This paper describes a model that has been developed in the Turgama Project at Leiden University to meet the challenges encountered in the computational analysis of ancient Syriac Biblical manuscripts. The small size of the corpus, the absence of native speakers, and the variation attested in the multitude of textual witnesses require a model of encoding---rather than tagging---that moves from the formal distributional registration of linguistic elements to functional deductions. The model is illuminated by an example from verb inflection. It shows how a corpus-based analysis can improve upon the inflectional paradigms given in traditional grammars and how the various orthographic representations can be accounted for by an encoding system that registers both the paradigmatic forms and their attested realizations.
Karamel is a system for finite-state morphology which is multi-tape and uses a typed Cartesian product to relate tapes in a structured way. It implements statically compiled feature structures. Its language allows the use of regular expressions and Generalized Restriction rules to define multi-tape transducers. Both simultaneous and successive application of local constraints are possible. This system is interesting for describing rich and structured morphologies such as the morphology of Semitic languages.
Fully unsupervised pattern-based methods for discovery of word categories have been proven to be useful in several languages. The majority of these methods rely on the existence of function words as separate text units. However, in morphology-rich languages, in particular Semitic languages such as Hebrew and Arabic, the equivalents of such function words are usually written as morphemes attached as prefixes to other words. As a result, they are missed by word-based pattern discovery methods, causing many useful patterns to be undetected and a drastic deterioration in performance. To enable high quality lexical category acquisition, we propose a simple unsupervised word segmentation algorithm that separates these morphemes. We study the performance of the algorithm for Hebrew and Arabic, and show that it indeed improves a state-of-art unsupervised concept acquisition algorithm in Hebrew.
Various methods have been devised to produce morphological analyzers and generators for Semitic languages, ranging from methods based on widely used finite-state technologies to very specific solutions designed for a specific language or problem. Since the earliest proposals of how to adopt the elsewhere successful finite-state methods to root-and-pattern morphologies, the solution of encoding Semitic grammars using multi-tape automata has resurfaced on a regular basis. Multi-tape automata, however, require specific algorithms and reimplementation of finite-state operators across the board, and hence such technology has not been readily available to linguists. This paper, using an actual Arabic grammar as a case study, describes an approach to encoding multi-tape automata on a single tape that can be implemented using any standard finite-automaton toolkit.
The adoption of semantic Query Expansion (QE) could be useful in the context of Question/Answering (Q/A) systems. For the Arabic language this is a challenging task since it has many particularities (short vowels, absence of capital letters, complex morphology, etc.). This paper presents an evaluation of a proposed semantic QE based on Arabic WordNet (AWN). Two types of experiments are conducted: the keyword-based evaluation which uses a classical search engine as passage retrieval system, and the structure-based evaluation that uses the Java Information Retrieval System (JIRS) which takes into account the structure of the question. Results show that the best performances in terms of accuracy and Mean Reciprocal Rank are reached when the proposed semantic QE together with JIRS are used.
