The Words of Facts: How British and Italian Students Use in Fact in Academic Writing

This study examines the frequency, meanings, lexico-grammatical and discursive features of the pragmatic marker in fact in an L1 subcorpus of the British Academic Written English corpus and in a learner corpus of essays written by Italian university student s. One of four meanings, ‘c ontrast’, ‘s pecification’, ‘factuality’ and ‘suppor t’, was assigned to each instance of in fact in both corpora. Similarities and differences emerged in the use of this pragmatic marke r. All four meanings are present in both corpora; however, the ‘c ontrast’ and ‘s pecification’ meanings occur more often in the L1 corpus, while the ‘factuality’ and ‘suppor t’ meanings occur more often in the L2 corpu s. These findings suggest that the Italian learners are not fully aware of the multiple meanings of in fact .


Introduction1
The processing and interpretation of communication is facilitated by the use of cohesive devices, which signal the relationship of a chunk of discourse with previous or later text (Povolná, 2012: 132).However, the specific type of relationship marked by a cohesive device may vary depending on the co-text in which it occurs (e.g. its preceding and following propositional content, its position in the utterance/sentence, its prosody).A case in point is the pragmatic marker in fact.
In fact signals speakers'/writers' awareness of how an utterance is relevant to its co-text (Aijmer andSimon-Vandenbergen, 2004: 1783) and also their attitude of 'evaluative certainty' toward the content of the previous co-text (Schwenter and Traugott, 2000: 7).2In the former case, it acts as a discourse marker, in the latter case, as an interactional marker.3 In fact originally conveyed the meaning of 'in truth' (Aijmer andSimon-Vandenbergen, 2004: 1790).In Present-Day English, its core meaning can be defined as announcing an immanent discrepancy in propositional attitude (Smith and Jucker, 2000).Within this general meaning, more specific meanings can be identified.One is introducing unexpected content (Oh, 2000;Tseronis, 2011: 480) which specifies "what is really the case" (Simon-Vandenbergen and Willems, 2011: 343), as in I know the mayor really well.In fact, I had dinner with her last week (Longman Dictionary of Contemporary English online).Another is introducing a strong opposing argument (Tseronis, 2011: 481), as in They told me it would be cheap but in fact it cost me nearly $500 (Longman Dictionary of Contemporary English online).
Lexicographers and scholars assign a varying number of meanings to in factone (e.g.Merriam-Webster online; Fraser, 1996;Furkó, 2014), two (e.g.Macmillan 1 The authors are jointly responsible for the design of the study as well as for data collection and analysis.Sara Gesuato wrote Sections 1, 2 and 5. Katherine Ackerley wrote Sections 3 and 4. 2 In fact has also been interpreted as a contrastive/elaborative conjunction, evidential marker, topic marker, elaborative marker, elaborative marker or stance adverbial (Oh, 2000: 244-245).3 We use the terms discourse marker and interactional marker to identify sub-types of pragmatic markers.We are aware that other terminological and conceptual distinctions are possible.For instance, Fung and Carter (2007: 411) include among discourse markers those units that signal transitions in conversation, the relation of an utterance to its preceding context, and "an interactive relationship between speaker, hearer and message".Similarly, Pulcini and Furiassi (2004: 109-110) consider discourse markers both those with an interactional/interpersonal function, "used to control cooperation between speakers", and those with a textual function, which serve to structure the text, signal cohesion between utterances or show a relation between what is being said and the situation.Dictionary online; Buysse, 2020) or more (e.g.Schwenter and Traugott, 2000).This suggests that, as is the case for other markers, in fact brings a 'meaning potential' to its surrounding discourse, which then activates a co-text-specific function.The compatibility of in fact with varied propositional content gives language users the opportunity to flexibly adapt this marker to their communicative purposes.However, this semantic indeterminacy may give rise to ambiguity, when contrasting interpretations are equally plausible in context.Additionally, in a domain that requires formal training and practice, such as scholarly communication, it may pose challenges to non-expert users, especially in a second/foreign language (L2/FL).
