When faced with the need to help a multi-lingual community interact better, not intimidate one region or another, and generally facilitate interaction, language can be a huge barrier.
I’ve recently started to investigate what could be done when faced with this challenge as this is a very real problem for me and our http://www.SolutionExchange.info community platform. Aggregation of user driven content can be a great thing but common publication processes like editing and translation are bypassed. The availability of tools to aid an individual to publish his or her thoughts and opinions is of course a good thing in the most part as it allows for people to interact more quickly and easily removing the barriers that actually once prevented any kind of sharing or interaction (e.g. you were never able to publically comment on a newspaper article or spread a story without significant effort and cost).
With a wide and varied community, I investigated the use of the Google translate API accessible via the Google AJAX language API to start a trial to see how this automated process can help our users gain some context about content that may not be written in their mother tongue. What is particularly useful, is that the API can detect the source language automatically, which is great when you have many languages within many sources.
The trial starts on the 6th August 2010 and I would like to run it over the course of a month to see whether this prototype evolves into something valuable for some of our users. The feature can be seen in the footer of the site http://www.solutionexchange.info and must be invoked manually as no choices are currently remembered. By design, this was a pro-active choice as I was keen to ensure users pro-actively decided to try out the feature and not become confused by auto-translated content that they had not expected. Auto-translated content then shows up appended with a green asterix to indicate that the related text has gone through automatic translation. Currently, Tweets, Solution Descriptions, and Community Feed items are just some of the sections under trial but this can be easily extended or refined depending on feedback.
I’d like to further and improve this trial so I’d happily take feedback here or through the feedback form on the site at http://www.solutionexchange.info/feedback.htm.
If you have any questions then feel free to pop them in a comment below.
@Danny: Currently, Tweets, Solution Descriptions, and Community Feed items are just some of the sections under trial…
Machine translation can give satisfactory translations (http://www.international-english.co.uk/mt-evaluation.html). However, to minimise the bad translations, the text must be optimised for machine translation. You can control your text, but you cannot control text that is written by other people.
For the best possible translations, tell your contributors how to optimise text for machine translation. For some basic rules, see Muegge’s ‘Rules for Machine Translation’ (http://www.muegge.cc/controlled-language.htm).
Thanks Mike. In this particular scenario, we are wanting to help our broader community get a flavour of what else is going on. There are certain bits of content where we can indeed optimise as suggested and other areas less so.
As this feature is only in trial at the moment, I’d like to simply raise awareness of it and then slowly and only maybe influence the user base for such optimisation.
As with anything you ask of anyone, we have to be mindful of the incentives and often Tweets for example are just shared without a lot of thought or optimisation for where they may be consumed.
Many thanks for adding to the conversation and for the useful references. I’ll take a read and see how this feature could be improved.