Fun with Unix Tools

Eileen Fitzpatrick
Steve Seegmiller

Montclair State University
Linguistics Department
Montclair State University
Upper Montclair, NJ 07043

The Unix operating system provides a set of flexible text processing tools that offer the user features beyond those of standard concordancers including the ability to compare and manipulate different types of text. We show how simple tools can combine to accomplish sophisticated tasks, using examples from lexicography and phonology.

The creation of an English-Karachay (a Turkic language) dictionary involves checking elicitations from native speakers, grammars, and glossaries against a series of authentic texts for accuracy, citations, and words that do not appear in our sources. Unix tools allow us to find words in context in the manner of traditional concordancers, but they also permit the creation of three lists -- words only in the dictionary, words only in the texts, and words in both -- that enable us to decide which words should be omitted from and which words entered into the dictionary.

Spanish text-to-speech systems are set at a faster rate than English systems. One reason for this might be that Spanish words and phrases are longer than their English counterparts. Unix tools allow us to estimate average syllable length per word by counting vowels in English and Spanish pronouncing dictionaries and showing that, on average, Spanish words have twice as many syllables as English words. Word counts of parallel corpora show that Spanish also has more words per text than English.

The operations discussed here involve standard commands available on any Unix (or Linux) system, do not require extensive training to use, and are re-usable for widely varying applications.