Article
Language and Literature
2020, Vol. 0(0) 1–16
© The Author(s) 2020
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/0963947020952200
journals.sagepub.com/home/lal
The role of empirical methods
in investigating readers’
constructions of authorial
creativity in literary reading
Fabio Parente
University of Nottingham, UK; University of Derby, UK
Kathy Conklin
University of Nottingham, UK; Max Planck Institute for Psycholinguistics, the Netherlands
Josephine M Guy
University of Nottingham, UK
Rebekah Scott
University of Nottingham, UK
Abstract
The popularity of literary biographies and the importance publishers place on author publicity
materials suggest the concept of an author’s creative intentions is important to readers’ appreciation
of literary works. However, the question of how this kind of contextual information informs literary
interpretation is contentious. One area of dispute concerns the extent to which readers’ con-
structions of an author’s creative intentions are text-centred and therefore can adequately be
understood by linguistic evidence alone. The current study shows how the relationship between
linguistic and contextual factors in readers’ constructions of an author’s creative intentions may be
investigated empirically. We use eye-tracking to determine whether readers’ responses to textual
features (changes to lexis and punctuation) are affected by prior, extra-textual prompts concerning
information about an author’s creative intentions. We showed participants pairs of sentences from
Oscar Wilde and Henry James while monitoring their eye movements. The first sentence was
followed by a prompt denoting a different attribution (Authorial, Editorial/Publisher and
Corresponding author:
Kathy Conklin, School of English, University of Nottingham, Trent Building, Nottingham NG7 2RD, UK.
Email: aezkcc@nottingham.ac.uk
Typographic) for the change that, if present, would appear in the second sentence. After reading the
second sentence, participants were asked whether they had detected a change and, if so, to describe
it. If the concept of an author’s creative intentions is implicated in literary reading this should
influence participants’ reading behaviour and ability to accurately report a change based on the
prompt. The findings showed that readers’ noticing of textual variants was sensitive to the prior
prompt about its authorship, in the sense of producing an effect on attention and re-reading times.
But they also showed that these effects did not follow the pattern predicted of them, based on prior
assumptions about readers’ cultures. This last finding points to the importance, as well as the
challenges, of further investigating the role of contextual information in readers’ constructions of an
author’s creative intentions.
Keywords
Authorial intention, eye-tracking, literary creativity, prose fiction, text editing, textual variants
1. Introduction
Within the broad discipline of English Studies, the concept of an author’s creative in-
tentions, and the role of this concept in literary interpretation, have long been contentious
issues. For the last half century or so, excepting the genre of literary biography and some
forms of text-editing, academic enquiry has tended to concentrate on investigating the
structures – linguistic, cultural and historical – through which such creative agency is
constituted by the reader, as well as the function of these constructs in the generation of
literary meaning. However, in a recent overview of the rich body of linguistic, theoretical
and philosophical literature devoted to these topics, Guy et al. (2018) pointed to the
continuing absence of empirical support for almost all current hypotheses about how
authors’ creative intentions are constructed in the proces s of literary reading. They further
suggested that understanding of this concept has been impeded by a lack of clarity about
the extent to which authorial intention can be considered a primarily linguistic or text-
centred phenomenon, and therefore the light that can be shed on it by stylistic analysis
alone. While it is uncontentious to assert that authorial intention begins ‘not at the moment
of reading the text, but in the reader’s culture’ (Stockwell, 2016: 160), the precise re-
lationship between these variables remains to be determined. Likewise, although Chatman’s
(1978: 147–150) elaboration of the concepts of the ‘implied’ and ‘real’ author have been
useful for pointing to the distinctiveness of the concept of authorial creativity inferred from
the text, the question of how precisely that construct is informed by contextual information
about the biographical author continues to be under-investigated. Guy, Conklin and Davies
ended their review by showing how the application of psycholinguistic methodologies,
including the use of eye-tracking technology to test readers’ actual (as opposed to hy-
pothesised) responses to texts, might be employed to address these issues. The present study
takes up this challenge.
Research by Carrol et al. (2015) has demonstrated how eye-tracking can be used to
determine which linguistic features, including fine-grained ones such as punctuation,
readers pay most attention to during literary reading. Experiments can therefore be set up
to test literary stylisticians’ hypotheses about what constitutes literary language, in the
sense of demonstrating whether readers are actually noticing the specific linguistic
2 Language and Literature 0(0)
features posited as being implicated in the generation of literary meaning, such as those
associated with foregrounding effects, or with stylistic novelty or complexity. In eye-
tracking experiments, as explained more fully below, the measure of noticing or attention
is the amount of time that the reader’s eye lingers on some words or phrases – experimental
regions of interest (ROIs) – relative to others. Eye-tracking can also be used to test whether
readers’ respon ses to certain textual features are affected by prior, extra-textual prompts.
