This paper presents a grammatical model for the air traffic controller's (ATC) commands using finite-state transition networks (FSTN). The grammatical representation is used by a syntactic parser and recognizer for the analysis of the grammatical structure of the commands. A grammatical description using FSTN is proposed for the ATC's commands assigning word categories and syntactic structure that can be followed by a syntactic parser for recognition and parsing.

This paper, also, presents an innovative model using "skip loops" for the implementation of a syntactic parser using finite-state transition networks to delete and remove incorrect or out of syntax words. These words could be the effect of mixed streams of words or errors in the conversion from spoken language to characters. The skip loop is an arc that allows the finite-state transition automata (FSTA) to delete a word that does not match the grammatical structure of the sentence, and continue the recognition process without affecting the syntactical definition of the sentence. This particular approach is especially useful in areas such as the command language of the air traffic controller (ATC)

The model uses FSTN with skip loops to model and recognize ATC's command language. The use of skip loops allows to delete words that may be present in the statement that are unrecognized or that do not fit into the grammatical structure of the ATC's language. This technique facilitates the recognition of the statement minimizing the possibility of declaring the statement as ill-formed. Two syntactic parser prototypes are implemented using Prolog and CLIPS.

These techniques are useful in applications like military tactical environments that are exposed to rapidly changing commands, streams of information, and different sources of background noise. Many critical decisions have to be made extracting the correct information from multiple input streams making difficult and uncertain the selection of the correct input information. The method presented introduces a certain degree of intelligence using current AI techniques to obtain an intelligent syntactic parsing of the input information. The parser syntax can be defined to dynamically adjust its model to follow a particular stream of information that sounds or looks appropriate for the particular context. The purpose of the parser will be to model the process that resembles the human ability to follow a single dialog in an environment where there are many conversations and background noise.