ARIES: A lexical platform for engineering Spanish processing tools

We present a lexical platform t h a t has been developed for the Spanish language. It achieves portabi l i ty between different computer systems and efficiency, in t e rms of lexical coverage. A model for t he full t r e a tmen t of Spanish inflectional morphology for verbs, nouns and adjectives is presented. This model permi t s word formation based solely on morpheme concatenat ion, driven by a feature-based unification g rammar . The runt ime lexicon is a collection of al lomorphs for b o t h s tems and endings. Al though not tested, it should be suitable also for other highly inflected languages, such as R o m a n ones. A formalism is also described for encoding a lemma-based lexical source, well suited for expressing linguistic generalizations: inheri tance classes, l emma encoding, morpho-graphemic al lomorphy rules and l imited type-checking. From this source base, we can automat ical ly generate an al lomorph indexed dict ionary adequa te for efficient retrieval and processing. A set of software tools has been implemented a round this formalism: lexical base augment ing aids, lexical compilers to build runt ime dictionaries and access libraries for t hem, feature manipula t ion libraries, unification and pseudo-unification modules , morphological processors, a parsing system, etc. Software interfaces among the different modules and tools are cleanly defined to ease software integrat ion and tool combinat ion in a flexible way. Directions for accessing our e-mail and web demons t ra t ion pro to types are also provided. This work has been suppor ted in par t by t he Spanish Plan Nacional de I+D, t h rough the project TIC91-0217C02-01, ARIES: An Architecture for Natural Language InterfacES with User Modelling. Some figures are given also, showing the lexical coverage of our platform compared to some popular spelling checkers.