Project Blog Archive

Back to list of blog posts

Panning for metaphors

Posted by Wendy Anderson on the 27th of September 2013

We’re currently up to our necks in data, to borrow a well-used metaphor. The most time-consuming part of our project involves working systematically through the automatically generated overlap data (see ‘Metaphorical links in the data’ blog post, below), teasing out and coding the metaphorical links, and discarding links which are due to other forms of polysemy, and homonymy. To give you a sense of scale, this means working through around 800 batches of data, each of which involves between around 500 and 50,000 lexical items.

This gets under your skin and worms its way into your mind. Last week, I found myself pondering the coding process while cleaning a rug at home. The rug is a large rectangle, made up of 54 small squares. While working out my strategy for cleaning it, I realised I was conceptualising this novel task in stages borrowed from the now very familiar data coding process: (1) tackle it one square at a time (=tackle our data one category at a time); (2) approach each square in stages – prepare surface, spray on cleaning foam, work it in, clean it off (=approach the data in stages – prepare files and reference books, do a first pass of the data, spend time on the complex links, run a quick check); (3) finally, hoover the complete rug (=run final checks on the coded data). The most significant difference has been in time – I finished the rug in four evenings, but the coding keeps going... This is an idiosyncratic metaphorical link, certainly not one which is likely to become conventionalised, but it made me think because I realised I was using our coding process as a source domain for something else.

A few days later, I was speaking to Ross on the coding team, and he made a comment that chimed rather nicely with my one-off rug/coding metaphor. He described the coding process as like panning for gold, i.e. what we’re doing is sifting gradually through pans of silt (=raw data) to find the nuggets (=metaphorical connections between categories). A much more vivid metaphor than mine. I then thought I’d ask around the team, to find out if anyone else has metaphors involving the coding process, as either source or target domain.

I got a quick reply from Heather, who told me that she thinks of coding as breaking a cipher or decoding a message: “you start with a jumble of words and data and your picture of the whole category becomes clearer and clearer as you go through the data and find out what parts of the data have meaning.” A nice one, and interestingly a largely abstract source category (breaking a cipher) mapped onto a similarly abstract target (data coding) – and one action (coding) seen in terms of its opposite (decoding).

For Jude too, the task of coding has been lodging itself in her brain. Jude lives in the country and has been splitting and storing logs in preparation for the winter. This very physical task has some connection for her with the mental task of identifying metaphors in data. The link for her is again structural – sizing up whether the logs are most suited to the wood stove or for the central heating boiler maps onto the process of deciding whether the connections in the data are metaphorical, literal, or ‘noise’. She suggests that perhaps the ‘metaphors’ in this scenario are the big juicy grubs that have set up home inside the logs, and which end up as treats for her hens. She concludes that “coding really does get into your head and I'm thinking about it a lot, even when I'm not in front of the computer screen.”

Christian had a few – she’s clearly been spending too much time coding! For her, data coding is like eating spaghetti: “you keep trying to twirl it round the fork in an orderly way, hoping that some of the sauce and the nice bits will stick.” This is linked to her second one, of “unfurling a ball of strands of wool of different kinds, all snaking (!) off in different directions”. Both are to do with making order out of disorder. Finally, and possibly because she’s been working recently on our Minerals category, “the image of chipping away with a hammer at a rock face” – creating something out of raw materials.

For Marc, the structuring metaphor is one of clarifying, like tuning a radio or TV set - "there's a whole mass of stuff and we want to tune in to only one set of them, the metaphors running through the dataset". The metaphors are hidden among "noise", like static on a radio, and "the coding is the rather tedious task of fiddling with the knob for each dataset" to identify sense among the noise.

Ellen similarly has some vivid metaphors for her coding, and commented that she had always thought of it as digging. I’ll quote this in full because it sums up so nicely our aims for the project: “I had been thinking about whether it was like beach-combing: searching through sand to find pretty shells, foreign packets, fishing nets and other interesting bits and pieces. That describes the variety of metaphors quite well I think, but not the sheer effort of the process! So I think it has to be proper, exhausting digging down into the earth in the hope of finding something worth the effort. An archaeological dig might be an appropriate way of conceptualising it: days and months of digging down through the earth and sifting through strata to find artefacts which might tell us something important about a culture over time...”