Wh-Movement and Extraction | Catalan Grammar

In Quin informe ha aprovat el consell?, quin informe is pronounced at the left edge but interpreted as the object of aprovat. The relation between that fronted wh-phrase and its unpronounced argument position is an extraction dependency. Catalan permits such dependencies across some clause boundaries, but not out of every constituent.

The advanced problem is not “put the question word first.” Speakers track argument structure, prepositions, clause type, information structure, and domains known as islands (academic). English allows preposition stranding and obliges auxiliary inversion; Catalan generally pied-pipes the preposition and uses its own subject-placement patterns, without inserting do. Review the ordinary inventory before this page under wh-questions.

The dependency has a pronounced edge and a gap

A neutral object question can be represented as follows:

[Quin informe]ᵢ ha aprovat el consell [ᵢ]?

Quina versió ha enviat finalment l'editora al client?

Which version did the editor finally send to the client?

Qui ha revisat els càlculs abans de la reunió?

Who reviewed the calculations before the meeting?

In the first, the wh-phrase is the object and l'editora is a postverbal subject. In the second, qui is itself the subject, so singular ha revisat agrees with it. An initial wh-phrase is therefore not automatically a subject. The verb's valency and the remaining arguments identify its role.

Wh-movement is compatible with a gap across an ordinary finite complement, especially after a bridge verb such as dir, creure, or pensar:

Quin pressupost dius que aprovarà el patronat la setmana vinent?

Which budget do you say the board will approve next week?

[Quin pressupost]ᵢ dius [que el patronat aprovarà ᵢ la setmana vinent]?

The dependency crosses que, but the object still belongs semantically to aprovar. Such long extraction is available in all registers, though deeply nested versions become more characteristic of planned (formal/academic) prose than conversation.

Prepositions move with their complement

Standard Catalan normally requires preposition pied-piping: the governing preposition accompanies the wh-expression to the left edge.

Amb quina assessora vas parlar del contracte?

Which adviser did you speak with about the contract?

De què depèn exactament la data de lliurament?

What exactly does the delivery date depend on?

Per a qui heu reservat aquesta plaça?

Who have you reserved this place for?

English commonly strands with, on, or for at the end. Catalan does not reproduce that neutral pattern: Qui vas parlar amb? is outside careful contemporary standard Catalan. In spontaneous contact-influenced speech one may hear stranded or resumptive-looking solutions, but they should not be presented as the shared Central/Valencian written norm. Choose the preposition selected by the Catalan verb, not the one visible in the English translation: dependre de, parlar amb, confiar en.

💡

Before moving a wh-phrase, reconstruct the ordinary statement: Vas parlar amb l'assessora → Amb quina assessora vas parlar? If the statement needs a preposition, carry that preposition with the question phrase.

Topics with clitics are not extracted wh-arguments

Catalan left dislocation can place a familiar topic before a complete clause and resume it with a weak pronoun:

D'aquest ajornament, què en pensa la direcció?

As for this postponement, what does management think about it?

Here d'aquest ajornament is a detached topic, en resumes its de-relation, and què asks for the content of the opinion. There are two dependencies, not one wh-phrase “moved twice.” By contrast, in De quin ajornament parla la direcció?, de quin ajornament is the extracted complement and there is no resumptive en in standard usage.

Likewise, an extracted direct object normally leaves a gap rather than a clitic:

Quina proposta defensarà demà la portaveu?

Which proposal will the spokesperson defend tomorrow?

Adding la — Quina proposta la defensarà...? — is not the neutral interrogative structure. It can only be salvaged under a special dislocated or echo interpretation with appropriate prosody, and careful writing should make that structure explicit.

Islands: constituents that resist extraction

An island is a domain from which an ordinary wh-dependency cannot escape. The asterisk below marks the intended reading as unacceptable, not every conceivable sequence of those words.

Complex noun phrases

Relative clauses and complements buried inside a noun phrase form strong islands:

❌ De quin autor has conegut la traductora que acaba de traduir una novel·la?

Unacceptable on the intended reading: Which author did you meet the translator who has just translated a novel by?

The gap would sit inside [la traductora [que...]], itself the object of conegut. Repair the sentence by asking within the relative proposition or by splitting the information:

✅ Has conegut la traductora? De quin autor és la novel·la que acaba de traduir?

Have you met the translator? Who is the author of the novel she has just translated?

Adjunct clauses

Temporal, causal, conditional, and concessive clauses are not selected object complements; extraction from them is sharply degraded:

❌ Quin expedient vas marxar abans que l'arxivessin?

Unacceptable on the intended reading: Which file did you leave before they filed?

✅ De quin expedient parlem, el que van arxivar abans que marxessis?

Which file are we talking about, the one they filed before you left?

The repair gives the wh-phrase a local relation in the main question and places a complete relative clause after it.

Subject islands

Extraction from inside a complex subject is also strongly restricted:

❌ De quin edifici dius que el terrat s'ha ensorrat?

Unacceptable on the intended extraction reading: Which building's roof do you say has collapsed?

