A Word Count of Modern Arabic Prose

Word-length distribution in modern Welsh prose texts

Although very little Celtic data has yet been examined within the Göttingen project on word-length distributions, one set of Q-Celtic data has already been processed – a set of 31 Scottish Gaelic e-mails, for which the best-fit distribution was the 1-displaced hyperpoisson distribution (Drechsler 2001). This study will add data from a P-Celtic language – Welsh – in order to obtain a preliminary...

Word Length in Estonian Prose

The present study deals with the problem of word length in Estonian prose. As is well known from quantitative and synergetic linguistics, word length is no isolated phenomenon; rather, it stands in close interrelations with word frequency, sentence and syllable length, and others, resulting in language as a dynamically balanced system. Moreover, the frequency with which words of a given length ...

Hybrid Word-Character Neural Machine Translation for Modern Standard Arabic

Traditional neural machine translation architectures use a word-level approach that assumes all important words have enumerable and relatively frequent surface forms. This assumption is invalid for the large number of non-analytic languages that form new words on the basis of complex morphological processes. Contributing to the quest of finding a universalist architecture that performs well for...

Word count:

Recognition of Modern Arabic Poems

We propose a machine learning method for recognizing modern Arabic poems based on the common poetic features of modern Arabic poetry. The poetic features include: rhyming, repetition, use of diacritics and punctuations, and text alignment. The method can classify text documents as poem or non-poem documents with a very high accuracy of 99.81%.