Although in fact has been extensively studied (see Section 2), some issues are underexplored.More specifically, although the literature reports on L1 interference phenomena in non-native use of in fact (e.g.Buysse, 2020), to our knowledge, research has not been carried out on its use by Italian learners of English, whose use of in fact might be influenced by the Italian cognate infatti.Also, while previous studies have mostly considered spoken data, an in-depth examination of in fact in written texts is missing.Finally, we do not know which distinct meanings are signalled by means of in fact in novice L1 academic writing.It is these knowledge gaps that we aim to address.
Below we review some of the literature relevant to the pragmatic marker in fact and specify the focus of our study (Section 2).We then outline our data collection and analysis (Section 3) and present our findings (Section 4).Finally, we discuss the findings (Section 5), and draw implications from them (Section 6).

Literature Review
We set the context of our study by highlighting the range of potential meanings of in fact and then outlining the semantic profile of the Italian cognate infatti.
Studies on in fact account for and exemplify its meanings.Some have examined the semantic evolution of this marker and the meanings it conveys in Present-Day English.Schwenter and Traugott (2000) reconstructed the evolution of three meanings of in fact.They observe that in fact first occurs as an adverbial at the Verb Phrase level, and introduces what the evidence suggests.In this case, it means "in practice, as far as can be told from evidence, in actuality", and answers the question "With respect to what?" (p.11).They then report that later in fact shows up as a sentential adverb indicating strong belief.Simon-Vandenbergen and Willems (2011) identified similar patterns of development for in fact.It first occurs in legal texts as a prepositional phrase meaning 'in reality' or 'in practice' , sometimes collocating with true or evident.Later on, in fact becomes an adversative adverb expressing certainty and contrast.Finally, it develops the sense of strengthening the preceding co-text, possibly from borderline cases conveying the notions of 'opposition' and 'strengthening' .
Scholars examining in fact from a synchronic perspective also draw attention to its different meanings.Aijmer and Simon-Vandenbergen (2004) distinguish three meanings of in fact, which are accounted for as generalised conventional implicatures.One is the adversative meaning of contradicting expectations.Combining the notions of 'certainly' and 'however' , it is glossed as 'but in fact' .Another is the additive meaning of reinforcing and correcting expectations.Glossed as 'and in fact' , it combines the notions of 'what's more and indeed' .The third is the enhancing meaning of justifying expectations.It is glossed as 'for in fact' , and introduces the reason for one's standpoint.
Other scholars focus on one specific meaning of in fact.For example, Fraser (1996: 186-188), Halliday (1985: 303) and Martin (1992: 212) classify it as an elaborative marker announcing a reformulation, clarification or refinement of previous discourse.Others classify in fact as an adversative marker signalling denial.For example, Halliday and Hasan (1976) interpret it as conveying the notion of 'contrastive avowal' , that is, as "an assertion of 'the facts' in the face of real or imaginary resistance" (p.254).Still others assign to in fact an in-between meaning, seeing it as a marker that introduces an emphatic, corrective statement.Viewed this way, in fact thus signals not only that the speaker is certain about the credibility of upcoming discourse (Bruti, 1999: 531;Hyland, 1998: 442;Oh, 2000: 262;Mortier and Degand, 2009: 343), but also that she does so by conveying discrepant information or alternative interpretations (Mortier and Degand, 2009: 342;Oh, 2000: 266;Smith and Jucker, 2000: 210;Tseronis, 2011: 481, 484).
