This tool will allow you to build complex dictionary entries and context overviews for a given word on the fly. Word relations are based on Constraint Grammar dependency analysis and grammatical functions, not just co-occurence. Relative and absolute frequency values are provided for each relation. Frequency values in red can be clicked to see detailed statistics and a set of corpus example sentences in concordance format. Languages marked with '*' do not require a subscription, just a login, and words beginning in 's' (e.g. 'sleep') are free for testing in all languages.
- Noun: car, horse, belief, promise
- Verb: eat, caress, promise, ask
- Adverb: carefully, tenderly
- Adjectives: big, large, high, tall
Advanced Option Reference
Lexical frequency threshold can be used to filter away rare words from the relations lists, independently of the frequency of the relation as such
- high = only very common words
- medium = common words (small dictionary)
- low = also some rarer words are listed (large dictionary)
- none = no lexical frequency filtering (unabridged dictionary)
Minimum occurrence can be set to rule out rare relations, or to explicitly include them. The default of 2 filters out some corpus anomalies and annotation errors, but for a full overview, use 1 (e.g. for lexicography). Values above 5 will index only very common patterns (e.g. for teaching). In the output, occurrence classes are added after the relative frequency values, separated by a colon. 4.27:2, for instance, means that the relative frequency of the relation is 4.27, and the occurrence class 2. Occurrence classes are calculated as the integer of the dual logarithm of the absolute number of occurrences, i.e. 0 (1 occurrence), 1 (2-3 occurrences), 2 (4-7 occurrences), 3 (8-15 occurrences) etc. 9 (more than 512 occurrences) is the cut-off value. Occurrence classes of 3 or higher are boldfaced.
Minimum relative frequency sets a threshold for co-occurence strength (mutual information) between the word and a given relation. The value is not intended to be used in absolute terms but allows users to focus on the most typical examples only.
Show top limits the number of relations shown for a given type (e.g. premodifiers or subjects) to the top n examples with the highest relative frequency.