Developing an interlingual translation lexicon using WordNets and Grammatical Framework

LinkedDictAbs = Cat ** { fun consecutive_01624944_A : A ; fun consequently_00061939_Adv : Adv ; fun conservation_06171333_N : N ; fun conspire_00562077_V : V ; fun sing_01362553_V2 : V2 ; ........ } The first line in the above given code states that the module ‘LinkedDictAbs’ is an abstract representation (note the keyword ‘abstract’). This module extends (achieved by ‘**’ operator) another module labeled ‘Cat2’ which, in this case, has definitions for the morphological categories ‘A’, ‘Adv’, ‘N’ and ‘V’. These categories correspond to the ‘adjective’, ‘adverb’, ‘noun’, and ‘verb’ categories in the WordNet respectively. However, note that in GF resource grammars we have a fine-grained morphological division for verbs. We sub-categorize them according to their valencies i.e ‘V’ is for intransitive, and ‘V2’ for transitive verbs. We refer to (Bringert et al., 2011) for more details on these divisions. Each entry in this module is of the following general type: fun lemma_senseOffset_t : t ; Keyword ‘fun’ declares each entry as a function of the type ‘t’. The function name is composed of lemma, sense offset and a type ‘t’, where lemma and sense offset are same as in the Princeton WordNet, while ‘t’ is one of the morphological types in GF resource grammars. This abstract representation will serve as a pivot for all concrete representations, which are described next. 2This module has definitions of different morphological and syntactic categories in the GF resource grammar library