A dictionary can tell you that every word in prendre una decisió exists, but a corpus can show whether speakers actually combine those words, in which genres, with which inflected forms, and across which regions. This is the central question in collocation research: not “Can a reader decode it?” but “Is this the conventional way Catalan packages this meaning here?”
Corpus evidence is powerful only when interpreted. Ten identical copies of a press release are not ten independent attestations; one prolific author is not a whole speech community; web frequency alone does not turn a Spanish or English calque into an established Catalan frame. A responsible workflow combines counts with dispersion, context, regional metadata, and reference works.
For the constructions themselves, compare the inventories of light-verb frames and regional phraseological variation. This page concentrates on how to test such candidates rather than supplying another vocabulary list.
Pattern grid: what exactly are you testing?
| Candidate frame | Fixed material | Variable slots | Search families |
|---|---|---|---|
| prendre una decisió sobre X | prendre
| article, number, tense, topic | prendre/pren/prenen/va prendre decisions |
| tenir en compte X | en compte | inflected tenir, direct object | té/tenia/han tingut en compte |
| donar suport a X | suport a | inflected donar, supported entity | dona/donen/donava suport a |
| fer una ullada a X | una ullada a | inflected fer, object of looking | fa/fes/va fer una ullada, fer-hi una ullada |
| adonar-se de X / que + clause | reflexive clitic; de before a noun phrase | person, tense, noun phrase or clause | m'adono que / s'adona de / se'n va adonar |
The grid forces you to state a hypothesis. If the question is whether tenir en compte is conventional, searching only the infinitive misses té en compte, tinguem en compte, and ho havia tingut en compte. If the question is whether the object can be pronominalized, a broad lexical query may hide the decisive clitic examples.
Choose the dataset just as carefully as the query. The IEC's CTILC is a large lemmatized written corpus with genre, date, and territorial metadata; the AVL's CIVAL is indispensable for Valencian evidence and includes material from different historical periods; the University of Barcelona's CCCUB adds contemporary oral corpora organized around dialectal, social, conversational, and register variation. Filter by date and medium before making a claim about contemporary speech: a historical written attestation and a present-day conversational pattern answer different questions.
Step 1: search a family, not one quotation
Catalan verbs inflect, articles vary, and clitics move or attach according to verb form. Begin with a lemma-aware query if the corpus supports it. Otherwise, build a controlled set of surface searches. For prendre una decisió, include singular and plural nouns and several high-frequency verb forms.
Després de consultar l'equip, prendrem una decisió definitiva dilluns.
After consulting the team, we will make a final decision on Monday.
Les decisions que es van prendre ahir afecten tots els departaments.
The decisions made yesterday affect every department.
These sentences instantiate the same collocation despite different tense, number, and syntax. Searching only the exact string prendre una decisió systematically undercounts it. Conversely, an unrestricted search for decisió says little about which verb it prefers.
For clitic-bearing frames, search both an explicit complement and its pronominal realization:
Cal tenir en compte les dades que falten.
The missing data must be taken into account.
Aquest límit no l'havíem tingut en compte.
We had not taken this limitation into account.
The second example shows that the variable factor is a direct object (l'), while en remains inside the fixed phrase en compte. A raw word count cannot reveal that analysis unless you inspect concordance lines.
Step 2: compare alternatives by rate and function
Raw totals can mislead when corpora differ in size. Compare candidates within the same dataset and normalize if possible—occurrences per million words, for example. More importantly, make sure the candidates express the same function. Prendre una decisió, decidir, and adoptar una decisió do not always distribute identically: the simple verb is compact, while a noun phrase can carry modification or institutional framing.
La comissió va prendre una decisió difícil després de tres hores de debat.
The committee made a difficult decision after three hours of debate.
La comissió va decidir ajornar la votació.
The committee decided to postpone the vote.
Both are idiomatic. The noun-based frame supports difícil, unànime, or definitiva; the simple verb efficiently selects a clause or infinitive. A good corpus comparison controls for that grammatical difference rather than declaring one form “more frequent” in the abstract.
