Back to Book Reviews

Back to Cercles



Investigating Dickens’ Style: A Collocational Analysis
Masahiro Hori

New York: Palgrave Macmillan, 2004
$75,00, 251 p., ISBN 1-4039-2051-6


Reviewed by Jacqueline Fromonot


Masahiro Hori’s study, based on a corpus-driven approach, falls into three main parts. The first one charts the theoretical background to the analysis of collocations; then the second narrows down the investigation to the uses of collocations in Dickens’s works; in the third, Bleak House is singled out as a case study. This well-documented book offers a thorough, fresh outlook on the much-discussed question of Dickens’s style.

In part one, the author undertakes a historical survey of the study of collocation, quoting JR Firth’s “Modes of Meaning” (1957) as one of the very first articles pointing to its importance in the investigation of semantic statements [3]. Research on collocation, Hori notes, hinges on three main periods. First of all, the sixties offered a definition of collocation and its associated terms–cluster, span, node, range, or lexical set [5]. Then, in the seventies, analysis provided more specific definitions of collocation, as “co-occurrence of two items within [a] specific environment” (Sinclair, English Lexical Studies) [5]. As Hori points out, this limited the investigation to “significant collocations,” i.e. those appearing more frequently than possibly expected, contrary to “casual collocations” [5]. This type of research rested on computer-assisted studies and only dealt with linear co-occurrence, thus excluding syntactic and semantic statements [6]. Lastly, the eighties saw the swift development of computers, thanks to which the COBUILD project (a.k.a. the “Bank of English”) came out, with the resulting release of the COBUILD English collocations on CD-ROM, in 1985 [7]. The tools required for computer-assisted research on collocation had come into full existence, and the present study clearly stands in the wake of this new era.

In order to make the most of the methodological principles defined by earlier critics, Hori explains that contrary to Sinclair, Firth retained not only general collocational patterns made up of common words but also particular patterns indicative of registers or personal style [8], thus contributing to the advance of descriptive linguistics [9]. Firth also favoured a diachronic approach, noting for instance that some typically 18th-century collocations were still common in the 20th century, while others had become obsolete [10]. Lastly, Firth is credited with insisting on the existence of creative, unique and personal collocations, hence the coining of such labels as “Swinburnese” collocations, for example those used in his study of the poet’s works [10-11]. In the same attempt at classification, McIntosh’s “Patterns and Ranges” (1966) offers promising tools. McIntosh distinguishes four stylistic modes as far as collocations are concerned: “(1) normal collocations and normal grammar, (2) unusual collocations and normal grammar, (3) normal collocations and unusual grammar, (4) unusual collocations and unusual grammar” [12]. Each category is aptly commented upon by Hori.
The first one applies to any significant network, such as “characters’ idiolects” or “semantic features” [12]. However, usual collocations do not have to be frequent to be significant in literary language, Hori notes, pointing out that in Dickens, “blue eyes,” a common but low-frequency collocation, is always meaningful, because it is associated with innocent and favoured characters, like Nell in The Old Curiosity Shop [12].
Then, McIntosh’s “unusual collocations and normal grammar” category emerges through the unique figure of the oxymoron, or “collocational clash” [13], which raises the problem of acceptability [13]. For Hori, this is a tricky issue, especially when it comes to older texts, owing to the impossibility to find native speakers for informant tests of acceptability, the absence of a machine-readable corpus and the shortage of historical corpora—although on this last point Hori mentions the existence of CD-ROMs for 18th and 19th century fiction [13].
McIntosh’s third category, “normal collocations and unusual grammar”, is likely to appear in fiction based on the exploration of the “stream of consciousness” techniques, free indirect speech or thought representation, as is the case with James Joyce’s prose or the English haiku [13]. Admittedly, the last type, “unusual collocations and unusual grammar,” is quite rare in fiction, Hori acknowledges, although it is easy to create “undecipherable and nonsense sentences” to illustrate it [14].
The next point in Hori’s study concerns the notion of semantic prosody, a relevant tool for the analysis of collocation, since in Sinclair’s terms it is the “phonological colouring […] capable of transcending segmental boundaries” [14]. For instance, the phrasal verb “set in” usually collocates with negative or unpleasant elements. However, semantic prosody changes through time, as shown in the evolution of the negative prosody associated with “utterly” [15].

The following sub-part of Hori’s introduction brings us closer to Dickens’s own style, and the critic notes that Dickens’s collocations are extremely varied and offer more unique patterns than those of 18th-century fiction-writers like Daniel Defoe, whose Robinson Crusoe contains only two instances of the collocation [adjective + “eyes”], contrary to Dickens [18]. Hori might have realized, however, that considering the plot of Defoe’s novel, the chances of there being numerous mentions of any pair of eyes at all is more unlikely than in Dickens’s typically crowded scenes. The occurrence of this pattern is to be found with Richardson, though, as suggested from the corpus of these collocations (whether usual or not) taken from Pamela [19]. The reference to the “eyes” obviously points to the characters’ feelings and states of mind [20], a technique particularly noticeable in Dickens’s Hard Times. In this case, not only do these collocations reveal the characters’ psychological state, but they also contribute to the individualization of each of them [20]. At this point, Hori ventures the conclusion that the “richness of collocation in Dickens’ works may be considered his contribution to the development of the language of English fiction” [20].

