Dictionary of Combinations
Words, like people, do not exist in a vacuum; it is only through their association with others that they take on their full meaning. Antidote’s dictionary of combinations provides a comprehensive inventory of word associations. Refresh one’s memory, a dazzling display, tender care—in all, there are hundreds of thousands of such word combinations that exemplify the lexical associations which help give the language its particular colour.
To create this enormous compilation, we put our computers to work in two ways. First, we scoured the Web to build a corpus of close to eight billion words, or 400 million sentences, drawn from many different sources. Among these are news sites such as The Washington Post, The Guardian and The Globe and Mail, as well as digital libraries such as Booksie and Project Gutenberg.
We then used Antidote’s analyzer to extract the most interesting associations from this huge corpus. We isolated the nouns and their modifiers, verbs and their complements, adjectives and their adverbs, and other such grammatical relations. A statistical filter allowed us to retain the most salient combinations, i.e., those that are significantly frequent and distinct. Thanks to the power of the analyzer, even distant combinations were discovered, through a relative pronoun or helping verb, for example, which increases the accuracy of the statistics relative to classic methods of simple proximity.
The resulting combinations were first classified by meaning whenever one of the words was polysemous, and then by syntactic role: modifiers, subjects, complements, etc. In addition, we chose to display each combination in full, with its most frequent determiner and inflected form, as for example love at first sight, a mother’s love, and love of one’s country. The result is an ordered list that is complete and easy to read.
In order to illustrate each combination, we developed a heuristic to select example sentences in accordance with various criteria. Hundreds of thousands of sentences from our corpus were extracted, providing on average more than two examples for each combination. The examples are displayed in a column on the right when a combination is selected, and they help to understand its meaning and use. Alternatively, you can also examine the definitions of the combining term.
Throughout the entire process, our linguists diligently reviewed the data, marking perceived combinations as familiar or slang. They also filtered out many instances of inappropriate or improperly formulated examples, problems that our automatic extraction heuristic could obviously not evaluate.
To help you find the perfect adjective for a given noun, the most suitable verb or the most evocative adverb, the dictionary of combinations, with its unique perspective on the language from the associative angle, will quickly become an indispensable aid.