Effort and accuracy during language resource generation: a pronunciation prediction case study

Abstract

When developing a language resource, there is generally a trade-off between the amount of effort invested in the resource creation process and the quality of the resulting resource. We (authors) argue that, in the developing world with its many resource-scarce languages, a ‘usable’ resource in multiple languages may be more valuable than a highly accurate resource for one language only. From this perspective we (the authors) investigate the resource validation process – determining whether a resource is sufficiently accurate– using the creation of a pronunciation dictionary as case study. We (the authors) show that the amount of effort required to validate a 20,000-word pronunciation dictionary can be reduced substantially by employing appropriate computational tools, when compared to both a fully manual validation process and a competing automatic process.