Collocations = typical word combinations
We are suppliers of high-quality collocation databases in many languages. The lists are generated automatically from an enormous authentic database of text (text corpora) produced by real users of the language. Our largest corpora in major languages contain texts with a total length of over 10,000,000,000 words. The use of authentic texts as a source ensures that the database reflects the real-life use of the language.
Collocation databases are important for language learning products and also for digital typing assistants designed to suggest more natural and idiomatically correct words.
Collocations are identified by the Word Sketch technology in Sketch Engine and supplied as a database download in a number of formats.
A relatively small corpus is generally sufficient to generate linguistic data for the most frequent 5,000 words because such words appear frequently enough in any text.
However, an enormous text database (corpus) is required to generate the same information for rare and infrequently used words. The only viable option of building corpora of billions of words is using an automatic procedure of downloading content from the web. Lexical Computing developed a sophisticated procedure for collecting only linguistically valuable content. A series of tools is used to focus on the right content and to deduplicate and clean the downloaded content. This ensures that the statistics are not skewed. This blog post gives more details.
Collocation database size
We are able to generate collocation databases of millions of items for all major languages. The size of the database for less major languages will be affected by the size of the source corpus. The actual size depends on the specifications provided by the customer.
Enriched collocation database
We are also able to provide additional information such as POS tags, lemmas, frequency ranks, or any other statistics or morphological information.
Collocation database sample
The easiest is to register a free trial account in Sketch Engine and use the word sketch tool to generate a wordlist. The advanced tab of the wordlist tool allows for detailed specifications to be used.
We will provide a quotation based on the exact specifications and the intended use of the database.
Collocation database download
The database will be made for download to you on a dedicated link within the agreed period of time. It normally takes a week or two to generate the data. Very complex databases can be computationally demanding and can take longer to produce.
Our natural language processing API supports the retrieval of collocations in real time.
The first few collocations of the Spanish word problema together with frequency and strength of collocation value.
object of ---------------------- resolver 179825 11.38 solucionar 116224 10.99 tener 401288 9.09 enfrentar 29200 8.82 haber 143262 8.75 causar 22885 8.41 plantear 21728 8.33 evitar 26164 8.3
Collocation database in these languages
Collocations for more languages can be made available or developed on request.