Quantifying written ambiguities in tone languages: A comparative study of Elip, Mbelime, and Eastern Dan

International audience Whether tone should be represented in writing, and if so how much, is one of the most formidable challenges facing those developing orthographies for tone languages. Various researchers have attempted to quantify the level of written ambiguity in a language if tone is not marked, but these contributions are not easily comparable because they use different measurement criteria. This article presents a first attempt to develop a standardized instrument and evaluate its potential. The method is exemplified using four narrative texts translated into Elip, Mbelime, and Eastern Dan. It lists all distinct written word forms that are homographs if tone is not marked, discarding repeated words, homophony, and polysemy, as well as pairs that never share the same syntactic slot. It treats lexical and grammatical tone separately, while acknowledging that these two functions often coincide. The results show that the level of written ambiguity in Elip is weighted towards the grammar, while in Mbelime many ambiguities occur at the point where lexical and grammatical tone coincide. As for Eastern Dan, with its profusion of nominal and verbal minimal pairs, not to mention pronouns, case markers, predicative markers, and other parts of speech, the level of written ambiguity if tone is not marked is by far the highest of the three languages. The article ends with some suggestions of how the methodology might be refined, by reporting some experimental data that provide only limited proof of the need to mark tone fully, and by describing how full tone marking has survived recent spelling reforms in all three languages.