✅ De quin edifici dius que s'ha ensorrat el terrat?

Which building's roof do you say has collapsed? (improved with the subject in its lower postverbal position)

The postverbal version is substantially better for many speakers. This asymmetry has an important analysis: a preverbal subject behaves like an island, whereas a postverbal subject can remain lower in the clause and be more transparent. Judgments vary with definiteness, predicate type, prosody, and processing load, so edited prose often chooses the unambiguously local De quin edifici s'ha ensorrat el terrat, segons dius? or recasts the relation as Quin edifici dius que té el terrat ensorrat? The variation is evidence about structure, not permission to ignore clarity.

Questions inside questions

An embedded interrogative is a wh-island. Extracting another wh-phrase across its edge is degraded:

❌ Quin article et preguntes qui va revisar?

Unacceptable on the intended reading: Which article do you wonder who reviewed?

✅ Et preguntes qui va revisar quin article?

You wonder who reviewed which article. (multiple-wh reading inside the embedded question)

The in-situ quin article remains within the embedded question. Such multiple questions can be demanding and often call for contextual support.

Coordination: extract from all conjuncts or none

Extracting from only one coordinated phrase violates the coordinate-structure constraint:

❌ Quin informe va arxivar la secretària i va destruir les notes?

Unacceptable on the intended reading where quin informe is only the object of arxivar.

If one dependency is shared by parallel gaps in every conjunct, across-the-board extraction (academic) is possible:

✅ Quin informe va corregir i va enviar la secretària abans de dinar?

Which report did the secretary correct and send before lunch?

The wh-object is understood with both corregir and enviar. Parallel argument relations are essential; merely placing two verbs together does not license a mismatched dependency.

Grammaticality is not the same as processing ease

A dependency may obey the grammar and still be difficult to process. Each intervening clause makes the listener retain an unresolved wh-phrase for longer:

Quina esmena diu la portaveu que creu el comitè que el ple acabarà aprovant?

Which amendment does the spokesperson say the committee believes the plenary will ultimately approve?

The chain through selected que-complements is structurally possible, but it is heavy. In (formal) prose, repeat the relevant noun or split the sentence; in conversation, speakers almost always do. Processing discomfort does not prove an island violation, while a short sentence can still violate one.

A passive can sometimes shorten a dependency, but it changes which participant is the grammatical subject and often changes the discourse topic. Use it only when that perspective is intended; splitting or locally reformulating the question is usually a more faithful repair for an island violation.

Scope and an attachment ambiguity

Quan vas dir que arribaria la directora?

When did you say that the director would arrive? (ambiguous: when was the saying, or when was the expected arrival?)

The Catalan sentence has two attachment sites:

High: [Quan vas dir [que arribaria la directora]]? — time of saying
Low: Quan vas dir [que arribaria la directora quan]? — time of arriving

Neutral prosody often favors the low reading, but context can select either. Punctuation alone cannot show the intended scope. Write En quin moment vas fer aquesta afirmació? for the high reading or Per a quin dia vas dir que arribaria la directora? for the low one.

A natural dialogue shows how context resolves it:

—Quan vas dir que arribaria la directora? —Dilluns vaig dir que arribaria avui, cap a les sis.

—When did you say the director would arrive? —On Monday I said she would arrive today, around six.

The answer explicitly supplies both times and exposes why the isolated question was ambiguous.

Common Mistakes

Stranding a preposition like English

❌ Qui confies en per coordinar l'equip?

Incorrect — en has been stranded.

✅ En qui confies per coordinar l'equip?

Who do you trust to coordinate the team?

Doubling an extracted object with a clitic

❌ Quina factura l'heu pagada avui?

Incorrect as a neutral information question — the extracted object should leave a gap.

✅ Quina factura heu pagat avui?

Which invoice did you pay today?

Extracting from an adjunct island

❌ Quines dades vas sortir de la sala abans que publiquessin?

Incorrect on the intended reading ‘Which data did you leave the room before they published?’ — the gap is inside a temporal adjunct.

✅ Quines dades havien de publicar abans que sortissis de la sala?

Which data did they have to publish before you left the room?

Treating a difficult grammatical sentence as automatically wrong

❌ Quin projecte dius que creu la Laia que aprovarem? — impossible només perquè és llarg.

Faulty analysis — the sentence is not impossible merely because the dependency is long.

✅ Quin projecte dius que creu la Laia que aprovarem? — gramatical, però difícil de processar.

Which project do you say Laia thinks we will approve? — grammatical, but difficult to process.

Key Takeaways

Catalan wh-extraction links a fronted phrase to an argument or adjunct position, normally carrying any required preposition with it. Selected complement clauses can allow long dependencies; relative clauses, complex noun phrases, adjuncts, embedded questions, subjects, and asymmetric coordination restrict them. Finally, keep three notions apart: a grammatical gap dependency, a topic resumed by a clitic, and a sentence that is grammatical but burdens working memory.

Now practice Catalan

Reading grammar gets you part of the way. The exercises are where it sticks — free, no signup needed.

Start learning Catalan →