On Vocabulary & Language

Controlled vocabulary - sounds like a great idea, esp. for computer systems (which don't handle ambiguity well). However the first decade or so of assessing website usability shows that differences in language are perhaps the single most difficult thing to get right when doing website design. I say "interlibrary loan", you say "document delivery" and administration or PR wants to call it "WhizBangPlus" (the older WhizBang Classic being something of a disappointment in initial user testing). Coming from a health sciences background I was used to singing the praises of MeSH and figured that all disciplines had such a tool, so why wouldn't we use it? We could even do some clever search engine tricks like synonym rings so if a user searched for "cancer" they got "neoplasm" too.

When first getting to meet other website designers I was horrified to learn that they had a best practice of creating a controlled vocabulary (if they were really good at their craft/trade) for each and every website! A virtual Tower of Babel - horrors! Then as we talked more it quickly became apparent - many other disciplines don't have the luxury of a MeSH. And even if we were talking health sciences there were many terms having to do with site navigation or the particular community which uses a particular website (i.e. a school, professional organization, patient support group, etc) that each site really might warrant its own limited vocabulary - after all if the users of your site call things something doesn't it make sense to also call it that thing? Do we really think we can "train them" to use "the right language"? I heard a story about a developer somewhere large (Intel?) who said the reason people couldn't find anything using their search engine was because they were "typing in the wrong words". Guess again, fair developer! We need to understand how they think and what language they use, nice as it would be for that to work both ways. Much like the OPAC being an excellent back end system for storing information about books, journals, etc a controlled vocabulary like MeSH is an excellent system for cataloging or classifying information in general - its just not good for everyday conversation, and most likely not the way people will be thinking when looking for things on a website.

Peter Morville, in his Ambient Findability book, talks about the rise of tagging at sites like Flickr and how the practice has caught on like wildfire yet hasn't exactly started to transform the way people find things other than at a pretty casual level. Its nice to be able to look at all the "German Shepherd Dogs" or "Bush Coronation" pictures but such an informal and idiosyncratic system is unlikely to replace a powerful set of descriptors like in MeSH anytime soon. But people can *use* tags without 3 days of MEDLARS training or special software so they do. Peter points out that there's a convergence likely to happen in the future, where unstructured vocabulary (like tagging or search engines that only look at keywords without any weighting or structuring like inside HTML tags) and structured vocabulary (like MeSH) learn to co-exist and even feed each other. Tags are potentially an excellent source of research for how people talk about things and the more common tags could be added to search engine synonym rings. They could also speed up the development of controlled vocabularies by fast tracking new terms which aren't just trendy but are actually useful and widely used to get them into our systems faster.

Robert McNeill's excellent documentary The Story of English (sadly out of print, hopefully coming soon to DVD?) is a great starting point for showing how dynamic language is, and how attempts to lock it down too far (i.e. The Queen's English) are doomed to fail. Recent years have seen normally staid publications like the Oxford English Dictionary starting to add words like "blog" and "bling" at a much faster pace than they used to. Seems like its time for libraries to loosen up and speed things up a little and start using some of this unstructured vocabulary to our advantage, to help our patrons use our websites better and, ultimately, find what they need faster and more painlessly. George Lakoff (of "Fire, Women & Dangerous Things" fame) reminds us that language and categorization are at the heart of how we make sense of the world as humans. If language itself is in a period of great churn we'd be wise to figure out how to use that to our advantage. Perhaps we librarians can come up with a way of allowing the churn to be our little petri dish in a laboratory and out of that come up with great easy to use systems that modern users can relate to. What we probably will never succeed at is "training" them to use our version of a controlled vocabulary as a way to find information quickly. Our controlled vocabularies are a great back end tool for configuring search engines and other software to quickly and efficiently retrieve information - but our users need to be able to talk to these systems or they may as well not exist at all.