Step 3: check dispersion, not just frequency
A plausible collocation should normally occur across several documents, authors, and dates. A hundred hits concentrated in one syndicated article show repetition, not broad entrenchment. Inspect at least three kinds of dispersion:
- document dispersion: does the phrase occur in independent texts?
- speaker or author dispersion: is it shared beyond one idiolect?
- time dispersion: is it current, historical, emerging, or declining?
Exact duplicates, machine-translated pages, legal templates, and copied product descriptions should be collapsed where possible. A rare frame can still be correct—especially a technical term—but its narrow distribution must become part of the conclusion.
Step 4: separate genre and register
A collocation that thrives in parliamentary records may sound pompous in a text message; a phrase common in dialogue subtitles may be too conversational for a research article. Filter or sample by genre: conversation, journalism, administrative prose, academic writing, literature, and online interaction.
Els resultats donen suport a la hipòtesi inicial, però la mostra és reduïda.
The results support the initial hypothesis, but the sample is small. (academic)
Sí, jo també hi estic d'acord; compta amb mi.
Yes, I agree too; count me in. (informal conversation)
Both can translate an English idea of “support,” but they answer different phraseological questions. Donar suport a una hipòtesi belongs to evidential prose; compta amb mi offers personal cooperation. Pooling their translations hides the Catalan distinction.
Register also explains why a grammatically possible alternative may be marked rather than wrong. Efectuar una revisió can occur in formal administrative language, whereas fer una revisió or revisar is often more proportionate elsewhere.
Step 5: inspect regional distribution
Catalan corpora may mix Central, Valencian, Balearic, Northwestern, Northern, and Alguerese material. A form concentrated in one territory is not automatically marginal or defective. It may be the ordinary local collocation.
Pega una ullada a la sol·licitud abans d'enviar-la.
Take a look at the application before sending it. (regional: Valencian, informal-neutral)
Fes una ullada a la sol·licitud abans d'enviar-la.
Take a look at the application before sending it. (Central baseline; broadly understood)
If a query returns mostly Valencian sources for pegar una ullada, the right annotation is regional distribution, not “bad Catalan.” Compare matched genres and periods before attributing the difference to region: a Valencian newspaper and a Central chat corpus differ along more than one axis.
Consulting IEC- and AVL-oriented reference sources can help interpret the result, but neither a corpus nor a dictionary replaces the other. A reference work establishes codification and meaning; corpus lines reveal actual distribution and grammatical behavior.
Three substitutable search frames
Build searches around a fixed core plus controlled slots:
- [form of prendre] + (article/adjective) + decisió/decisions: test the verb–noun association while allowing modification.
- [form of donar] + (clitic) + suport + a + X: test the selected preposition and the range of supported entities.
- [reflexive clitic] + [form of adonar] + de + noun phrase / que + clause: test de-complements and que-clauses separately; also search se'n va adonar and its inflected relatives when en replaces a previously mentioned de-complement.
For each frame, sample concordance lines manually. Automated counts cannot reliably distinguish literal donar suport (“provide a physical support”) from idiomatic donar suport a una proposta without contextual filtering.
Literal versus idiomatic evidence
Consider donar suport. In a literal engineering context, a beam can donar suport a una estructura; in institutional or academic language, a person, organization, or finding can donar suport a una proposta o hipòtesi. The surface words are the same, but subject type and discourse domain reveal the reading.
Aquesta biga dona suport a la coberta.
This beam supports the roof. (literal physical support)
Les dades donen suport a una interpretació més prudent.
The data support a more cautious interpretation. (idiomatic academic frame)
A comprehensible but nonidiomatic candidate can also produce hits through translationese:
Els resultats suporten aquesta conclusió.
The results support this conclusion. (comprehensible English-influenced wording; standard academic Catalan prefers donar suport a)
One or two attested calques do not make the preferred collocation disappear. Corpus work measures patterns; it does not vote on correctness by isolated hit.
A miniature investigation
Suppose you want to verify the English-inspired candidate fer una decisió. First search inflected forms of fer near decisió, then the corresponding family with prendre, and finally simple decidir. Read the surrounding clause and classify every genuine decision-making use. Exclude examples such as fer pública una decisió, where fer belongs to fer pública, not to a light-verb collocation with decisió.