These may include contextual information about the words in front of them, such as who
authored them. It is this latter facility that we exploit in the study reported here. Our aim
was to determine whether assigning interpretative significance to given textual features is
dependent on a framework of understanding brought to (rather than derived from) the text,
when that fram ework involves information about how a text was authored.
There is much anecdotal evidence, including the enduring popularity of literary bi-
ography and the significance given to author publicity materials (such as photographs and
interviews) in the marketing of literary fiction, to suggest that the concept of an author’s
creative intentions remains an important element in most readers’ appreciation of literary
works. Yet little empirical research has been undertaken to investigate whether and how
the proliferation of information about ‘real’ biographical agents, and their creative in-
tentions towards their texts, is deployed in literary reading. An important exception,
discussed by Guy et al. (2018), is that of Claassen (2012), whose study of author rep-
resentations in literary reading is different to that undertaken here. Because Claassen
focused on an issue that does not pertain to all literary reading (how readers perceive the
moral content of a story), and on atypical texts (those which had proved highly pro-
vocative, and about which there had been significant publicity), her findings are not easily
generalisable to the question we address here: the nature of the relations hip between
textual and cultural factors in readers’ construction of an author’s creative intentions.
Testing the impact of the many and diverse kinds o f information about authors and
authorship on literary reading is clearly a challengin g task. Our study focuses on just one
kind of evidence commonly held to exhibit an author’s creative intentions: the textual
variations to be found in the processes of re-writing and revision which are recorded by
text editors (for further discussion of the controversies surrounding the recording of these
variants and their creative significance, see Guy et al., 2016). These variants are a useful
resource for such enquiry both because they are authen tic, meaning they are not the result
of artificial manipulation. And because they typically involve small-scale changes to
single textual features, such as a lexical item, or a change in punctuation, and this feature
makes it easier, under experimental conditions, to isolate the effect of a prior input on the
attention readers pay to them. (For a discussion of the limitations of psycholinguistic
enquiry, in these respects, see Conklin and Guy, 2019; Conklin et al., 2018).
In the current study, we used eye-tracking to determine whether the attention readers
paid to changes made to small-scale textual variants was influenced by prior information
about the kind of creative agency behind them. The technology allows for natural
reading – or natural reading from a computer screen – providing a rich moment-to-
moment record of eye movements for actual readers. Eye-tracking assumes an ‘eye-mind
equivalence’, whereby what is being fi xated is thought to be what is being processed at any
given time (Pickering et al., 2004), meaning that the amount of time that the eyes spend
looking at a word or section of text indicates how much cognitive effort is being expended
Parente et al. 3
to process the input. More precisely, eye-tracking technology tells us where people’s eyes
land (fixation), how many times they land in that position or region (fixation count), and
how long each fixation lasts (fixation duration), as well as movements back to previously
read sections of text (regressions). Longer and/or greater fixations, as well as longer and/or
more regressions, provide an indication of greater processing effort or attention.
Ascribing ‘value’ to a textual variant necessitates readers looking at examples of the
‘choices’ that were made – be they by an author freely exercising his/her creative de-
cisions, or by an editor or publisher, or perhaps by a typesetter or printer – alongside each
other. If a variant is interpretively significant, readers’ conscious and/or unconscious
behaviour should reflect this. In the current study, we used eye-tracking to examine the
effect of three different prompts on two sorts of textual changes (lexical and punctuation):
(1) a prompt informing readers that they were looking for potential textual changes that
had been brought about by a canonical, named literary author (in this case, Henry James or
Oscar Wilde); (2) a prompt attributing potential changes to an editor, an agent not usually
thought of as authorially creative in the same sense; or (3) a prompt indicating that
potential changes were the result of an ‘accident’, such as a typographical error. When
readers are presented with two extracts that are identical except for a change in one textual
feature, we would expect that the second reading of the text would be read faster than the
first (e.g. Hy
¨
on
¨
a and Niemi, 1990; Levy et al., 1992). The key question is how a textual
variant is read when it has a different attribution. We hypothesised that if a concept of an
author’s creative intentions was strongly implicated in literary reading, and that if it
derived – at least in part – from information brought to the text, there would be a dif-
ferential between the amount of attention paid to the ROIs (where the textual change was
located) depending on the prior prompt. We also anticipated that the kind of prompt would
have a differential effect on readers’ accu racy reporting lexical versus punctuation
changes, with information about authorial agency leading them to expend more effort, and
achieve greater accuracy, in the reporting of the latter changes. Notably, if a particula r
attribution leads to more and longer fixation and/or greater accuracy in detecting variants,
it is important to note that this does not establish how the change is interpreted, nor
whether and how it may be judged as ‘artistic’. Alternatively, if a reader’s concept of an
author’s creative intentions, as some literary stylisticians have assumed, is largely text-
centred, there should be no (or only a very minimal) difference between attention to the
ROIs for the three different prompts; with the prompt also making no difference to the
amount of attention readers pay to ROIs involving lexical rathe r than punctuation changes .
