the governing word in the centre and its dependent placed on the side the language uses. Head-final languages such as Korean lean left almost row by row; head-initial ones such as Irish lean right. Mixed languages show where the tendencies give way. The thin bar under each row shows how all languages coded for that pairing divide, so a row that goes against the grain is easy to spot.

The figures here are counts of the languages WALS codes for both features in each pair, so the totals change from row to row. They describe the atlas’s sample, which covers families broadly but is not a full list of the world’s languages. Large families with a fixed order can weigh on a count; the family pages show how uniform each family is on these features.