Overall, the above studies show that in fact is polysemous, and that its meanings do not necessarily have clear-cut boundaries.That is, in fact appears to be contextually adaptable and ambiguous: it conveys different meanings in different stretches of discourse and it is interpretable in alternative ways in the same piece of text.However, we do not know how such polysemy and indeterminacy play out in academic discourse -in particular, which meanings of in fact are attested in novice academic writing -and which are most frequent.Contrastive PragmaticS 4 (2023) 274-288 Infatti is the Italian cognate of English in fact.It is also used as a pragmatic marker, but in ways that are only partly comparable to in fact.It can thus be considered a false friend.Studies on Italian infatti have been carried out on spoken data.Bruti's (1999) research, conducted on dialogic discourse, shows that, although both in fact and infatti signal "thematic connectedness" (p.531), they differ in three respects.First, infatti is backward-oriented, linking what is about to be said to what has been said previously, while in fact is forward-oriented, paving the way for what is going to be stated next.Second, infatti signals the imminent closure of the topic being discussed, thus potentially bringing an exchange to a close.Instead, in fact signals the imminent development of a new (sub-)topic, contributing to the progress of the interaction.Third, infatti displays support or confirmation of what has been stated before through exemplification, demonstration, mention of consequences or more general thematic connectedness.On the other hand, in fact introduces discourse that may either confirm or refute what has been said before by means of correction, specification, reformulation and/or clarification in terms of precision or intensity.Bruti's dialogic data show that in fact mainly introduces an autonomous, confident assertion, elucidating corrections.Infatti, instead, typically found in second-speaker responses, expresses some degree of agreement with the previous speaker's utterance.Similarly, Battaglia (2021) illustrates how infatti expresses the general notion of 'confirmation' of a previous claim.In samespeaker utterances, it introduces a justification or consequence of a previous utterance.In dialogue, it shows agreement with or affiliative response to a previous utterance.Finally, Andorno (2016) shows that infatti is typical of replies to statements and questions (p.115), introducing confirmation of previous assertions or alignment with the previous speaker's opinion (pp. 109-110).
Overall, the above findings show that, in spoken data, Italian infatti expresses support of previous discourse.In same-speaker utterances, it introduces the possible development and conclusion of a topic by confirming it; in second-speaker utterances, it conveys agreement and compliance with a previous speaker's statement.Only the former function minimally overlaps with one of those served by in fact, namely supporting previous discourse with a stronger elucidating statement.Given the formal and partly functional similarity between in fact and infatti, it is possible that Italian EFL speakers use in fact in the same way as infatti (cf.Buysse, 2020: 29).However, to our knowledge, research has not been carried out on Italian speakers' use of in fact in writing.
On the basis of the above findings and considerations, we set out to compare and contrast how in fact is used by novice writers, both native speakers and Italian EFL learners.The research questions (RQ s) we address are the following: To address the above research questions, this study considers two corpora of written English: one composed of texts written by university students who are L1 speakers of English, the other of texts produced by Italian university students of English.The initial part of this section will describe the two corpora.
It will then go on to describe the tools and procedures used for their analysis.

Data
The first dataset consists of a subcorpus of the British Academic Written English corpus (BAWE).It includes contributions from "undergraduate and Master's students in a wide range of disciplines" (Alsop and Nesi, 2009: 71) studying at Warwick University, Reading University, Oxford Brookes University and Coventry University.The subcorpus used for this study includes assessed assignments produced by L1 users of English only.This dataset comprises 1,952 assignments, for a total of 4,849,001 tokens.The average length of the texts is 2,484 words.Henceforth, it will be referred to as L1-BAWE.
Our second dataset is a learner corpus of 264 argumentative essays (totalling 170,183 words) produced as exam texts.The average length of these L2 essays is 645 words.The corpus was compiled in the 2020-21 academic year by the authors of this paper.The texts were written by third-year undergraduates majoring in modern languages at the University of Padua, Italy.Henceforth, it will be referred to as IT-essays.The essays focus on various topics, including Brexit, online shopping, Covid, and language learning.According to course requirements, the students should have achieved a C1 level by the time they wrote the texts included in the corpus.Not all students showed that they had reached this level in their written English, but their essays are included in the corpus nevertheless.Only texts written by students who had granted permission for them to be used for research purposes were included.
Although both corpora exemplify expository, argumentative writing, they differ in a couple of respects.First of all, the texts in L1-BAWE come from a range of academic disciplines and text types (including essays and reports), while the texts written for the IT-essays corpus are argumentative essays on topics of general interest written by students of modern languages.Another difference is the length of the texts (see above).Despite the differences, we expected to find instances of in fact in both corpora for two reasons: in fact is used frequently in academic writing as a cohesive device that signals the logical connection between neighbouring statements, and it is not topic-dependent.

