The assumption that authorial style changes over time is a
reasonable one that is widely accepted in authorship attribution.
Some important studies concentrate on individual authors,
using a wide array of methods, and yielding varied results,
however, so that is unwise to generalize (see the overview in
Stamou 2007). Henry James’s style, for example, changes in a
remarkably consistent and extreme way (Hoover 2007), and
Austen’s stylistic development also seems consistent (Burrows
1987), but Beckett’s texts are individually innovative without
showing consistent trends (Stamou 2007: 3). For some authors,
different investigators have obtained inconsistent results. We
know of no general study that investigates how authorial style
changes over a long career. We thus explore how working and
otherwise living abroad for periods exceeding two years may
affect an author’s vocabulary and style.
Our presentation begins a broadly based study of the growth
and decay of authorial vocabulary over time. Although we limit
our study to nineteenth-century prose, to reduce the number
of possible variables to a manageable level, we compare French
and English authors, allowing us to test whether any discovered
regularities are cross-linguistic. Because authors’ vocabularies
seem related to their life experiences, we compare authors
who spent most of their lives in a single geographical area
with others who traveled and lived extensively abroad. For the
purposes of this study, we defi ne extensive living and working
abroad as at least two consecutive years, somewhat longer than
the contemporary American student’s or faculty member’s
stay of a semester or a year. This differentiation allows us to
investigate in what signifi cant ways, if any, extensive foreign
travel affects an author’s vocabulary.
Our investigation of these questions requires a careful selection
of authors and texts. Although research is still ongoing, we
have selected the following eight authors–two in each of the
four possible categories–for preliminary testing and further
biographical study: Domestic authors:
• George Meredith (1828-1909). German schooling at age
fi fteen, but apparently no signifi cant travel later; fi fteen
novels published between 1855 and 1895 available in digital
• Wilkie Collins (1824-1889). No foreign travel mentioned in
brief biographies; 20 novels available over 38 years.
• Honoré de Balzac (1799-1850). He traveled little outside
of France until 1843. Subsequent excursions abroad in
the last seven years of his life, mainly for romance, were
relatively infrequent and never exceeded fi ve consecutive
months. Over two dozen novels in digitized format are
• Jules Barbey d’Aurevilly (1808-1889). Raised in France and
schooled in the law, he was devoted to his native Normandy
and seldom ventured abroad. Initially cultivating the image of
the dandy, his novels and novellas create a curious paradox
in his later writing between sexually suggestive themes and
nostalgia for earlier aesthetics and a defense of Catholicism.
His literary productivity can be divided into the three
periods of 1831-1844 (fi rst published novel and novella),
1845-1873 (return to Catholicism; work as both literary
critic and novelist), and 1874-1889. At least fourteen novels
or novellas are available in digitized format.
• Dickens (1812-1870). Seven novels available 1836-43,
foreign travel for 3 yrs 1844-47, and more travel in 1853-55;
four novels after 1855 available.
• Trollope (1815-1883). Travel to Brussels in 1834, but
briefl y; six novels available before 1859, Postal missions to
Egypt, Scotland, and the West Indies, 1858-59; 1871 trip
to Australia, New Zealand, and US; travel to Ceylon and
Australia, 1875, South Africa 1877, Iceland 1878; fi ve novels
available after 1878.
• Arthur de Gobineau (1816-1882). Raised partly in France,
partly in Germany and Switzerland, he learned German
and began the study of Persian in school while his mother,
accused of fraud and estranged from her military husband,
kept the family a step ahead of the French police. Three
periods encompassing his publication of fi ction and often
related to career travel can be divided as follows: 1843-
1852 (at least 4 novellas and 1 novel); 1853-1863 (1 novella);
1864-1879 (10 novellas, 2 novels). Living mainly in France
from 1831-1849, he was a protégé of Alexis de Tocqueville,
who brought him into the French diplomatic service in
1849. Gobineau was then stationed in Switzerland and
Germany (1849-1854), Persia (1855-1858, 1861-1863),
Newfoundland (1859), Greece (1864-1868), Brazil (1868-
1870) and Sweden (1872-1873). Following travel through
Europe in 1875-77, he left the diplomatic service. His fi rst
short fi ction was published and serialized in 1846. At least
a dozen novellas written throughout his career (mostly
written and published as collections) and two novels (1852
and 1874) are available in digitized format.
• Victor Hugo (1802-85). Raised in France except for a
six-month period in a religious boarding school in Madrid
(1811-12), Hugo began writing his fi rst novel, Bug-Jargal
(1826) in 1820. This initial literary period includes 1820-
1851. Aside from a few short trips lasting less than three
months, Hugo lived and wrote mainly in his homeland
until his exile on the Island of Guernsey during the reign
of Louis-Napoléon from 1851-1870. The third period
encompasses his triumphant return to Paris in 1871 until his
death, celebrated by a state funeral, in 1885.
