Concordances Syntagmatiques et Analyse de Surface

English Abstract

This paper describes two procedures for the establishment of syntagmatic concordances through pre-coding and automatical analysis for the nominal syntagm. The first is pre-coding of the text, in this instance the poetry of P. Valéry, in accordance with taxonomic grids enumerating the typical patterns followed by the syntagm to be analyzed. From a strictly mathematical point of view, there are too many possible combinations (several thousand), and the number of typical patterns is therefore reduced as much as possible. In these taxonomic grids the analytical structures are poor. The second is the automatic analysis of simultaneous occurrences to left and right of a pivot term set by pre-coding. All the categories known as the “parts of speech” are precoded. In this process of recognition, the categories comprised by the units of the chain and their syntactical functions are also pre-coded.

The grid is based on the shortest sequence to include the six following descriptive factors: 1. determiner; 2. antenominal adjective; 3. the noun, pivot of the syntagm; 4. postnominal adjective, verbal adjective; 5. syntagmatic expansion a (prepositionde in French and other prepositions), 6. syntagmatic expansion b (so-called relative clauses introduced byqui, que, dont, etc.). Four different syntagmatic concordances have been achieved: 1. according to the identical distribution of the elements on each side of the pivot term of the nominal syntagm; 2. identical distribution and subclassification according to the function of the noun in the syntagm; 3. the same type as 2 with a reversal of the preceding classification; 4. classification only according to the noun's function in the syntagm.