Next, compare document dispersion and genre. If prendre una decisió appears across journalism, conversation, and institutional prose, while fer una decisió occurs mainly in learner text or machine translation, the evidence favors prendre. The simple verb decidir remains an idiomatic alternative with different syntax.
—Has pogut verificar quina forma és habitual? —Sí. He cercat també les formes flexionades, he eliminat duplicats i he separat els textos valencians dels centrals. —I el resultat? —La construcció possible no era la preferida: els autors feien servir prendre una decisió o decidir.
—Were you able to verify which form is usual? —Yes. I also searched the inflected forms, removed duplicates, and separated Valencian texts from Central ones. —And the result? —The possible construction was not the preferred one: writers used 'prendre una decisió' or 'decidir.'
Common Mistakes
❌ He trobat tres resultats exactes per a «prendre una decisió», per tant és una expressió rara.
Faulty method: an exact infinitive search misses inflected, plural, modified, and reordered instances.
✅ He cercat les formes prendre, pren, prenen, prendrà i va prendre amb decisió i decisions.
I searched several inflected forms with both singular and plural decision nouns.
❌ La frase apareix cinc-centes vegades, així que cinc-cents autors diferents la fan servir.
Faulty inference: copied or syndicated texts can multiply one source.
✅ Després d'eliminar duplicats, la frase apareix en autors, mitjans i anys diferents.
After removing duplicates, the phrase appears across different authors, outlets, and years.
❌ Les dades suporten la hipòtesi perquè aquesta traducció literal té alguns resultats al web.
English word-by-word assembly supported only by isolated web hits; standard Catalan prefers donar suport a.
✅ Les dades donen suport a la hipòtesi.
The data support the hypothesis.
❌ No ho vam tenir compte quan vam dissenyar l'enquesta.
Incorrect frozen phrase: en belongs inside tenir en compte.
✅ No ho vam tenir en compte quan vam dissenyar l'enquesta.
We did not take it into account when we designed the survey.
⚠️ Idò, qui no s'arrisca no pisca, segons demostren les nostres dades.
The Balearic marker and proverb may be natural in conversation, but here they replace an evidential step in academic prose; the problem is register and argumentation, not grammaticality.
✅ Les dades suggereixen una associació, però cal una mostra més gran per confirmar-la.
The data suggest an association, but a larger sample is needed to confirm it. (academic)
Key Takeaways
- Search inflected families, articles, number variants, clitics, and selected prepositions—not only one exact string.
- Compare normalized frequency only after ensuring that candidates perform the same grammatical and pragmatic function.
- Check dispersion across documents, authors, dates, genres, and regions; repeated text is not independent evidence.
- Read concordance lines to separate literal meanings, idioms, translationese, and unrelated syntactic structures.
- A regional concentration calls for a regional label, not automatic rejection; triangulate corpus evidence with IEC- and AVL-oriented references.
Now practice Catalan
Reading grammar gets you part of the way. The exercises are where it sticks — free, no signup needed.
Start learning Catalan →Related Topics
- High-Frequency Verb–Noun CollocationsA2 — Use fer, tenir, donar, prendre, posar, and treure with the nouns Catalan conventionally selects instead of translating English combinations word by word.
- Light-Verb ConstructionsB2 — Use Catalan support verbs such as fer, donar, prendre, and tenir with the nouns they conventionally select.
- Regional Phraseological VariationC1 — Recognize regionally preferred Catalan phraseological frames without treating one variety as the only authentic model.
- Academic CollocationsC1 — Build precise Catalan academic prose with conventional research, argumentation, and evidence frames without bureaucratic inflation.
- Avoiding Translationese from English and SpanishC2 — How to revise calqued word order, passives, nominalizations, connectors, and collocations according to Catalan information structure.
- One Language, Pluricentric StandardsA2 — Understand how IEC and AVL codification, regional standard forms, register, and mutual intelligibility fit within one pluricentric Catalan language.