by Dirk Siepmann
Cover picture: https://pxhere.com/en/photo/718737
What’s crime fiction? The question will seem old hat to specialists and fiction lovers alike. It’s usually about a murder that’s happened before the story begins, they’ll say, and the author plants clues about that murder to keep us guessing until the detective solves the crime and tracks down the murderer. This is the classic stuff of university seminars on genre theory – where you’ll also learn that much crime fiction has moved away from the simplistic plot-puzzle formula towards the harsh social realities faced by detectives, victims and criminals. In the end, you might be told that there is ‘no consensus’ among literary scholars on what crime fiction is, and a detective novel is a detective novel if you read it as one. This might leave you a little dissatisfied.
What if, instead, we were to make the heretical claim that the defining feature of a whodunit is that it contains some of the following word groups: ‘lean back in one’s chair’, ‘his/her lip trembled’, ‘light/get out a cigarette’, ‘raise one’s hand to’ ? Or that much of highbrow literature is just a stylistic extension of the trash novel? Where lowbrow authors make a simple statement like ‘There was silence.’, their highbrow colleagues will be tempted to elaborate: ‘An oppressive silence surged out of the place like a cloud.’ (Iris Murdoch) Proust was not far off the mark when he famously noted that ‘fine books are written in a kind of foreign language’. Captured in this bon mot is the realization that literature can be readily identified by its linguistic features. This is obvious with poetry, whose exceptional characteristics – metre, rhyme, stanzas, etc. – have long been acknowledged, but it is perhaps less evident in the case of prose fiction and its various genres.
So, how did the foreign language that is prose fiction arise? Historically speaking, a distinctively literary prose style is a comparatively recent development. It evolved only from the 1830s, when a rift developed between the bohemian world of postromantic artists and the bourgeoisie. This self-empowerment of the artistic sphere went hand in hand with an endorsement of art for art’s sake, an experimentation with new stylistic practices and an interest in literary style on the part of literary critics. Graphic evidence for this development can be had from the Google n-gram viewer, a tool that lets you research how words and phrases have evolved over the past few centuries: the terms ‘literary language’ and ‘literary style’ show a constant rise, from the 1830s/1840s to around 1900.
The systematic study of literary language is even more recent. It was Ferdinand de Saussure’s theory of the sign that turned the spotlight on language as a social phenomenon and on the literary text as an embodiment of a particular set of linguistic usages arising out of the collective langue. The Russian formalists – foremost among them Roman Jakobson – took this idea further by claiming that literariness is a linguistic effect produced in relation to other kinds of knowledge or discourse, the most important of these effects being ‘defamiliarization’ and ‘foregrounding’.
Modern stylistics, however, which did not really emerge until the 1950s, has often taken issue with such claims, arguing that literary language is not special since any formal feature termed ‘literary’ can be found in other texts. Literary stylisticians try to capture the specificity of literary language in functional categories, such as the distinction between the author and the narrator. In this view, the subjective impression of literariness conveyed by even the shortest passage of imaginative prose would merely be an epiphenomenon – an incidental accompaniment subordinate to the unfathomable rules of the artistic craft. Yet, significantly, this assumption of a functional difference without a formal correlative is without parallel or precedent in linguistics.
This is where ‘corpus stylisticians’ step in, a relatively recent brand of digital humanities scholars who’ve turned the traditional ‘top-down’ research process on its head: instead of laboriously sifting through dozens of, say, nineteenth-century novels to find that elusive stylistic feature that will buttress their hypothesis, people like Franco Moretti of the Stanford Literary Lab or Iva Novakova of Université Grenoble Alpes now use the computer to trawl through thousands of books in a matter of seconds. What they find is that, to paraphrase the corpus pioneer John Sinclair, ‘novels look rather different when you look at a lot of them at once’. Corpus research draws out the hidden patterns that form the subconscious basis of how literary texts construct reality. In Moretti’s words: ‘quantitative hermeneutics is the art of understanding conventions – forms, genres, styles, practices – better than their (i.e. in this case nineteenth-century) society ever did’. While this may be anathema to traditional literary criticism, still deeply rooted in the idea that facts are a subjective construction, it is in keeping with the more recent philosophical concept of new realism. New realists like Markus Gabriel have eventually unmasked both constructivism and metaphysics as being one-sided: for constructivists, reality is the world of spectators, for metaphysicians it’s the world without spectators. Actually, the world is both (and more), and facts do exist. Traditional close reading can build on statistical facts to tease out the social within the individual.
