A Semantic and Syntactic JourneyThrough the Dylan Corpus

Abstract

A Semantic and Syntactic Journey Through the Dylan Corpus* Jean-Charles Khalifa (bio) The original title of this paper, as presented at the Caen colloquium, was "Dylanesque Syntax," which was meant as a joke of sorts, the idea being to scare the audience into submission before the first line was even spoken. Of course, there is no such thing as "Dylanesque syntax," in the sense that we are by no means dealing with a language that is separate from (American) English and has a distinct, separate syntax (even though it may be argued to have a distinct, all but separate phonology). What I intend to do is simply apply some of the tools of linguistics, especially corpus linguistics, to the songs recorded by Dylan over the past 44 years, and, in line with the general theme of the colloquium, see if the findings can teach us something about his artistry. I will then leave it to true Dylan scholars to interpret some of the data I am presenting, in the context of their own approaches to Dylan's writing and poetry. I would like to add, in the way of an introduction, that it is indeed the lyrics I will be concerned with, not the music, even though I am aware that the question of music as syntax is of paramount importance to some theorists. The corpus compiled simply brings together as many of the Dylan lyrics as I managed to compile. As always with the "Bard of Hibbing," it is impossible even to approach exhaustiveness: there are 401 songs in total, which doesn't quite cover the whole body of the officially recorded songs. Still, for my purposes in this study, this selection will be taken as representative enough, if not close enough to completeness. I simply copied all the songs back to back into one single file (which proved to be something of a headache), saved it in ".txt" format, and ran it through a concordance program1 to see what would happen. [End Page 162] The concordance program displayed the basic figures about the file: 111,555 words, and a 8,170-word vocabulary. There are of course a number of ways of viewing these raw figures; the only significant figure is the vocabulary count, and on that criterion alone—compared to Shakespeare's alleged 25,000 to 30,000-word vocabulary—Dylan might indeed appear as a poor writer. On the other hand, random comparisons with other classic writers yield results that make him compare very favorably (James Joyce's Dubliners: 67,000 words, with a 7,600 word-vocabulary, Defoe's Robinson Crusoe: 121,000 words and 6,500 word-vocabulary provide comparable examples). Another significant figure in textual analysis is the type-token ratio (111,555 / 8,170 = 13.65 in the case of the Dylan corpus), which measures lexical density and richness. It should be pointed out, however, that such a ratio will prove different for specific literary genres. Indeed, we are dealing with song, which makes extensive use of repetition in verses and choruses. Very little comparative use can then be made of that ratio, until we further investigate other bodies of song. I found it a lot more interesting, as it were, to take a quick look at word frequencies and ranks; but here again, the raw data had to be pre-evaluated since, and not surprisingly, the program reported back to me that the most frequent word in the corpus was "the"! However, once the words were sorted out by categories, and certain grammatical items (i.e., determiners, conjunctions, pronouns, and the auxiliaries have, be, and do) set aside, a much more interesting picture emerged. Here is a list of the top 10 lexical nouns in the corpus: man, time, love, baby, night, day, mind, eye, lord, and heart. The list itself, however, would hardly be worth commenting on if we didn't have a reference corpus for assessing it. To this end, I have chosen the British National Corpus (B.N.C.) for reasons of simplicity (it is easy to use online, and there are interfaces available where frequencies within word classes are readily obtainable).2 Of course, it...