As touched on above, eye-movement and reporting behaviour only provi des evidence
for the presence or absence of an interpretative act. Our assumptions about interpreting
this behaviour derive from an eye-tracking study that presented readers with texts in which
some words were replaced by semantically similar ones which were either in linguistic
focus or not in the sentence; in other words, the change was either foregrounded by the
preceding context or not (Ward and Sturt, 2007). Readers had more fixations and longer
reading times for changed words when they were in focus, but not when they were out of
focus. Because focus indicates the information that is the most important or prominent in
a sentence (Halliday, 1967), one might argue that the findings by Ward and Sturt
demonstrate that greater attention is paid to what are interpretively signi ficant changes
in a text.
4 Language and Literature 0(0)
Also important for the current study is previous research on how readers process
punctuation (Hill and Murray, 2000; Hirotani et al., 2006). While most studies compared
reading in sentences with no punctuation to those that were punctu ated, two eye-tracking
studies investigated the impact of changes in sentence internal punctuation to already well
punctuated sentences, by presenting readers with extracts from the 1846 and 1867 editions
of Charles Dickens’s Oliver Twist and from the 1881 and 1908 editions of Henry James’s
The Portrait of a Lady (Carrol et al., 2015; Parente et al., 2019). In these two studies,
readers encount ered pairs of sentences that differed in some way between the two editions
in terms of lexis and/or punctuation. Carrol et al. (2015), in sentences that had one or two
changes, showed an increase in reading times to the ROI containing the change relative to
the rest of the sentence that remained unchanged. There was no evidence in the reading
record to suggest that changes to punctuation were less noticeable than changes to lexical
items. However, changes in lexis triggered more re-reading of the whole sentence,
whereas changes to punctuation did not, indicating that readers may implicitly ascribe
more ‘semantic load’ to lexical changes, which causes them to reconsider the sentence as
well as the change itself. This was not the case for punctuation changes, which may
suggest that such features are deemed minor variations with limited interpretative sig-
nificance. Parente et al. (2019), using pairs of sentences that only had one change, in-
vestigated the influence of reader expertise and whether performance was influenced by
a task-specific ‘spot-the-difference’ effect. They found ROIs with punctuation changes
required greater processing effort, demonstrating that identifying changes in punctuation
is more effortful, which aligns with the participants’ difficulty with consciously identi-
fying changes in punctuation. They also found that expertise played little role in readers’
greater sensitivity to lexical rather than punctuation changes and that the advantage for
identifying lexical changes persisted when the time interval between exposures was
increased. This evidence confirmed earlier findings that small-scale features – like changes
to punctuation – may not possess the creative significance predicated of them by critics
and text editors. The current study explores whether this is the case when readers are
alerted to the source of the variant: whether that source is the author, the editor, or the
change is due to a typographical error.
2. The study
2.1. Methods
2.1.1. Participants. Thirty-six participants from a UK university were paid for their
participation. One was excluded from the data set due to poor data quality (i.e. too much
track loss). Participants were undergraduate and postgraduate students from various
departments (aged between 18 and 45, M = 24.63, SEM = 0.97).
2.1.2. Materials. We selected 60 pairs of sentences from two versions (the 1890 periodical
text and 1891 book-text) of Oscar Wilde’s The Picture of Dorian Gray (variants are
printed in Bristow, 2005) and 60 from two editions (the 1881 Macmillan first book-text
and the 1908 Scribner’s New York Edition text) of Henry James’s The Portrait of a Lady
(select variants a re printed in James, 2011), yielding a total of 240 experimental variants.
Parente et al. 5
