The dynamic mainstays of the digital catalog—sorting, searching, filtering, and otherwise using descriptive metadata to find things—these are what have always separated the digital from its analog counterpart. The end-user gets exactly what they’re looking for from these mechanisms—results of univalent significance.
On the other hand, keywords change the game. Keywords are metacontent (content about content) that can provide multiple viewpoints from which one may derive larger, more complex meanings. So, to the extent that good keywords lead to interpretation, the digital catalog begins to exceed the limits of the analog rather than to exist as a mere enhancement.
Choosing keywords is like playing a game of concentration, except all the cards are exposed. There are no hard and fast rules; the only thing to ask yourself before you begin is, “What could the end-user possibly find here?” Other than that, your choices are entirely determined by the nature of your subject.
The best way for me to demonstrate the concept of keywording is to use an artist many of us can easily imagine, so I’ll use Mary Cassatt. The obvious place to start is with the objects that populate her pictures, from armchairs to windows and everything in between: curtain, flowers, fruit, mirror, newspaper, opera glasses, toy, tree, vase, wall covering, washstand, water pitcher, etc.
Less easy are people including, but not limited to baby, boy, brother, daughter, family, father, girl, man, mother, nurse, woman, etc. I say, “less easy,” because a grandmother is always a mother, and a son is sometimes a brother. However, knowing who the sitters are and the contexts in which they appear will help sort out this difficulty.
As the list begins to grow, keywords other than objects naturally suggest themselves and categories begin to emerge. Now we can begin collecting keywords under rubrics besides objects, such as activities, people, the body, clothing, accessories, locales, colors, and animals.
You may have noticed that with activities we have embraced verbs along with the nouns we were using. Besides the more general ones (bathing, boating, picking fruit, combing, playing, reading, riding), there are other verbs (gerunds to be exact) that are Cassatt-specific, such as admiring, caressing, clasping, kissing, reaching, embracing, resting, sitting, even smiling and sucking finger(s). If we add looking we can also append forward, left, right, and up to it and make gerund phrases.
Interestingly, once we are able to choose from lists of subjects (people) and find them doing things (activities) a kind of descriptive grammar begins to reveal more about the pictures:
a grandmother is reading to her grandchildren,
a toreador is smoking a cigarette,
a woman / looking / up is picking / fruit from a tree.
To the colors—pink, blue, green, I add plaid and stripes. Now we have adjectives to tell us even more about our subjects:
a woman in black is sitting on a striped / sofa,
a nude / child is playing with a toy,
a girl in a garden has a white / dress and a hat with a pink / bow.
It’s one thing to compile lists of things we expect users will be looking for. It’s quite another to reveal something (possibly unexpected) about the work itself. Certainly the keywording process is different with every painter and is radically different when dealing with other kinds of cultural objects. But, if thoughtfully considered, the application of content about content will always yield more profitable results than a single category or the simple enumeration of physical parts.
Roger Shepherd is the Creative Director of panOpticon.
Have questions you want to ask or topics you want to recommend for discussion? Please go to our question page to post them. Thanks.