Khaled Shaalan

The interlingua approach to Machine Translation (MT) aims to achieve the translation task in two independent steps. First, the meanings of source language sentences are represented in an intermediate (interlingua) representation. Then, sentences of the target language are generated from those meaning representations. In the generation of the target sentence, determining sentence structures becomes more difficult, especially when the interlingua does not contain any syntactic information. Hence, the sentence structures cannot be transferred exactly from the interlingua representations. In this paper, we present a mapping approach for taskoriented interlingua-based spoken dialogue that transforms an interlingua representation, so-called Interchange Format (IF), into a feature structure (FS) that reflects the syntactic structure of the target Arabic sentence. This approach addresses the handling of the problem of Arabic syntactic structure determination in the interlingua approach. A mapper is developed primarily within the framework of the NESPOLE! (NEgotiating through SPOken Language in E-commerce) multilingual speech-to-speech MT project. The IF-to-Arabic FS mapper is implemented in SICStus Prolog. Examples of Arabic syntactic mapping, using the output from the English analyzer provided by Carnegie Mellon University (CMU), will illustrate how the system works.

Arabic is a Semitic language that is rich in its morphology and syntax. The very numerous and complex grammar rules of the language could be confusing even for Arabic native speakers. Many Arabic intelligent computer-assisted language-learning (ICALL) systems have neither deep error analysis nor sophisticated error handling. In this paper, we report an attempt at developing an error analyzer and error handler for Arabic as an important part of the Arabic ICALL system. In this system, the learners are encouraged to construct sentences freely in various contexts and are guided to recognize by themselves the errors or inappropriate usage of their language constructs. We used natural language processing (NLP) tools such as a morphological analyzer and a syntax analyzer for error analysis and to give feedback to the learner. Furthermore, we propose a mechanism of correction by the learner, which allows the learner to correct the typed sentence independently. This will result in the learner being able to figure out what the error is. Examples of error analysis and error handling will be given and will illustrate how the system works.

Semantic analysis is the system that takes as input a sentence and outputs a list of prominent concepts that characterize the contents of the input sentence, and for each concept, gives the set of attributes that discuss the concept along with their relevancies. This paper presents a system that employs a machine learning approach that automates the semantic analysis process of spoken Arabic into interlingua representation. An experiment has been conducted to measure the performance of our approach. The results were promising and assured the ability of this approach in capturing the semantics of Arabic utterances taken from the travel and tourism domain.

This paper presents a novel Arabic morphological generator (AMG) for Modern Standard Arabic (MSA) which is designed and implemented using Prolog. The AMG is used to generate inflected forms of words used for the indexing of Arabic audio. These words are also the relevant terms in the Arab authority system (library information retrieval system) used in this study. The AMG generates inflected Arabic words from the root according to pre-specified morphological features that can be extended as needed. The Arabic word is represented as a feature structure which is handled through unification during the morphological generation process. The inflected forms can then be inserted automatically into a speech recognition grammar which is used to identify these words in an audio sequence or utterance.