Lexical Functional Grammar in speech recognition

~he syntax component of the speech recognition system IKAROS t is described, clhe usefidness of a probabilistic Le~jcal Functional Grammar both for cow,straining bottom-up hypotheses and top-down predicting is showtL 1. I n t r o d u c t i o n The most impor t an t p rob l em in al l speech recognit ion systems is the inherent uncer ta inty associa ted with the acous t ic -phonet ic decoding process at the basis of such a system. One approach taken in many existing system to overcome these difficulties is to integrate higher level knowledge sources that have a certain a-pr ior i knowledge about specific problem areas. Following this line of thought, the system architecture adopted in the I K A R O S p r o j e c t a ssumes d i f f e ren t l eve ls of k n o w l e d g e ( r e p r e s e n t a t i o n s ) e .g . a c o u s t i c p a r a m e t e r s , p h o n e m e s , w o r d s , c o n s t i t u e n t s t ructures etc. The in te rac t ion between these knowledge sources is cont ro l led by a central blackboard control module (like in HEARSAY II). This whole system is embedded in an objectoriented environment and communicat ion between the modules is realized by message passing. Within IKAROS particular attention is given to the p r o b l e m of us ing the same k n o w l e d g e representa t ions both for da ta -dr iven bot tom-up hypo thes iz ing and expec t a t i on -d r iven top-down predict ion and to the problem of provid ing a general f ramework of uncer ta in ty management . According to this rationale, the main purpose of the syntax component is to constrain the number of word sequences to be dealt with in the recognition process and to predict or insert poorly recognized words. Grammaticaless in itself is of no importance to us. Quite to the contrary , in a rea l l ive application a certain degree of error tolerance is a desired "effect. 1 Research in IKAROS is partially funded by the ESPRIT programme of th6 European Community under contract P954 In the syntax component of IKAROS we work within the formal f ramework of a p r o b a b i l i s t i c Lexical Functional Grammar. Certain modif icat ions to the formalism as expounded in / B r e s n a n 1 9 8 2 / have been made to suit our purposes. We use as an implementat ion an event -dr iven char t -parser that is capable of all the necessary parsing strategies i .e . top-down, bot tom-up and lef t to-r ight and rightto-left parsing. 2. Probabi l i s t ic context . f ree G r a m m a r s 2 .1 . The event-dr iven p a r s e r The interact ion between the b lackboard manager and the syntax component is roughly as fol lows: the b lackboard manager sends a message to the syntax component indicating that a part icular word has been recognized (or rather "hypothesized") at a certain posit ion in the input stream (or in charto parser terminology with starting and ending ve r t e~ ~ number ) toge the r wi th a cer ta in numer ica l c o n f i d e n c e score . The syn t ax c o m p o n e n t accumulates information about these (in arbitrary order) incoming word hypotheses and in turn posts hypotheses about predicted and recognized words or constituents on the blackboard. The job of the syntax c o m p o n e n t now is to decide between several conf l i c t ing (or compe t ing) cons t i tuen t structures stored in the chart i.e. to choose the best grammatical structure. 2 .2 . The f o r m a l i s m We assume a probabil ist ic context-free grammar G = < V N , VT, R , S > : VN denotes the nonterminal vocabulary Nonterminals are denoted by A, B, C .... strings of these by X, Y, Z... lexical categories by P, Q . . . . VT denotes the terminal vocabulary terminals (words) denoted by a, b, c . . . . . strings of both types of symbols are denoted by w, x, y, z . R denotes the set of rules {R1, R2 . . . . . Ri} with each rule having the format Ri = < Ai -> Xi , qi > where qi indicates the a-priori