= Linguistic typology
<Linguistic typology> compares structural properties across <languages> using explicit criteria for grouping constructions. It can ask which patterns recur, which combinations are rare, and whether a proposed generalization survives a wider sample. The <World Atlas of Language Structures> exemplifies this approach by coding documented properties as comparable features.<WALS Online by Matthew S. Dryer and Martin Haspelmath>[project description, structural properties and descriptive source materials]{ref}
Consider <word order>. Three distinct elements have $3!=6$ possible <permutations>, but logical possibility alone says nothing about their distribution in <languages>. A published comparison must define which constructions count: the atlas's chapter on <subject (grammar)>, <object (grammar)> and <verb> order restricts its main comparison to specified transitive declarative constructions.<Order of Subject, Object and Verb by Matthew S. Dryer>[§1, scope of comparison and six dominant-order types]{ref}
For an original counting example, a sample containing six languages of one type and four of another gives proportions $0.6$ and $0.4$. Those proportions describe that sample. <Sampling bias>, missing observations and shared historical relationships can prevent them from estimating how common the types are more generally. An unrecorded feature must also remain distinct from an observed absence.
Back to article page
