Moll Flanders: Calculating Voice

  1. 1. Larry Stewart

    Department of English - College of Wooster

Work text
This plain text was ingested for the purpose of full-text search, not to preserve original formatting or readability. For the most complete copy, refer to the original conference program.

Literary critical discussions of Daniel Defoe's The Fortunes and Misfortunes of the Famous Moll Flanders have long focused on the protagonist's, Moll's, narrative voice. Dorothy Van Ghent (1953) notes the "relatively great frequency" of Moll's "use of words naming that kind of object which constitutes material wealth" (49) and indicates that these words connote "the counting, measuring, pricing, weighing, and evaluating of...things in terms of the wealth they represent and the social status they imply...' (50). More recent criticism has problematized the identity of Moll's voice or, at least, the voice that purports to be Moll's. These studies frequently call attention to the fictional editor of the novel, an editor who reports on the difficulty of putting Moll's words "into a Dress fit to be seen" and attempts to frame Moll's narrative as a story of repentance for a life of sin. For example, Thomas Grant Olsen argues that the editor's claim to have re-written Moll's prose "obliterates the notion that the narrative is Moll's at all" (468). Thus, Moll's narrative voice is alternately viewed as one that uses specific kinds of words and has a specific identity or one that cannot be discovered, possibly an absence rather than a presence or, in the view of some, a female voice preempted by that of a male.

The question posed by this paper is whether quantitative analysis can provide a useful way of way of thinking about this problem and supplement more traditional critical approaches. More generally, this is part of an on-going series of explorations and test-cases to determine whether the kind of stylometric techniques used in studies of authorship attribution can distinguish among different narrative voices of an individual writer and, thus, provide useful insights concerning literary-critical issues focused on narrative voice. (See Stewart, 2003.)

The study began by utilizing the Burrows' technique to compare the voice of the fictional editor of Moll Flanders to the voice found in the rest of the novel. The relative frequencies of the forty most common words were used as variables in a principal components analysis of the novel's preface and the rest of the novel divided into seven approximately equal parts. The assumption was that if the voice of the preface is statistically similar to that of the other parts of the novel there is at least some evidence to support recent theories that Moll's voice is not separable from that of the editor. In fact, however, the principal components analysis displayed on a scatter graph shows a wide discrepancy between the editor's voice and the voice in the remainder of the novel. (See Fig. 1)

Figure F1
Figure 1

Of course, this discrepancy does not necessarily show the voices to belong to different characters. Baayen and others have convincingly argued that "differences in genre override differences in education level and authorship"; one would assume the same to be true in trying to differentiate the voices of different characters by an individual writer. The genre of the preface may simply differ from the genre of narration. Clearly, there needed to be a different method of determining whether Moll's voice is "her own."

It seemed worthwhile then to compare the voice in Moll Flanders to the voices of narrators in Defoe's other novels as well as to narrative voices in several other eighteenth-century narratives. (The other novels are Susanna Rowson's Charlotte Temple, Sarah Fielding's The Governess, Clara Reeve's The Old English Baron, Henry Fielding's Jonathan Wild, Samuel Richardson's Pamela, Jonathan Swift's Gulliver's Travels, Charlotte Lennox's The Female Quixote, and Eliza Haywood's Miss Betsy Thoughtless.) Again, the Burrows' technique was used. Several different analyses were made: one with the whole of each of Defoe's novels, one with each of Defoe's novels, several of his essays, and the group of sample novels from the same time period, and one with each of Defoe's novels divided into sections. As well, a cluster analysis was done using each of those sets of texts. In every case, the results are remarkably similar to those shown in Figures 2 and 3.

Figure F2
Figure 2

Figure F3
Figure 3

All the analyses show a clear distinction between Defoe's five novels narrated by male characters (Robinson Crusoe, The Further Adventures of Robinson Crusoe, Captain Singleton, Col. Jacque, and A Journal of the Plague Year) and those narrated by female characters (Roxana and Moll Flanders). As well, the analysis using sections of each novel shows overlapping among the sections of Moll Flanders and Roxana but no overlapping between sections of those novels and sections of Defoe's other novels. At least in quantitative terms, Defoe's two female narrators seem to speak in nearly the same voice.