Research on the French authors is being facilitated by use of
PhiloLogic and the ARTFL database, complemented by local
digitized works and tools that are also used to investigate the
Preliminary testing must be done on all the authors to discover
overall trends or principles of vocabulary development
before investigating any possible effects of foreign travel. The
importance of this can be seen in fi gures 1 and 2 below, two
cluster analyses of English traveling authors, based on the 800
mfw (most frequent words). Dickens’s novels form two distinct
groups, 1836-43 and 1848-70, a division coinciding with his
1844-47 travels. For Trollope, the match is not quite so neat,
but it is suggestive. Ignoring Nina Balatka, a short romance set
in Prague, and La Vendee, Trollope’s only historical novel, set in
France in the 1790’s, only two remaining novels in the early
group were published after his travel to Egypt, Scotland, and
the West Indies in 1858-59.
Fig. 1 – Fifteen Dickens Novels, 1836-1870 Unfortunately, comparing cluster analyses of the English
domestic authors casts doubt on any simple correlation
between travel and major stylistic change. Meredith, like
Dickens and Trollope, shows a sharp break between his two
early novels 1855-57 and the rest of his novels (see Fig. 3). And,
ignoring Antonina (an historical novel about the fall of Rome),
Collins’s novels also form early and late groups (see Fig. 4).
Fig. 3 – Fifteen Meredith Novels, 1855-895
Fig. 4 – Twenty Collins Novels, 1850-1888
Furthermore, although our results for French authors are
more preliminary, a test of six works by Gobineau shows that
two of his earliest texts, written before any extensive travel,
are quite similar to some of his latest texts (see Fig. 5).
Fig. 5 – Six Texts by Gobineau
We are still developing the techniques for the study of the
effects of travel, but preliminary testing based on a division of
each author’s career into early, middle, and late periods allows
us to check for consistent trends rather than simple differences
between early and late texts, and to begin comparing the four
categories of authors. Choosing novelists with long careers
allows us to separate the three periods, selecting natural gaps
in publication where possible, but creating such gaps where
necessary by omitting some texts from the study. For traveling
authors, these divisions also take into account the timing of
their travel, creating gaps that include the travel.
Three stylistic periods create six patterns of highest, middle,
and lowest frequency for each word in the an author’s texts.
Depending on the number and size of the novels, we include
approximately the 8,000 to 10,000 most frequent words, all
those frequent enough to show a clear increase or decrease
in frequency; we delete words that appear in only one period.
Thus, as shown in Fig. 6, each of the six patterns would be
expected to occur about one-sixth (17%) of the time by
Fig. 6 –Six Patterns of Change (“E” =
early; “M” = middle; “L” = late)
Results for the English authors, shown in Fig. 7, are both
surprising and suggestive (note that the axis crosses at the
“expected” 16.7% level, so that it is easy to see which patterns
are more or less frequent than expected for each author.
Gradual decrease in frequency, E > M > L, is the only pattern
more frequent than expected for all four authors (Meredith’s
fi gure is only very slightly more frequent than expected), and
both M > E > L and L > E > M are less frequent than expected
for all four authors. Although the patterns for these authors
suggest some regularity in the growth and decay of vocabulary,
no simple relationship emerges. Consider also the surprising
fact that vocabulary richness tends to decrease chronologically
for Dickens, Trollope, and possibly Collins, while only Meredith
shows increasing vocabulary richness. (These comments are based on a relatively simple measure of vocabulary richness,
the number of different words per 10,000 running words;
for more discussion of vocabulary richness, see Tweedie and
Baayen, 1998 and Hoover, 2003.) These facts contradict the
intuitive assumption that the main trend in a writer’s total
vocabulary should be the learning of new words.
Similar comparisons are being developed for the French
authors in question. The conference presentation will include
not only a comparison of style within each language group,
but between language groups. Such comparisons also build
on strengths of corpus stylistics important to translation
(Goldfi eld 2006) and a possible related future for comparative
literature (Apter 2005).
Fig. 7 – Patterns of Frequency Change
in the Words of Six Authors.
Apter, Emily (2006) The Translation Zone: A New Comparative
Burrows, J. F. (1992a). “Computers and the study of
literature.” In Butler, C. S., ed. Computers and Written Texts.
Oxford: Blackwell, 167-204.
Goldfi eld, J. (2006) “French-English Literary Translation
Aided by Frequency Comparisons from ARTFL and Other
Corpora,” DH2006: Conference Abstracts: 76-78.
Hoover, D. L. (2003) “Another Perspective on Vocabulary
Richness.” Computers and the Humanities, 37(2), 151-78.
Hoover, D. L. (2007) “Corpus Stylistics, Stylometry, and the
Styles of Henry James,” Style 41(2) 2007: 174-203.
Stamou, Constantina. (2007) “Stylochronometry: Stylistic
Development, Sequence of Composition, and Relative
Dating.” LLC Advance Access published on October 1, 2007.
Tweedie, F. J. and R. H. Baayen. 1998. “How Variable May a
Constant Be? Measures of Lexical Richness in Perspective”.
Computers and the Humanities 32 (1998), 323-352.
If this content appears in violation of your intellectual property rights, or you see errors or omissions, please reach out to Scott B. Weingart to discuss removing or amending the materials.