3.2
Classification and Analysis This study adopted a Contrastive Interlanguage Analysis (CIA) approach (Granger, 1996(Granger, , 2015)), comparing L1 and L2 corpora to identify similarities and differences between native speaker and learner English.In conducting this analysis, we were faced with a common methodological question in CIA (see Werner, 2017), that is, whether and to what extent L1 data can serve as a suitable standard model against which to compare learner production.In the case of L1-BAWE, the L1 writers are themselves students, who might not yet have fully mastered all the complexities of academic writing, least of all the nuances in meaning of in fact.This study, then, observed the use of in fact in L1-BAWE, without taking it as a standard model that should, in some way, be emulated by learners.
To respond to the research questions, that is, to investigate the frequency of use of in fact, and to observe how it is used by L1 and L2 writers, we first searched for in fact in the two corpora.A total of 841 concordance lines for in fact were retrieved from L1-BAWE, using Sketch Engine (Kilgarriff et al., 2014).As our aim was to analyse L1 students' use of in fact, we eliminated all cases of in fact that occur in quotations from published works.This left us with 815 cases of in fact (168.08 per million words (pmw)).During a preliminary reading of the concordance lines, we identified the following categories: 1) 'Contrast' (when in fact could be paraphrased as, and thus be replaced by 'but instead'); e.g.
(1) In Milgram's experiment, it was predicted that only one in a thou- sand would continue to the end.In fact, 26 out of 40 subjects continued until the end.2) 'Specification' (when in fact could be paraphrased as, and thus be replaced by 'more accurately' or 'not only … but …'); e.g.
(2) Due to the immense pressure the caregiver is under, elder abuse is not uncommon; in fact physical abuse by trusted caregivers is among the most common type of violent offense experienced by older adults.3) 'Factuality' (when in fact could be paraphrased as, and thus replaced by 'effectively' , 'in actual fact'); e.g.
(3) It also demonstrates that concepts are, in fact, comprehended at different ages.
Downloaded from Brill.com 10/08/2023 03:52:04AM via Open Access.This is an open access article distributed under the terms of the CC BY 4.0 license.https://creativecommons.org/licenses/by/4.0/4) 'Support' (when in fact could be paraphrased as, and thus replaced by such formulae as 'for example, consequently, that is, as to be expected, indeed, sure enough').This 'support' meaning, which signals that what follows confirms previous information, corresponds to the meaning of Italian infatti; e.g.
(4) however, it is laying the foundations for learning a language during the whole first year of its life.In fact, before it even starts to experiment with producing words, it will learn to make eye contact and to interpret gesture, facial expression and tone of voice.5) 'Unclear' (for unclear cases resulting from poor writing, or ambiguous cases that were compatible with more than one interpretation).This category will not be analysed in the present study, but an illustrative example, where in fact could be replaced by instead or effectively, is provided below: (5) The James-Lange Theory said that emotion is in fact the conse- quence of the bodily changes associated with its expression, rather than the cause.The above meanings correspond to those reported in the literature review for in fact (1, 2, and 3) and for infatti (4).4We independently proceeded to classify each instance of in fact.We then compared classifications and discussed cases we had initially disagreed on (final inter-rater agreement: 95.5%).In this study we only discuss the 778 instances of in fact on which we both agreed.
Using the software AntConc (Anthony, 2014), we retrieved a concordance of 77 tokens of in fact from IT-essays.We applied the same classification procedure adopted for L1-BAWE.We agreed on the classification of 74 instances of in fact (final inter-rater agreement: 96.1%) and it is these that we considered in our study.

Results and Discussion
In this section we first report on the range of meanings attested in the corpora (RQ1) and then on their frequency of occurrence (RQ2).As the two corpora are of differing sizes, a normalised frequency of the occurrences of in fact pmw is provided.
Contrastive PragmaticS 4 (2023) 274-288 As outlined in Section 3.2, four meanings of in fact are present in L1-BAWE: 'contrast' , 'specification' , 'factuality' and 'support' , the last of which corresponds to the meaning of Italian infatti.These four meanings are also present in IT-essays, though with differing frequencies.
Figure 1 shows the occurrences of each meaning as a percentage of the total number of occurrences of in fact in L1-BAWE.While there is not a marked difference between the frequency of the first three meanings (26% for 'contrast' , 29% for 'specification' , and 22% for 'factuality'), the frequency of the fourth meaning, 'support' , is strikingly lower.In fact, only 3% of the instances of in fact are used in the meaning of 'support' .
