Many years ago, as a graduate student in comparative literature at the University of Wisconsin-Madison with a focus in part on the linguistic model in literary criticism, I turned my attention to beyond-the-sentence topicality. Scholars have parsed the sentence since ancient time, but they have paid less attention to the way sentences connect to each other.
One of the applications of this line of research is for machine translation. How does the translation engine determine, for example, whether the word lead in a text refers to the heavy metal or to the concept of leadership?
One way to try to answer that question is to look for patterns in the denotative and connotative qualities of the lexicon of the passage. Looking for themes that interlace the selection of words might help to determine the sense in which they are being used and also help to highlight the passage’s main concerns.
Consider, for example, the opening of Flaubert’s Salammbo. (I picked this text up from a website that did not identify the translator but simply gave the credit “etext prepared by John Bickers”)
It was at Megara, a suburb of Carthage, in the gardens of Hamilcar. The soldiers whom he had commanded in Sicily were having a great feast to celebrate the anniversary of the battle of Eryx, and as the master was away, and they were numerous, they ate and drank with perfect freedom.
The captains, who wore bronze cothurni, had placed themselves in the central path, beneath a gold-fringed purple awning, which reached from the wall of the stables to the first terrace of the palace; the common soldiers were scattered beneath the trees, where numerous flat-roofed buildings might be seen, wine-presses, cellars, storehouses, bakeries, and arsenals, with a court for elephants, dens for wild beasts, and a prison for slaves.
Fig-trees surrounded the kitchens; a wood of sycamores stretched away to meet masses of verdure, where the pomegranate shone amid the white tufts of the cotton-plant; vines, grape-laden, grew up into the branches of the pines; a field of roses bloomed beneath the plane- trees; here and there lilies rocked upon the turf; the paths were strewn with black sand mingled with powdered coral, and in the centre the avenue of cypress formed, as it were, a double colonnade of green obelisks from one extremity to the other.
Far in the background stood the palace, built of yellow mottled Numidian marble, broad courses supporting its four terraced stories. With its large, straight, ebony staircase, bearing the prow of a vanquished galley at the corners of every step, its red doors quartered with black crosses, its brass gratings protecting it from scorpions below, and its trellises of gilded rods closing the apertures above, it seemed to the soldiers in its haughty opulence as solemn and impenetrable as the face of Hamilcar.
Looking at this passage we can easily pick out certain themes and observe how the author’s vocabulary reinforces them. (This is just a quick demonstration and not an attempt at a fully worked-out critical analysis.) Clearly Flaubert meant to impress us with the magnitude, exoticism, and lushness of his setting and with the concepts of military authority and its obverse of excess and lack of restraint. He works hard to create an intensified vividness of setting. Notice how his words align to emphasize certain concepts.
EXPANSE / VASTNESS | VIVIDNESS/ HYPER-SATURATION | LUSHNESS OF NATURAL SETTING | OPULENCE | WARFARE | RESTRAINT / LACK OF RESTRAINT |
great perfect reached numerous stretched masses field strewn double extremity far broad large every |
bronze gold-fringed purple pomegranate shone white black coral green yellow mottled ebony red brass gilded |
gardens terrace trees fig-trees wood sycamores verdure vines branches pines roses bloomed plane-trees lilies cypresses trellises terraced |
gold-fringed purple palace shone ebony opulence strewn marble court |
soldiers commanded battle captains cothurni surrounded stables arsenals prison slaves vanquished prow protecting galley |
master freedom prison slaves extremity vanquished apertures wild beasts galley gratings |
I limited myself to six columns for fit in my text window. Of course there are many words relating to feasting and celebration. There is also an element of danger and impending violence associated with the martial aspect and references to scorpions and the like that I could have picked out, as well as a sort of architectural exoticism, but these six topical groupings fairly well underscore the main themes of the passage. In fact, they predict pretty well the overall themes of the entire book.
Now fast forward to the present. I this that search engines use a similar analysis to determine the topicality of web pages. Some such analysis could be done on the page itself. Web search specialists use the term keywords to describe the words they hope to rank for in search engine results pages. But notice that this sort of analysis can help to determine not just the keywords but the context in which they are being used. A similar analysis could be done to the anchor texts and surrounding words in links leading to the web page.
Poets and other literary types have always been sensitive to these sorts of word associations. Even in nonliterary prose, the careful web author would be wise to do likewise.
.
Kate Clifford
Well, I’ve seen google do something like that, but yahoo always screws up the order of the search terms, so I’m thinking it doesn’t care about topicality.