Many of the methods developed for Western European languages and used widespread to produce annotated language resources cannot readily be applied to Central and Eastern European languages, due to the large number of novel phenomena exhibited in the syntax and morphology of these languages, which these methods have to handle but have not been designed to cope with. The process of morphological tagging when applied to Hungarian data to produce corpora annotated at least at the morphosyntactic level is most indicative of this problem: several of the algorithms (either rule-based or statistical) that have been used very successfully in other domains cannot readily be applied to a language exhibiting such a varied morphology and huge number of wordforms as Hungarian. The paper will describe a robust tagging scenario for Hungarian using a relatively simple stochastic system augmented with external morphological processing, which can overcome the two most conspcicuous problems: the complexity of morphosyntactic descriptions and most importantly the huge number of possible wordforms.

A speech recognizer is a device that translates speech into text. Many current speech recognizers contain two components, an acoustic model and a statistical language model. The acoustic model indicates how likely it is that a certain word corresponds to a part of the acoustic signal (e.g. the speech). The statistical language model indicates how likely it is that a certain word will be spoken next, given the words recognized so far. Even though the acoustic model might for example not be able to decide between the acoustically similar words "peach" and "teach", the statistical language model can indicate that the word "peach" is more likely if the previously recognized words are "He ate the". Current speech recognizers perform well on constrained tasks, but the goal of continuous, speaker independent speech recognition in potentially noisy environments with a very large vocabulary has not been reached so far. How can statistical language models be improved so that more complex tasks c...

NYU's research objective is to determine whether non-local, linguistically-based word preferences can be used to enhance speech recognition. We are working jointly with BBN, and our system takes as input the N-best hypotheses generated by BBN (with acoustic and n-gram language model scores for each hypothesis). Our goal is to generate scores based on both intersentential dependencies (related to topic coherence) and intrasentential dependencies (connected by syntactic relations) to complement the usual contiguous-word (n-gram) dependencies. We describe our sublanguagemodel, which is intended to capture the effects on vocabulary of topic coherence within an article. We report several measures of this model, including its effect on word error rate when combined with the BBN acoustic and language model scores. We also briefly describe our initial efforts at applying a syntactic language model, and a word model using syntactic relations (a "semantic " model). 1. Overview NYU's research o...

This paper describes work performed withing the CRATER (Corpus Resources And Terminology ExtRaction, MLAP-93/20) project, funded by the Commission of the European Communities. In particular, it addresses the issue of adapting the Xerox Tagger to Spanish in order to tag the Spanish version of the ITU (International Telecommunications Union) corpus. The model implemented by this tagger is briefly presented along with some modifications performed on it in order to use some parameters not probabilistically estimated. Initial decisions, like the tagset, the lexicon and the training corpus are also discussed. Finally, results are presented and the benefits of the mixed model justified. 1 Introduction This paper describes the adaptation work carried out to retarget the Xerox Tagger to Spanish 1 . The Xerox Tagger [Cutting et al., 1992] has as one of its virtues the characteristic of being based on a simple probabilistic model, as it will become clear below. It is also claimed to be languag...

In this article, we apply to natural language parsing and tagging the device of triggerpair predictors, previously employed exclusively within the field of language modelling for speech recognition. Given the task of predicting the correct rule to associate with a parse-tree node, or the correct tag to associate with a word of text, and assuming a particular class of parsing or tagging model, we quantify the information gain realized by taking account of rule or tag trigger-pair predictors, i.e.

Conversational speech recognition is a very challenging task due to the large amount of variability compared to read speech and the corresponding lack of training data. Where sources of variability are systematic, however, recognition performance can be improved by modifying the structure of the language and/or acoustic model, which mainly comprise a speech recognizer. The focus of this thesis is on incorporating the discourse structure of conversational speech into a language model using mixture distributions. We extend previous work in this area with improved estimation techniques that use clustering to reduce model order, class-based smoothing techniques, and a new strategy for unsupervised training to use additional unlabeled data. In addition, we introduce unsupervised dynamic cache adaptation in order to capture topic changes as well as discourse dynamics. Experimental results on the Switchboard corpus show that discourse mixtures give better results than topic mixtures, with the best discourse mixture model giving an 1.9% reduction in word error rate over a trigram language model. Further gains are achieved by adding a dynamic cache.

Abstract not found

In this article, we apply to natural language parsing and tagging the device of triggerpair predictors, previously employed exclusively within the field of language modelling for speech recognition. Given the task of predicting the correct rule to associate with a parse-tree node, or the correct tag to associate with a word of text, and assuming a particular class of parsing or tagging model, we quantify the information gain realized by taking account of rule or tag trigger-pair predictors, i.e. pairs

Abstract not found

Abstract not found