Stylometry, Chronology and the Styles of Henry James

Contemporary stylistic and stylometric studies usually focus on an author with a distinctive style and often characterize that style by comparing the author’s texts to those of other authors. When an author’s works display diverse styles, however, the style of one text rather than the style of the author becomes
the appropriate focus. Because authorship attribution techniques are founded upon the premise that some
elements of authorial style are so routinized and habitual
as to be outside the author’s control, extreme style
variation within the works of a single author seems
to threaten the validity of the entire enterprise. This
apparent contradiction is only apparent, however, for
the tasks are quite different. Successful attribution of a diverse group of texts to their authors requires only that each author’s texts be more similar to each other than they are to texts by other authors, or, perhaps more accurately, that they be less different from each other than from the other texts. The successful separation of texts or sections
of texts with distinctive styles from the rest of the
works of an author takes for granted a pool of authorial similarities and isolates whatever differences remain.
Recent work has shown that the same techniques that are able to attribute texts correctly to their authors even when some of the authors’ styles are quite diverse do a good job of distinguishing an unusual passage within a novel from the rest of the text (Hoover, 2003). Other quite subtle questions have also been approached using authorship attribution techniques. Nearly 20 years ago, Burrows showed that Jane Austen’s characters can be distinguished by the frequencies of very frequent words in their dialogue (1987). More recent studies have used
authorship techniques to investigate the sub-genres
and varied narrative styles within Joyce’s Ulysses
(McKenna and Antonia, 2001), the styles of Charles Brockden Brown’s narrators (Stewart, 2003), a parody
of Richardson’s Pamela (Burrows, 2005), and two
translations of a Polish trilogy made a hundred years apart (Rybicki, 2005). Hugh Craig has investigated chronological changes in Ben Jonson’s style (1999a, 1999b), and Burrows has discussed chronological
changes in the novel genre (1992a).
I am using authorship attribution techniques to study the often-remarked differences between Henry James’s early and late styles.1 I begin by analyzing a corpus of 46
American novels of the late 19th and early 20th century (12 by James and 34 by eight other authors) to determine the extent to which multivariate authorship attribution techniques based on frequent words, such as principal components analysis, cluster analysis, Burrows’s Delta, and my own Delta Primes, successfully attribute James’s early and late novels to him and distinguish them from novels by eight of his contemporaries.2 Because all of these techniques are very effective in this task, all are appropriate for further investigation of the variation within James’s style, but DeltaLz produces especially
accurate results, correctly attributing all 40 novels by members of the primary set in eleven analyses based on the 2000-4000 most frequent words. All of the results also reconfirm recent findings that large numbers of frequent words are more effective than the 50-100 that have been traditionally used, and that the most accurate results (for novel-sized texts) often occur with word lists of more than 1000 words (see Hoover, 2004a, 2004b). The PCA analysis in Fig. 1, based on the 1001-1990 most frequent words, clusters the novels quite well–better, in fact, than analyses that include the 1000 most frequent words.
When cluster analysis, PCA, Delta, and Delta Prime techniques are applied to nineteen novels by Henry
James, they show that the early (1971-1881) and late styles (1897-1904) are very distinct indeed, and
that an “intermediate” style (1886-1890) can also be
distinguished. DeltaLz again produces especially accurate
results, correctly identifying all early, intermediate, and late novels in 24 analyses based on the 200-4000 most frequent words. These results paint a remarkable picture of an author whose style was constantly and consistently
developing, a picture that is congruent with James’s
reputation as a meticulous craftsman who self-consciously transformed his style over his long career. A comparison
with Charles Dickens and Willa Cather shows that
Dickens’s early and late novels tend to separate, but
do not fall into such neat groups as James’s do, and that Cather’s novels form consistent groupings that are not chronological. These authors seem not to have experienced
the kind of progressive development seen in James. It is dangerous, then, simply to assume chronological
development of authorial style.
