On h-Lexicalized Automata and h-Syntactic Analysis

Following some previous studies on list automata and restarting automata, we introduce a generalized and refined model – the h-lexicalized restarting list automaton (LxRLAW). We argue that this model is useful for expressing transparent variants of lexicalized syntactic analysis, and analysis by reduction in computational linguistics. We present several subclasses of LxRLAW and provide some variants and some extensions of the Chomsky hierarchy, including the variant for the lexicalized syntactic analysis. We compare the input languages, which are the languages traditionally considered in automata theory, to the so-called basic and h-proper languages. The basic and h-proper languages allow stressing the transparency of h-lexicalized restarting automata for a superclass of the context-free languages by the so-called complete correctness preserving property. Such a type of transparency cannot be achieved for the whole class of contextfree languages by traditional input languages. The transparency of h-lexicalized restarting automata is illustrated by two types of hierarchies which separate the classes of infinite and the classes of finite languages by the same tools.