Machine translation Was RE: TRANSLATIONS

Subject: Machine translation Was RE: TRANSLATIONS
From: Bill Burns <bburns -at- scriptorium -dot- com>
To: "TECHWR-L" <techwr-l -at- lists -dot- raycomm -dot- com>
Date: Thu, 25 Jan 2001 08:35:00 -0700

I would like to toss in a few words about machine translation, since that was apparently what the original post has in mind.

I agree with what most of the responses have said concerning off-the-shelf machine translation software. Anyone who has played with these features on the web (like Babel Fish) can verify that they're good for "gisting" only (that is, determining the gist of a web page or document). The assumption I'm hearing, though, is that those off-the-shelf packages are the state of machine translation today. That's not wholly the case. It's also not wholly the case that you can't have reliable translation sentences. (Note that I didn't say "error free.")

Machine translation is used in some industries where the technical language changes slowly, the language sets are limited, and the complexity level of the material is relatively low. Machine translation is typically combined with a strenuously enforced program of controlled language. That is, you don't get to write however you please. You use a constrained vocabulary and formulaic sentence structures. As I understand, writers of controlled English use tools that parse as the writers compose, so noncompliant syntax or vocabulary is flagged as such.

Coupled with the use of controlled-language strategy is an extensive glossary and a system rigid linguistic rules for handling the translation. Essentially, terms have a limited range of possible interpretations based on their context. Just as the syntactical options for the source language are constrained, so are the target segments. The machine translation process usually (if not always) involves an editing pass.

Mind you, I'm not advocating machine translation. It's impractical for a number of reasons:
- An industrial-strength machine-translation system is expensive.
- It takes a lot of planning and buy-in from whatever company adopts it.
- It's not suitable for all language sets.
- The rapid change in language for some industries might make controlled language difficult to implement.
- For consumer software, the controlled-language approach might not produce quality documentation.

I found a white paper of sorts on this subject at http://clwww.essex.ac.uk/~doug/book/book.html. A search on "machine translation" produces a mixture of hits from companies producing gisting packages to full-blown systems.


Bill Burns, Senior Technical Consultant, Scriptorium Publishing
FrameMaker ACE, WebWorks Publisher Certified Trainer
bburns -at- scriptorium -dot- com - 208-484-4459
http://www.scriptorium.com


^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Develop HTML-Based Help with Macromedia Dreamweaver 4 ($100 STC Discount)
**WEST COAST LOCATIONS** San Jose (Mar 1-2), San Francisco (Apr 16-17)
http://www.weisner.com/training/dreamweaver_help.htm or 800-646-9989.

Sponsored by DigiPub Solutions Corp, producers of PDF 2001 Conference East, June 4-5, Baltimore/Washington D.C. area.
http://www.pdfconference.com or toll-free 877/278-2131.
---
You are currently subscribed to techwr-l as: archive -at- raycomm -dot- com
To unsubscribe send a blank email to leave-techwr-l-obscured -at- lists -dot- raycomm -dot- com
Send administrative questions to ejray -at- raycomm -dot- com -dot- Visit http://www.raycomm.com/techwhirl/ for more resources and info.


Previous by Author: RE: Framemaker, Word and Robohelp
Next by Author: Re: WebWorks question: WebWorks Help vs. JavaHelp
Previous by Thread: RE: Wax paste-up (Was OT: (almost) Blank Pages)
Next by Thread: RE: Machine translation Was RE: TRANSLATIONS


What this post helpful? Share it with friends and colleagues:

Sponsored Ads


Sponsored Ads