Finally, these same techniques show that the heavily revised versions of The American (1877), Daisy Miller (1878), and The Portrait of a Lady (1881) that appear in the New York edition of James’s novels (1907-09) are consistently and dramatically closer to the style of the
later novels. Yet even his notoriously detailed and extensive
revisions do not allow PCA to group the revised early
novels with the late novels. Instead, the revised versions
fall at the border between the early and intermediate
novels in PCA graphs (see Fig. 2), and consistently join with their original versions in cluster analyses. The
results obtained using Delta show that even the errors make sense. In analyses that are not completely accurate,
The Portrait of a Lady and Washington Square, the
latest of the early novels (both 1881) are sometimes
identified as intermediate. The other errors involve the identification of The Spoils of Pointon, the first of the late novels (1897), as intermediate; no analyses incorrectly
identify an early novel as late or a late novel as early.
In the analyses that are completely correct for the 19 unambiguously early, intermediate and late novels, the New York edition versions of earlier novels are identified as follows: the revised early novels The American and
Daisy Miller are universally identified as early, the
revised intermediate novel The Reverberator is universally
labeled intermediate, and the revised early The Portrait of a Lady is usually labeled intermediate, but sometimes
early. This is an intuitively plausible result, with the
latest of the early novels pulled far enough toward the late novels to appear intermediate, but, interestingly enough, DeltaLz, which produces much more accurate results overall, labels all of the novels according to their original
publication dates in eighteen analyses based on the
200-2800 mfw. In the remaining six analyses, the
early Daisy Miller is identified as late, and in one
analysis (based on the 4000mfw) The Portrait of a Lady is
identified as intermediate. Further investigation of the
implications of these results is ongoing.
Authorship attribution techniques thus confirm the
traditional distinction between early and late James,
establish the existence of an intermediate style, and lay the groundwork for a fuller analysis of the linguistic and stylistic differences upon which they rest. Based as they are on a very large proportion of the text of the novels (the 4000 most frequent words typically comprise more than 94% of a novel), these results provide a wealth of material for stylistic analysis. The huge numbers of words involved will, however, require new methods of selection, analysis, and presentation if they are not to
prove overwhelming and incomprehensible. Meeting
these challenges will advance and refine authorship
attribution techniques, and, at the same time, further
illuminate the linguistic bases of James’s style and his process of revision. Fig. 1. 46 Novels by Henry James and 8 Other Authors
Fig. 2. Twenty-Three Novels by Henry James (revised versions marked with triangles)
References
Burrows, J. F. (1987). Computation into Criticism.
Oxford: Clarendon Press.
Burrows, J. F. (1992a). “Computers and the study of
literature.” In Butler, C. S., ed. Computers and Written Texts. Oxford: Blackwell, 167-204.
Burrows, J. F. (1992b). “Not unless you ask nicely:
the interpretative nexus between analysis and
information,” LLC 7: 91-109.
Burrows, J. F. (2002a). “‘Delta’: a measure of stylistic difference and a guide to likely authorship”. LLC 17: 267-287.
Burrows, J. F. (2002b). “The Englishing of Juvenal: computational stylistics and translated texts.” Style 36: 677-99.
Burrows, J. F. (2003). “Questions of authorship:
attribution and beyond.” CHUM 37: 5-32.
Burrows, J. F. (2005). “Who wrote Shamela? Verifying the authorship of a parodic text,” LLC 20: 437-450.
Craig, H. (1999a). “Contrast and change in the idiolects of Ben Jonson characters,” CHUM 33: 221-240.
Craig, H. (1999b). “Jonsonian chronology and the
styles of a Tale of a Tub. In Butler, M. (ed.).
Re-Presenting Ben Jonson: Text, History, Performance.
Macmillan. St. Martin’s, Houndmills, England,
210-32.
Hoover, D. L. (2003). “Multivariate analysis and the study of style variation.” LLC 18: 341-60.
Hoover, D. L. (2004a). “Testing Burrows’s Delta.” LLC 19: 453-475.
Hoover, D. L. (2004b). “Delta prime?” LLC 19: 477-495.
McKenna, C. W. F. and A. Antonia. (2001). “The
statistical analysis of style: reflections on form, meaning, and ideology in the ‘Nausicaa’ episode of Ulysses,” LLC 16: 353-373.
Rybicki, J. (2005). “Burrowing into translation: character idiolects in Henryk Sienkiewicz’s trilogy and its two English translations,” LLC Advance Access
published on March 24, 2005. doi:10.1093/llc/fqh051
Stewart, L. (2003). “Charles Brockden Brown:
quantitative analysis and literary interpretation,” LLC 18: 129-138.