The author points out next that Dickens’s use of collocations can be unique in his own language as well [20], using the evidence provided by R. Quirk (“Charles Dickens, Linguist”, 1974), in his general classification of Dickens’s style in four categories, “his use of language for individualization; for typification; his use of it structurally; and his use of it experimentally” [21]. Hori again picks up collocational patterns including the word “eye(s)” to show that typification by language is essential in Dickens’s style, with such occurrences as, among plenty of others, “a smith’s eye” (Great Expectations), contributing to characterization through the mention of the character’s occupation [21], while a certain “brightness” of the eyes will hint at discontent or pent-up emotions [21]. Structural language use also contributes to the opposition between characters, like Dora and Agnes in David Copperfield. Thus, while the “eyes” of David’s first spouse collocate with adjectives expressing physical beauty, Agnes’s are associated with words pointing to her psychological state [22]. Finally, Dickens’s much celebrated experimental use of language falls into the last category distinguished by Quirk, “unusual or deviant collocations”, of which “choking eyes” (Bleak House) provides a striking specimen [22].

Last but not least, Hori carefully tackles various methodological problems he had to address prior to starting his research [23]. Firstly, he warns us, acceptation of a word in collocation cannot be equated with a “lemma,” its basic form. Accordingly, “word forms that differ in number, tense or aspect are treated as different words” [23], since, for instance, a word may collocate differently whether it is in the singular or in the plural. Another problem is raised by identical word forms which belong to different grammatical categories and therefore collocate differently (“little,” for instance), a distinction a computer cannot make [24]. This implies that the syntactic relationship between a word and its collocate(s) has to be examined through a wide range of combinations [25]. The last point in Hori’s introduction concerns the computer-assisted method per se, which is applied to investigate usual as well as unique collocations [26]. This computerized approach is to be completed with native informant tests and comparisons with the databases of electronic texts of 18th and 19th -century fiction mentioned above.

To conclude briefly on this detailed introduction, we may say that the protocol seems highly reliable, thanks to the variety and consistency of the tools chosen and the elaborate scientific apparatus. What remains to be seen is to what extent the findings on Dickens’s style are up to such high standards. The results, which appear in parts two and three of the book, are—sadly enough—somewhat disappointing.

To begin with, part two investigates usual and creative collocations in Dickens’s works. Hori firstly deals with his methodological approach. He indicates that the Dickens corpus (containing approximately 4.6 million words), has been stripped of such words as proper nouns, function words, and “be” and “”do” (verbs and auxiliaries) among others [33]. Concerning usual collocations of highly-frequent nouns, adjectives and adverbs, the observations are based on the 100 highest-frequency content words [33]. Among the first results the database yields, one remarkable feature is the high frequency of the word “gentleman” [34], a sharp contrast with its rarer occurrence in other 19th-century writers’ fiction (with the notable exception of Thackeray) and in 20th-century fiction, which “may reflect a change of the word’s meaning and usage” [37]. Do we need, however, computer data to be aware that the issue of the status of a “gentleman” is indeed at the heart of these two novelists’ fiction? What is considered a finding is, at best, a confirmation. What is more, the list of usual words that collocate with “gentleman”, among which “old, sad, young, very” [39] hardly brings any clue as to Dickens’s preoccupations and least of all style. Indeed, Hori’s approach is rather descriptive at this point, as shown in the painstaking lists of nouns, adjectives and adverbs [38-53]—but this is hardly surprising, since these are the very pitfalls of such an approach.

The next chapter [57-96] deals with creative collocations, classified in eight juxtaposed categories: “metaphorical” (e.g. “malevolent baboon” referring to Uriah Heep in David Copperfield), “transferred” (in which the adjective actually applies to another element of the collocation, like “[Twemlow’s] innocent head” in Our Mutual Friend), “oxymoronic” (“curious indifference”, in Bleak House), “disparate” (in which the oxymoron does not create a clash but a mere disparity, like “[Miss Murdstone’s] delicious hand” in David Copperfield), “unconventional” (“raw afternoon” in Great Expectations), “modified idiomatic” (where Dickens modifies the idiom by truncating it or changing some of its elements, like “to the last extent” used instead of the standard “to some extent” in Dombey and Son), “parodied” (“ducky, ducky, come to Mrs Bond and be killed”, in David Copperfield, a parody of a line in a nursery rhyme) and “relexicalized” collocations (involving the novelist’s disregard for the original meaning and a personal reconfiguration, as in the case of Captain Cuttle’s numerous nautical terms in ordinary conversation in Dombey and Son.) [passim, 57-96].

