Content area
Full Text
Abstract
A multimodular Cognitive Translation Model (CTM) offers algorithms of recognition, identification and equivalency in a model of translation replicating human cognitive-linguistic competency. Translation adequacy is achieved by triangulating unique alpha-numeric addresses of CMT elements.
This protocol (1) relates a word-form to a referent; (2) identifies one referent's POV spectrum -. an inventory of all Points of View of any aspect of the referent. (3) identifies members of the class of referents created by a POV; ((4) assigns each POV a unique alpha-numeric address. (5) connects POV to its four proprietary sets of Musketeer classes of operants serving all referents grouped by the POV. (6) identifies that each referent word-form pair claims unique coordinates of (a) all POV addresses in its spectrum and (b) all addresses of operants bundled with each POV. (7)Textual equivalency between coda is established and verified by finding referents with identical coordinates.
Key-words:Machine Translation,Cognitive Translation, Equivalency, Translation Model, Artificial Intelligence, Cognitive modeling;
1. Universal Platform of Shared Reality
The paradox of machine translation (MT) is this - to achieve competent results in processing fruits of cognition by machine, we must replicate the tree of human thinking. However, the heuristic-cognitive nature of translation is not yet a friend of MT developers. Our logical-statistical means are not adequate to formalize and integrate mental processes into an MT paradigm. To resolve this paradox we need metalanguage tools.
Back in the dawn of MT era, I had a dream of a 'dictionary' that, "given the rules of use, anyone can generate semantically, lexically and grammatically correct sentences in another language'. (Gorbis, 1976). This is one possible interpretation of this vision..
The assumption of most MT efforts is that correlating means of expressing our thoughts in one coda1 as strings of words or signs in another is a lexical-grammatical task. This assumption is incorrect. Just because the end product of translation is a text; identifying correlations and establishing equivalency are not linguistic procedures.
The basis for translating 'texts' of any length is competence in universally shared reality previously defined as "a sum total of all concepts and their instrumental and qualitative characteristics shared (potentially) by all humans regardless of when and where they happen to communicate." (Gorbis 2006 (a)). Possession of databases of 'word-forms',...