Análisis de los fenómenos lingísticos de los mensajes de correo electrónico en catalán desde la perspectiva de la traducción automática

Emails contain linguistic phenomena that deviate from standard language norms and may cause machine translation errors. In order to design an email translation environment for the Interlingua project developed at UOC, we classified deviations from the standard in a corpus of emails in Catalan and analysed their impact on the machine translation quality in Spanish. Here we present this analysis. Curiously, most translation errors are caused by the lack of linguistic competence of the sender. The impact of characteristic email phenomena (smileys, performance errors, sms vocabulary, etc.) is not so strong.