The use of in fact in IT-essays is noticeably different (see Figure 2).In particular, the high use of the meaning 'support' , corresponding to the meaning of Italian infatti, is clearly visible.In fact, 'support' accounts for over 50% of the occurrences of in fact in IT-essays, compared to just 3% in L1-BAWE.Furthermore, 26% of occurrences in L1-BAWE denote contrast, as opposed to 5% in IT-essays; and 29% of occurrences of in fact in L1-BAWE denote 'specification' , as opposed to 7% in IT-essays.Only the frequency of the meaning 'factuality' is not strikingly different (22% of the total number of occurrences of in fact in L1-BAWE, compared to 18% in IT-essays).
This study confirms what has been observed in the literature (Aijmer and Simon-Vandenbergen, 2004;Buysse, 2020;Schwenter and Traugott, 2000), namely that in fact is a multifunctional marker.We identified four meanings, 'contrast' , 'specification' , 'factuality' and 'support' , all of which are used by both the L1 speakers and Italian learners of English.The meanings 'contrast' and 'specification' are those used most frequently by the L1 students, but least frequently by the Italian students.On the other hand, the third meaning, 'factuality' , occurs over twice as often in the L2 corpus as it did in L1 corpus.The fourth meaning, 'support' , occurs 42 times more in the L2 corpus than in the L1 corpus.The presence of this fourth meaning in L1-BAWE was unexpected, as it was not identified in the literature on L1 use of in fact.Its correspondence to the Italian cognate infatti (see Bruti, 1999) led us to presume it would only appear as a false friend in IT-essays and, in our preliminary reading of the L1-BAWE concordance lines, we initially overlooked its presence.Repeated readings, however, revealed that it was, in fact, also used with this function by L1 writers.
We also noticed an overlap in meanings in some instances of in fact.This should not be surprising, considering that borderline cases have been identified in diachronic analyses by other scholars (e.g.Simon-Vandenbergen and Willems, 2011).With RQ2 we intended to identify the frequency of the meanings of in fact in both corpora.In L1-BAWE, the 'contrast' , 'specification ' , 'factuality' and 'support' meanings of in fact occurred 44.13, 48.46, 36.71 and 5.77 times pmw, respectively.We noticed a fairly even distribution of the 'contrast' (26% of the total number of occurrences), 'specification' (29%) and 'factuality' (22%) meanings of in fact, but only a few instances (3%) of the 'support' meaning.In contrast, 'support' was the most frequently occurring meaning of in fact in IT-essays (occurring in 53% of cases, 240.92 pmw), while the other meanings of in fact were less frequently attested (5% for 'contrast' , 7% for 'specification' , and 18% for 'factuality').
Overall, in fact occurs almost three times as frequently in IT-essays (452.45 pmw) as in L1-BAWE (168.28 pmw).In addition, while the three most frequently occurring meanings of in fact appear to be fairly evenly distributed in L1-BAWE, the most frequent meaning in IT-essays corresponds that of the Italian cognate infatti.This is in line with previous research, which suggests that pragmatic markers are challenging to learners, because of their complex semantic attributes or multifunctionality (Chen, 2010: 46).As a result, they may not only be over-or under-used (e.g.Povolná, 2012;Werner, 2017), but also used for a narrower range of functions than is typical of native-speaker discourse (e.g.Fung and Carter, 2007;Haselow, 2021).The Italians students' over-reliance on one particular meaning of the pragmatic marker probably indicates a lack of awareness of its other three meanings, as well as likely L1 interference.

Conclusions
This study has explored the range of meanings instantiated by the pragmatic marker in fact in the academic discourse of English L1 and Italian EFL students.Our findings show that, in both the L1 corpus and the learner corpus considered, in fact conveys the same meanings, namely 'contrast' , 'specification' , 'factuality' and 'support' , but also that there are unclear cases to which it is impossible to assign one specific meaning.In addition, in the L1 corpus, in fact most often conveys the 'contrast' and 'specification' meanings, while in the learner corpus, in fact most often conveys the 'support' meaning.