Some of the Stanford lab’s results have been justly described as ‘less than earthshaking’ by the New York Times. For example, Moretti and his colleagues looked at emotions associated with various places in fictional London between 1700 and 1900. Not entirely unsurprisingly, the West End was found to be associated with happiness, the City with fear. Less accessible to intuition and more in line with the concerns of traditional stylistics are the insights gained from another study, in which the lab explored sentence patterns in the nineteenth-century novel. When Anne Radcliffe writes ‘When Peter perceived the village, he burst into a shout of joy’, we (and she) are wholly unaware that she is following a typical nineteenth-century novelistic pattern. The pattern consists of a dependent clause that contains words referring to a setting or movement and an independent clause featuring expressions of emotions. That same sequence of space and emotion is a ubiquitous stereotype in the nineteenth-century novel.
Unfortunately, the Stanford lab hasn’t delved much deeper (which is a pity given their huge database). They might draw inspiration from a number of European linguists and literary scholars who consider style to be more than just words slotted into clauses. These ‘phraseologists’ extend their analysis beyond the word level to include sequences of words, such as simple strings of words that occur together (must have been) or relatively fixed word combinations (the strengthening sun, nervously light a cigarette). It is such word sequences that create a sense of literariness in the reader. Each genre has its own specific sequences, most of which are so discreetly conventional as to escape conscious notice. Many of these sequences act as narrative devices with a specific function. Thus, his thoughts / reflections / musings were interrupted when / by … is a transition marker many post-war writers use to shift from an internal to an external point of view. Such sequences turn out to be a more reliable indicator of genre than any description of typical plots by literary scholars.
Dominique Legallois of the Paris-based Sorbonne Nouvelle has ingeniously complemented the word-based approach by considering sequences that combine fixed word forms (e.g. puts), variable word forms (e.g. put, puts, putting) and grammatical units (e.g. present tense verb). These motifs, as he calls them, give us insight into previously unknown features of genres and individual authors, be it the bawdy scenes of Harlequin novels or the exquisitely chiselled style of a Zola or Flaubert. Motifs are pervasive in nineteenth-century fiction, and yet a hundred years of profound scholarship has failed to notice most of them. One of the most straightforward examples is the following structure: the (le/la) + noun + of the (du/de la) + noun which (qui) V in (dans), where the slots are filled by, respectively, nouns denoting wind (vent, brise, bise), nouns denoting night (nuit, soir), and verbs denoting death (mourir, gémir, pleurer, tomber, se briser). More intuitive once you’ve been alerted to it is the following recurrent structure from sentimental fiction: (s)he + past tense verb + possessive + noun (body part) + on + possessive + noun (body part), as in he put his hands on her hips, she felt his hands on her breasts or he fastened his mouth on her neck.
Using the motif approach, Iva Novakova of Grenoble University and her colleagues compare the literary patterns found in British and French authors. The results show that even such closely related cultures avail themselves of differing narrative patterns. Typically, English writers of fantasy novels, for example, will frame their dialogue with the rather pedestrian ‘(s)he said’, where French writers opt for a much larger variety of phrases. Similarly, where French writers might describe a male character as ‘advancing at a resolute pace’ (avancer d’un pas résolu), using a verb and a complex adverbial, their British counterparts will content themselves with a verb plus an adverb: ‘stride briskly’.
The last example nicely illustrates how linguistic analysis assists the literary scholar by uncovering the general stereotypes underlying much writing. Not just gender scholars will note that the behaviour denoted by ‘stride briskly’ reiterates the traditional link between masculinity and activity. In other words, motifs are potent world-building tools: they help writers shape characters, plot and setting along particular lines, and they activate familiar patterns in readers’ minds, allowing them to make sense of texts.
The quiet revolution sparked by the discovery of motifs will also impact on literary translation and writing classes. It allows us to describe the style of an author with almost mathematical precision and thus helps translators render recurrent constructions far more consistently across a novel. It will teach literary aspirants that good writing is not the result of unbridled ‘self-expression’ but rather requires close observation of other authors – neuronal networks are becoming uncannily good at this.
Let’s not forget, of course, that language and literature, like everything else, are in a state of perpetual flux, and that new ways of putting things are created just as babies are born while you read this. However, in the same way that babies are partial copies of their parents, so linguistic variation may well conceal underlying cognitive constants, as with the interrupt + thought example, which, in Stevenson’s day, would have corresponded to I was so far in my reflections when … – itself an exact calque of the present-day French equivalent j’en étais là de mes réflexions quand … There is more unity in historical diversity than either authors or readers realize but we have a lot more data crunching to do to bring it all to light.