This section offers a useful insight into Dickens’s creative genius, with plenty of material to comment upon and a relevant, if not illuminating, attempt at classification. Surely these corpora are precious material for further research. It is completed by the next chapter’s discussion of Dickens’s “role as a linguistic innovator” [98], through the study of his first-citations as recorded in the OED2, in the wake of Sorensen’s research on Dickens’s neologisms (Charles Dickens: Linguistic Innovator, 1985). A lot of Dickensian first-citations are dialectal or slangy. Others are collocational neologisms, which include many occurrences of the [noun + noun] type (“allotment-garden”), the [adjective + noun] combination (“lazy-legs”) or the reduplicative mode (“bon-bon”) [109]. There follows a flurry of compound adjectives built on various patterns, from “beef-faced” to “politico diplomatico” [110]. This allows Hori to say that “one cannot help but recognize Dickens’s power of expression, through which two or more words may be seemingly combined or juxtaposed at will” [108], and to acknowledge Dickens’s “acute consciousness of English, as an excellent linguistic recorder or discoverer of the Victorian age” [113].

The last part of the book is the case study of Bleak House, a choice which Hori justifies quite rightly (1) by the novelist’s “unusual strategy of dual narration” [117], involving at once a first-person narrator (a young woman) and a third-person narrator (a male adult) and (2) by the book’s attempt to typify the English society, a tendency which is likely to be represented through a great many collocations [117]. The investigation is conducted according to the same pattern as before, providing huge databases followed by attempts at analysis and synthesis. Thus, we learn that “the characteristics of language and style (…) are found in collocations or combinations of words rather than in the vocabulary of the novel, per se” [123]. In other words, many collocations are made up of rather usual words whose association itself produces originality. Then Hori proceeds to compare collocations taken from the third-person narrative, on the one hand, and from Esther’s narrative, on the other. The assumption is that the existence of two distinct narrators in charge of two distinct accounts is likely to produce stylistic differences, with potential echoes in the collocational strategies, or with collocations offering a choice perspective. Esther sounds more sympathetic, it appears, as is evidenced by the high frequency of the collocation [“poor little” + noun] in her own narrative [132]. Similarly, her extensive use of “so very” and its absence in the male characters’ speeches may imply, according to Hori, that Dickens found it a typically feminine type of diction [133]. The expression “so very” sounds indeed a little precious, but it shows little evidence for such a sweeping, gender-oriented judgement, might we say. More relevantly, though, Esther’s narrative style is defined elsewhere as “unpretentious” [141], as suggested by her handling of reporting clauses in the presentation of direct speech. Indeed, among the four possibilities available—initial, medial, final position, and the absence of clause (see Bonheim, The Narrative Modes, 1982)—Esther discards the more conspicuous options (initial and medial positions) and tends to use the least assertive patterns, now using a clause in final position, now not using any at all, as if to play down her own presence as a speaker. The grid showing the ratio of the positions of reporting clauses used by other characters indicates, by contrast, Richard’s and Skimpole’s predominant tendencies to prefer the medial position [141].
Another of Hori’s findings is worth mentioning as well, because it challenges the widely accepted idea among Dickensian critics that the third-person narrative ranks higher in terms of linguistic experimentation and poetic creativity than Esther’s narrative, traditionally seen as simple, or even banal and dull [151]. The results obtained by computerized research belie this assumption and substantiate Esther’s own significance by underscoring her experimental use of collocations [170]. To this end, a great deal of material—informant tests on the degree of unusualness of collocations in the two narratives, classifications of unusual collocations and finally comparisons between unusual collocations taken from the two narratives—has been carefully processed.
The wonder of computerized data is indeed the ability to confirm, question or contradict what readers or critics more or less intuitively feel, by giving quantified, objective information which seems undeniable. It is a pity, however, that so little sometimes comes out of such an elaborate set up. This is clearly the case in Hori’s study, as some sections (the last one, for example, devoted to new compound words seen as collocations in Bleak House) are entirely made up of corpora of classified examples, i.e. include little or even no commentary. This creates an impression of inconclusiveness which is detrimental to the initial project of defining Dickens’s style.

This book obviously offers new insights. It explores new perspectives of investigation, both regarding the theoretical approach to collocation in various kinds of texts, and the stylistic study of a particular writer. Indeed, in the general conclusion to the book, Hori reasserts the importance of collocational analysis, and expresses his hope that it will “bring about a shift in research emphasis concerning the English language, whether in literary or non-literary contexts” [208]. However, as far as Dickensian style is concerned, Hori’s investigation is a mere starting-point, that is a database from which further, deeper analysis should stem out. It certainly suffers from the blatant disproportion between the complexity of the technological tools in use and the meagre findings regarding Dickens’s stylistic techniques and choices. Identifying style is indeed a baffling issue. What is style anyway? The answer is a misty and mysterious one, and at any rate not (yet) within the scope of data-mining.


All rights are reserved and no reproduction from this site for whatever purpose is permitted without the permission of the copyright owner. Please contact us before using any material on this website.