Browse Prior Art Database

Regular Language Transcription Machine

IP.com Disclosure Number: IPCOM000049503D
Original Publication Date: 1982-Jun-01
Included in the Prior Art Database: 2005-Feb-09
Document File: 4 page(s) / 19K

Publishing Venue

IBM

Related People

Roth, JP: AUTHOR

Abstract

Among the many difficulties encountered in attempting to translate language by machine are the following: 1. Multiple meanings for any given word. 2. Irregular and grammatical structures in the languages to be translated, irregular conjugations, declensions, irregular word plurals, etc. 3. Complex grammatical structures.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 31% of the total text.

Page 1 of 4

Regular Language Transcription Machine

Among the many difficulties encountered in attempting to translate language by machine are the following:
1. Multiple meanings for any given word.
2. Irregular and grammatical structures in the languages

to be translated, irregular conjugations, declensions,

irregular word plurals, etc.
3. Complex grammatical structures.

From experience gained in connection with an attempt to construct a DESIGN MACHINE which transforms an architectural description of design, for example, of a computer into a detailed design in terms of the hardware prescribed by the desired technology, it was discovered that the method of design must be in some regular form for it to be possible to design a machine which was deterministic and which could, if actually constructed, be deterministically tested for the existence of failures. For this purpose a Regular notation, for short, R-notation, was defined to prescribe architectural descriptions of machines at a high level. Likewise, the "target" machine in a given technology was also prescribed to have a regular form, a Regular Logic Design and a Regular Hardware Design.

With these restrictions it, in fact, became possible to achieve a desired result, including many side dividends.

This suggests an analogy to the problem of translating from language A to language B. Firstly, the languages involved shall be regularized as well as the correspondences between the words, phrases, idioms, etc. Secondly, different meanings and different grammatical forms of words are distinguished by appropriate subscripts, removing ambiguity as well as proliferation of different grammatical forms of a word, eliminating these problems with irregular verbs and nouns. Thirdly, the grammatical structure is regularized as well as forbidding certain complicated grammatical structures. It is then agreeable that the problem of transcription of "regular languages", to build a machine to perform the same, is in fact feasible. It could have direct application, for example, in the transcription of manuals, instruction guides and other factual straightforward text from one Regular Language to another. The Regular Language Transcription (RLT) Machine would have storage dependent primarily upon the number of words in each language as well as the number of languages between which transcription is desired. The mechanism to accomplish the transcription is in itself quite simple. One such process is outlined.

How easy would it be to write in a Regular Language and how easy would it be to understand a Regular Language Transcription? After a little practice the writer might find it easier to write in the regular version of one's native language than to compose a telegram in the same. It is certainly clear that a student knowing only language A could learn to transcribe text Ra in the corresponding Regular Language RA. Thus, the student is acting as a preprocessor. Thence, the RLT Machine could transcribe Ra into te...