What it is
Some units are fixed and open at once. A frame settles part of the wording and leaves a position free, and that position is filled again on every occurrence. This is why frames are so often translated from scratch: the variable part moves, so the whole looks new and the settled part goes unrecognized.
Renouf and Sinclair gave the construct its first shape, a discontinuous pair of function words standing one word apart, with [a __ of], [an __ of], [too __ to] and [many __ of] as their worked cases. Their interest was in what fills the middle, and their finding is that the fillers are not an open set: they cluster on quantities, on part-whole relations and on types. That coherence is their argument that a frame is a unit rather than an accident of adjacency. Wray states the wider claim, that a sequence, continuous or not, can be stored and retrieved whole instead of being assembled by the grammar.
Two boundaries. A bundle is contiguous and admitted by a count, while a frame is discontinuous and defined by the position it leaves open, which is why a frequency list of strings does not contain it. Where a term fills the slot, the term record keeps the term and this record keeps the frame around it.
Where the fixity comes from
Nothing prescribes a frame. It is attested, but harder to read than a contiguous string, because the evidence is a distribution over fillers rather than a count of one sequence.
- The two measures disagree. Ranking fillers by how often they occur inside the frame is not the same as ranking them by how often they occur at all: of the twenty words Renouf and Sinclair list for [a __ of], man is the most frequent in the corpus at large and lot only the ninth, while lot heads the frame. They read that promotion as evidence of a tight association.
- Their second measure, the share of a word's occurrences falling inside the frame, is fragile. Stefanowitsch's replication on the British National Corpus shows the ranking survives only because those twenty words were pre-selected by raw frequency: without it, 477 words occur exclusively inside the frame and all score 100 per cent, misspellings included.
- A frame is worth no more than the association measure behind it, and a bare proportion is a poor one.
Consequences for translation
The frame is settled and the slot is not, and the two need different handling.
- Settle the frame once and vary only the slot. Translating it again at each occurrence produces variants of something that did not change, and the document ends up carrying three shapes of one construction.
- The slot is a class and not a blank. What may go into it is restricted, so a filler that is correct on its own can be wrong inside the frame.
- A frame is discontinuous, so it does not survive a search for a string and it can straddle a segment boundary. Checking consistency means looking for the frame's two ends, not for a phrase.
- Where an instrument prescribes the frame, as with certification wording and prescribed forms, the warrant changes and another record takes it over. The handling does not change: the frame is placed, not composed.
Examples
EN>NL
- Source: in one of the cases referred to in Article 2(4)
- Target: in een van de in artikel 2, lid 4, bedoelde gevallen
- Comment: Council Framework Decision 2002/584/JHA, Article 4, point 1. The English frame follows the noun and holds together. The Dutch one wraps around the slot, so de ... bedoelde gevallen is interrupted by the reference it governs, and a search for a string will not find it.
EN>DE
- Source: in one of the cases referred to in Article 2(4)
- Target: in einem der in Artikel 2 Absatz 4 genannten Fälle
- Comment: Same provision. German wraps as well, and the frame reaches into its own surroundings: the participle takes its ending from the noun after the slot. Two of these languages put the frame around the reference and English puts it after: a fact about the frame, not about this sentence.
EN>ES
- Source: in one of the cases referred to in Article 2(4)
- Target: en uno de los casos citados en el apartado 4 del artículo 2
- Comment: Same provision once more. Spanish keeps the frame after the noun, as English does, and reorders the slot itself, naming the paragraph before the article. Neither the frame's position nor the slot's contents carry over unchanged.
References
- Renouf, A. and Sinclair, J. (1991). "Collocational frameworks in English." In K. Aijmer and B. Altenberg (eds), English Corpus Linguistics: Studies in Honour of Jan Svartvik, 128 to 143. London: Longman.
- Wray, A. (2002). Formulaic Language and the Lexicon. Cambridge: Cambridge University Press.
- Stefanowitsch, A. (2020). Corpus Linguistics: A Guide to the Methodology. Textbooks in Language Sciences 7. Berlin: Language Science Press, at pp. 263 to 266.
- Council Framework Decision 2002/584/JHA of 13 June 2002 on the European arrest warrant and the surrender procedures between Member States, Article 4, point 1, English, Dutch, German and Spanish language versions.