Relative clause formation encompasses diverse grammatical strategies: wh-movement and relative pronouns, resumptive pronouns, headless/free relatives, and correlative constructions. The choice of strategy interacts with the Accessibility Hierarchy, island constraints, and language-specific morphosyntax.
You have studied relative clauses as a grammatical category — structures that modify nouns by embedding a dependent clause — and you've worked with the syntactic machinery of wh-movement. Now the question shifts from *what* relative clauses are to *how* different languages build them, and what that variation reveals about the underlying grammar. Languages do not all use the same strategy to form relative clauses, and understanding the range of strategies illuminates both typological universals and language-specific constraints.
The strategy you know best is wh-movement: a relative pronoun (who, which, that) moves from its underlying position in the embedded clause to the front of the relative clause, leaving a gap — an empty position where the head noun's role is interpreted. "The student [who_i I recommended t_i]" — the relative pronoun *who* has moved from the object position (the gap *t_i*) to the clause-initial position. English and most European languages use this gap strategy. The key property is that the extraction site is phonologically silent; the moved element encodes the grammatical relation. Island constraints, which you've seen in the context of wh-movement generally, restrict which positions can be relativized this way — you cannot extract out of a subject island or a complex noun phrase island without producing ungrammaticality.
When wh-movement is blocked or disallowed, some languages use a resumptive pronoun strategy instead: rather than leaving a silent gap, the extraction site contains an overt pronoun that is coreferential with the head noun. Hebrew, Arabic, and many West African and Bantu languages employ resumptives systematically. "The man [that I saw *him*]" — the pronoun *him* is the resumptive, filling the position where a gap would appear in English. Resumptives are not errors or awkward repairs; they are the grammatically regular strategy in these languages. Interestingly, even in English, resumptive pronouns appear in long-distance contexts where speakers feel gap-formation is too demanding — suggesting that resumptives may be a universal option that is simply suppressed in English under normal conditions.
The Accessibility Hierarchy (Keenan & Comrie, 1977) is the key typological universal in this domain. It ranks grammatical positions by how easily they can be relativized: subject > direct object > indirect object > oblique > genitive > object of comparison. The universal is that if a language can relativize a lower position on the hierarchy, it can also relativize every higher position — but not vice versa. Some languages relativize only subjects; some add direct objects; English and many others go all the way down. This hierarchy makes precise, cross-linguistically testable predictions. Correlative relative clauses — found in Sanskrit, Hindi, and Turkish — represent yet another strategy: rather than embedding the clause inside the noun phrase, the relative clause appears as a separate left-peripheral clause and is linked to the main clause through a correlative pronoun. These are not subordinate in the traditional sense but are biclausal structures where a demonstrative in the main clause picks up the reference established in the correlative. Together, these strategies show that the semantic function of relativization is universal, but the syntactic machinery for achieving it varies along principled, constrained dimensions.