The aim of this PhD thesis is to develop automatic methods for Noun Phrase Recognition in Modern Greek corpora. The linguistic research is placed within the framework of Harris' distributional and transformational theory (1951;. For the detailed and systematic description of Modern Greek syntax, we adopted Gross ' methodological model (1975), called Lexicon-Grammar, which can provide extremely rich and exhaustive linguistic information. According to Gross (1975), in order to proceed to syntactic analysis, we should first study the structure of elementary sentences. Based on the structure of elementary sentences in Modern Greek, we delimit our research and define noun phrases within our research. More specifically, we focus on noun phrases that are non-prepositional first complements of verbs.As far as Noun Phrase Recognition Is concerned, we constructed and applied to corpora machine-readable grammars. These grammars, which have the formalism of Recursive Transition Networks (RTN), are very often used to represent linguistic data (Gross, 1993;Roche & Schabès, 1997). In total, we constructed 925 grammars (graphs), consisting a linguistic resource of great importance as they recognise a big part of noun phrases. Because of their detailed descriptions, these grammars were also used to eliminate morphological ambiguities in noun phrases. The latter task was evaluated in terms of recall and precision.