Wordlink With WordPress
The POOLS-T project has now reached the end of its funded period. During the course of the project the basic concept of a tool that enables the checking of any unknown words in a text in an online dictionary has been developed substantially. The original “Webpage Textblender”, invented by Kent Andersen, is now complemented by Caoimhín Ó Donnaíle’s Wordlink and Multidict tools, and the range of languages supported has been massively expanded.
To mark the completion of the project Gordon Wells has placed all 40 texts from Series One of Island Voices (20 English and 20 Gaelic) on WordPress platforms linked to Wordlink/Multidict. Although the funded period of the project is now over further refinements of the tools will continue, as more languages are incorporated and in response to comments from users. Interested readers are invited to take a look at the Wordlinked Island Voices materials. Please note that, while the default display is “Split Screen”, there are alternative options, such as “New Tab” and “Pop-up”. A choice of dictionaries is also available, depending on the language in question.
Of course, dictionaries are by no means the answer to every language learning problem – and some dictionaries are better than others. But the facility to present online text with this added dictionary access function may be attractive to both teachers and learners. Comments are welcome here, and will be relayed to the developers.
Leave a comment
Island Voices Videos
Island Voices - Guthan nan Eilean 
Unfortunately too many dictionaries still go with a paper-style “headword” search — the tool is of very limited use if it’s linked to a dictionary that only deals with headwords. The link to the Internet Archive is pretty cool though — I imagine Kevin’s just created a list of first words by page and just directs to the appropriate page that way. It’s actually more handy than a direct tool, as you’re quite likely to end up on the same page as the headword if you click the inflected form (except for those nasty initial mutations).
I see the tool as it stands as something that is most useful for reading random pages. For use in specific pages aimed at learners, it would be handy if the page author was to use some corpus-like mark-up as a hint for the dictionary, so if I click on a verb, it only returns verbs, and the like.
Niall’s comments are all spot-on. Yes, for languages with a significant amount of inflection (i.e. most languages, with the notable exception of English), the usefulness of the tool is much reduced if the online dictionary only accepts dictionary headwords, not wordforms. A notable dictionary which does accept inflected forms is the wonderful Vertalen dictionary, which does 9×9 European language pairs but not Gaelic unfortunately.
It would be nice to extend the project and get Multidict to somehow do some lemmatisation, either rule-based or table-based, when the dictionary doesn’t. At the moment, the only thing it does is to remove the ‘h’ from the second place of Scottish Gaelic or Irish Gaelic words.
Yes, the Web Archive “page image” dictionaries enable you to bypass the lemmatisation problem to some extent. And yes, the lookup works in exactly the way you guessed. For each dictionary, I created an index of the first word on each page. Volunteers welcome to do other dictionaries! (At the moment it is a bit of a nuisance having the useless search pane taking up room on the right of the Web Archive dictionary page, and having to zoom in each time, but I have a bug report in on this for the Gnu Bookreader and I have hopes that an improvement will shortly be on hand.)
The trick of indexing the first word on each page works not just with the Web Archive dictionaries, but with other page-image dictionaries too. The most useful Gaelic dictionary for use with Wordlink, at least for learners of a particular range of ability, is probably the one labelled “Briathrachan Beag (images)”.
The idea of Wordlink (and of Kent Andersen’s standalone tool, the Textblender) is to avoid authors having to do any special markup. So no plans along those lines. It would be possible to imagine a cleverer version of Wordlink which would do some local grammatical parsing and give a report, with appropriate links to dictionaries, but that is getting a bit advanced for now.