The rise of Wordle

3 June 2010 at 9:02 am 8 comments

We’ve been having a debate in the office about the merits of Wordle. These are Russ’ thoughts.

Russ Wilson writes:

Wordle, word clouds, Tagxedo:  all online apps for taking a load of data in the form of words and presenting it in a design friendly way.  As a lover of language I’m all for anything that encourages people to explore words, think about how and why they’ve been used and analyse their meanings. However I’m not really sure that any of these tools do this.

I have two main issues with Wordles, and they’re exemplified in the wordle above, based on David Cameron’s coalition speech. First, they remove the word from its immediate context.  Take the word interest, represented as one of the more frequently occurring words.  But it could equally indicate curiosity and engagement or interest payments. The Wordle doesn’t help; it only tells us the word occurred often in the speech. Similarly, coalition also figures prominently. But it doesn’t help with context. We can’t tell, for  example, whether they said ‘this is a coalition’ or ‘this is not a coalition’!

The second issue is that frequency is being proposed as an indicator of importance, but that’s not how we actually interpret speech. Imagine a Wordle which captures responses to a question such as ‘What do you think of the coalition?’ One person might say the new government is ‘absolutely the most important and exciting change in politics in living memory’; others might respond that it is ‘quite troubling’, ‘not very troubling’ or even ‘not troubling’?  The Wordle would look, well, troubling:

Frequency of use is simply that – frequency of use.

Wordles do look good. But they become dangerous when presented as meaningful analysis. They don’t tell the right story, and worse, they are also capable of telling a completely different story altogether. Yet the mainstream media are happy to present them as semi-serious analysis: The Guardian says that from its Wordles for Nick Clegg and David Cameron’s acceptance speeches ‘you can get a good idea of the two leaders’ use of language – and which words were important to them’.  As a linguist I know there are many ways to explore their language use, but I don’t think I would include a Wordle as a method of analysis or of display. Their visual appeal gives them more credence than they deserve.

As a final test, here is a Wordle of this post – do you think it reflects the views I’ve expressed above?

Entry filed under: insight, language, trends. Tags: , , , .

Facing off about privacy Opening up in Latin America

8 Comments Add your own

  • 1. thenextwavefutures  |  5 June 2010 at 9:23 pm

    While I agree with the overall point of Russ’ post, there’s an interesting post over at Energy Bulletin which uses Wordle as a tool to compare the contrasting rhetoric of the oil interests, the White House, and Greenpeace, on the Dee[water Horizon oil disaster. Which seems quite a good way to use word clouds as a form of analysis.

    Reply
  • 2. wonder_j  |  8 June 2010 at 2:25 pm

    I agree with the overall point of Russ as well, and would like to add 2 things:
    the most important: STOP PUTTING VERTICAL WORDS in a word cloud! The human brain cannot assimilate it properly, ie it cannot compare the sizes of horizontal and vertical words. Same thing with mixing fonts and colors: it may look better but it adds confusion.
    also, Russ points out the issues of word clouds when compiling words of a speech, he’s right its not ideal in that situation. However, a cloud may be much more convenient when representing a series of key words, from a web search for example, such as yahoo! or google allow you to discover with their instant word trends.

    Reply
  • 3. Alex Steer  |  15 June 2010 at 11:57 am

    Nice post, Russ. Wordles are especially bad at analysing conversation (as opposed to isolated speech/writing), because conversations aren’t disconnected utterances about things we think are important. We (obviously) follow up on what our fellow speakers are saying, unless we’re being deliberately rude. The need to participate in a meaningful and polite exchange, together with our susceptibility to linguistic effects like lexical and syntactic priming, means we tend to gravitate towards the same words and phrases as the people we’re talking with. The agenda for most conversations, and hence their repertoire of words, is often set pretty early on.

    The TV debates offer a good example of the problem. All three leaders, despite their differences, were primed to talk about certain issues. Despite their differing positions, their stock of words was largely conditioned by the questions fired at them by the audience. The words used in a given utterance (especially one as calculatedly voter-friendly as a speech in a leadership debate) are in large part a function of prevailing voter opinion. This probably means that chronological frequency analysis gives more interesting insights into political discourse in Britain: e.g. has there been a notable shift in the co-occurrence of ‘government’ and ‘data’ in political speeches, journalism, analysis, etc., in recent years?

    Reply
  • 4. Russ  |  5 July 2010 at 5:36 pm

    I think Andrew’s point is interesting in that does show how language analysis can shine a light on competing rhetorics. But again I would suggest that Wordles still prove problematic for the same reasons outlined above – here a more appropriate methodology would be Critical Discourse Analysis which evolved out of the very desire to make clear the differences in competing ideologies’ viewpoints.

    Alex – interesting point about what it might show about changes in vocabulary in certain types of discourse. Again the unknown polarity of the words means that any inferences would be limited but at a push I might concede that it could show changes in issues that are being talked about within a certain context! But then used in this way they’re just a very rudimentary form of corpus analysis…

    Reply
  • 5. Edward04  |  5 December 2011 at 10:52 am

    Great post, and I agree with the first point totally – context is all in language and meaning. The second point indeed is true – including or leaving out a “not” before a word such as troubling is crucial. My own experience with Word cloud has been that it’s just something that hits the eye, they look great – and no, they shouldn’t replace proper interpretation. however, I would argue that they have a place in a presentation if properly contextualised.

    Reply
  • 6. LoveStats  |  14 August 2012 at 12:00 pm

    I look at this in the same way as research. Bad research misdirects readers. Bad word clouds misdirect readers. If you’re going to use any tool, you need to use it properly. Don’t blame the word clouds. Blame the creator of the word clouds. And the creator isn’t wordle or tagxedo, the creator is YOU.

    Reply
  • 7. J.  |  14 August 2012 at 3:19 pm

    All excellent points. I have not used word clouds in research for my clients, but I do talk about them in regards to competitive and company analysis and for those people I know looking for work in relation to job descriptions.

    Reply
  • 8. freerangeresearch  |  14 August 2012 at 6:28 pm

    EXACTLY: “Wordles do look good. But they become dangerous when presented as meaningful analysis. They don’t tell the right story, and worse, they are also capable of telling a completely different story altogether. Yet the mainstream media are happy to present them as semi-serious analysis”

    Thank you for writing about this. This is a big pet peeve of mine. The underlying assumption is that 1 word= 1 concept, and that is just not true.

    …Although sometimes I wonder if a wordle could be more meaningful with a little bit of coding… (butthen, would it still be a wordle?)

    Reply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Trackback this post  |  Subscribe to the comments via RSS Feed


The Futures Company blog

The Futures Company was created through the merger of Henley Centre HeadlightVision and Yankelovich in 2008. This is the blog of the new company - but the former posts from the former Henley Centre Headlightvision blog still can be found here.


WPP? Leaders in Advertising,Branding,Marketing

%d bloggers like this: