Computerisation of phrase-to-phrase matching from a standard marine communication phrases corpusA preliminary empirical study

Resumen

Standard Marine Communication Phrases (IMO, 2002), formerly Standard Marine Navigational Vocabulary (IMO, 1985), are a collection of phrases conceived by the International Maritime Organization as a restricted language in the specialized setting of maritime communications to enhance maritime safety by avoiding language misunderstandings at sea. These standardized phrases were created for use over VHF radio in bridge external communications and land-based station exchanges or face to face on board communications. Its prominent characteristics include the lexical and grammatical restrictions (controlled vocabulary, modality limitations) and the controlledurgency and safety procedural patterns, broadcast entries) they are submitted to. The standardized language restrictions in the vocabulary and phrase structure selection were designed taking into account predictable areas of language confusion and error. Availability of training material suited to the peculiar pedagogy that a controlled language involves is of utmost importance. In this context, several studies have been made (Strevens & Johnson, 1983; Losey, 2000; Pritchard & Kalogjera, 2000; Cole, Pritchard & Trenkner, 2007; CAPTAINS, 2012). However, there is still a lack of computerised corpus-based material oriented towards the specific gradual practice of the SMCP through its phraseology in semantic scenarios that may train and prepare the learner (or user) for building up predictable questions-to-answers about situations that may emerge at sea. Within the NLP framework, the present corpus-based study attempts to develop an application for automatic generation of SMCP message phrase replies based on the Question Answering (QG) system for restricted domains (RDQA) (Mollá, Vicedo, 2007). Findings revealed during this preliminary empirical study shall be presented and examined. Finally, it is expected that this approach, which to this author´s knowledge has not been explored so far, may contribute to SMCP training by e-learning, blended, face-to-face courses or Intelligent Tutoring Systems (ITS).