Syntax-Directed Recognition of Spoken Words in Real-Time