Obviously, such evidence does not indicate that Moll has a unique voice. However, insofar as her voice is statistically indistinguishable from the voice of Defoe's other female narrator, there seems no reason to believe her voice has been preempted by the voice of the editor. She speaks as Roxana speaks, and no one has suggested Roxana's voice to be other than her own. (At the conference, a much fuller analysis would be given of the differences in voice between Defoe's female and male narrators.)

In attempting to understand Moll's voice, another set of tests was used. Taking a cue from Dorothy VanGhent's comment that Moll's vocabulary is characterized by words of measuring and evaluating and from other critics who remark on the emphasis on morality and religion, I attempted a content analysis by inductively generating two lists. One list is comprised of the language of calculation of which VanGhent speaks and the other of what I call the language of religion and morality. The procedure was to create a single alphabetical list of all the words used in Defoe's novels as well as in the other eighteenth century novels and then, with no knowledge of the specific novels from which they came, produce a list of words that seemed to connote in the one case calculation and evaluation (words indicating measurement, finances, and ranking) and, in the other, religion and morality (words referring specifically to religion such as deity, pastor, lord, etc. and those referring more generally to morality such as chastity, innocence, guilt, repentance, etc.). A list of 200 lemmata resulted and would be shown at the conference. Clearly, such a procedure is open to subjective factors and results need to be used with extreme caution. As Burrows has stated, "it has become customary, in recent years, to allow variables to 'declare themselves,' thus obviating, as far as possible, the danger of a predetermined outcome" (268). The words used in these analyses are not self-declared.

However, the results do show that Moll Flanders uses these words of calculation at a rate significantly higher than that of other novels in the study. Among Defoe's narrators, Moll and Roxana have narrative voices in which such words occur most frequently. More importantly, these words occur significantly more frequently than they do in the language of the editor. On the other hand, the language of religion and morality is used much more frequently by the fictional editor than by Moll herself. (See Figure 4) The editor, in fact, uses words connoting religion and morality at a frequency about six times that of Moll.

Figure F4
Figure 4

The results of these analyses seem to have implications that bear on critical discussions of Moll Flanders, implications that would be discussed in detail at the conference. There seems to be no stylometric evidence that Moll's voice has been preempted by the voice of the male editor, particularly given the similarity of Moll's voice to that of Roxana. As well, whereas the editor's voice makes heavy use of the vocabulary of morality and religion and relatively little use of the vocabulary of calculation, Moll uses religious language sparingly but uses the language of calculation with great frequency. If the male editor does indeed attempt to preempt Moll's story and her voice, the irony of the novel may be found in his failure to do so.


1. Baayen, R. H., Tweedie, F., van Halteren, H., and Krebbers, L. (2000). Back to the cave of shadows: stylistic fingerprints in authorship attribution. ALLC/ACH 2000 Conference Abstracts: 156-158.
2. Burrows, J. (2002). 'Delta': a measure of stylistic difference and a guide to likely authorship. Literary and Linguistic Computing, 17: 267-287.
3. Olsen, T.G. (2001). Reading and righting Moll Flanders. Studies in English Literature 1500-1900. 41: 467-481.
4. Stewart, L. (2003). Charles Brockden Brown: quantitative analysis and literary interpretation. Literary and Linguistic Computing 18: 129-138.
5. VanGhent, D. (1953) The English Novel: Form and Function. New York: Harper.

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.

Conference Info



Hosted at Göteborg University (Gothenburg)

Gothenborg, Sweden

June 11, 2004 - June 16, 2004

105 works by 152 authors indexed

Series: ACH/ICCH (24), ALLC/EADH (31), ACH/ALLC (16)

Organizers: ACH, ALLC

  • Keywords: None
  • Language: English
  • Topics: None