The 'contrast' , 'specification' and 'factuality' meanings, which are more prominently exemplified in the L1 corpus, are also attested in diachronic studies of in fact, suggesting that present-day use of this marker has retained traces of its semantic evolution, which are more salient to L1 speakers.The 'support' meaning, which mainly characterises the L2 corpus, corresponds to that of Italian infatti, suggesting that interference from Italian is at play.Whether this is indeed the case could be explored in future research by examining the use of infatti (its meaning/s, possible collocations and position in the clause) and checking the extent to which this use matches the patterns of use of in fact in Italian students' writing.Further insights are likely to emerge by exploring possible inter-speaker variation in the use of in fact, that is, by determining the dispersion of the various meanings of in fact across writers and across texts.This would clarify whether and to what extent given meanings -including unclear cases -are a widespread phenomenon in language use or are to be attributed to individual communicative episodes.
The presence of a relatively large minority of cases that are hard to classify in both corpora might be motivated by the students' limited expertise in academic discourse, indicating that the conventions of scholarly communication are not picked up spontaneously, but need focused instruction.Alternatively, it may be that in fact has also started to be used as a generic filler which only signals the progress of one's argumentation more than a specific logical relationship between neighbouring utterances.Whichever the case may be, a fruitful line of research may involve comparing the use of in fact between groups of writers at different levels of expertise, both as L1 as FL users of English.In addition, further insights might emerge from examining both novice and expert non-academic written discourse.
The results of the present study could serve to develop corpus-informed pedagogical materials for advanced English language students of Italian -or other L1 backgrounds -and, indeed, L1 student writers who need to refine their writing skills.Such teaching materials should aim, first, at raising students' awareness of the polysemy of in fact and, secondly, at encouraging students to experiment with its contextual adaptability in their own writing.An initial activity could involve presenting students with contextualised examples of in fact unambiguously illustrating its various meanings.These could be accompanied by glosses, that is, explicit paraphrases, of the functions of in fact.A further activity could present examples of in fact in context and require students to identify its various meanings, by replacing it with one of its possible glosses.This is likely to be especially relevant to Italian EFL students, who may indeed find it hard to appreciate that in fact, unlike infatti, can also serve a contrastive function.We thus fully embrace the view put forward by Seidlhofer (2002: 220), who observes that FL pedagogy has to be local, designed for specific learners and settings.This is because different L1 backgrounds may pose different challenges to FL learners.
We are aware that oversights in the use of in fact in argumentative writing is not likely to lead to serious miscommunication.Misuse of in fact may "only" cause temporary, local misunderstandings, without counting as breach of proper interactional conduct.Yet, at an advanced level of proficiency, accurate use of in fact not only ensures communicative effectiveness, but also reduces decoding efforts, which is a manifestation of reader-friendliness.The cumulative effect of a proper use of pragmatic markers such as in fact is likely to make the reader cognitively and emotionally well-disposed towards the writer, even below their level of consciousness.Therefore, teaching the uninitiated how to use in fact, and other pragmatic markers, is a worthwhile goal.

Figure 1
Figure 1 Tokens of in fact in L1-BAWE as a percentage of the total number of instances.R.d. stands for 'rater disagreement' Downloaded from Brill.com 10/08/2023 03:52:04AM via Open Access.This is an open access article distributed under the terms of the CC BY 4.0 license.https://creativecommons.org/licenses/by/4.0/Contrastive PragmaticS 4 (2023) 274-288 Combining the epistemic function of certainly and the adversative function of however (p.11), it roughly means 'in truth' (p.15).In more recent occurrences, in RQ1) What meanings of in fact do L1 English and Italian students use in academic written discourse?Do these meanings also include the 'support' meaning characteristic of infatti?RQ2) Which meanings of in fact are most frequently used by L1 English and Italian students in academic written discourse?

Table 1
Frequency of occurrence of the meanings of in fact pmw in L1-BAWE and IT-essays (absolute frequencies in brackets); r.d.stands for 'rater disagreement' Downloaded from Brill.com 10/08/2023 03:52:04AM via Open Access.This is an open access article distributed under the terms of the CC BY 4.0 license.https://creativecommons.org/licenses/by/4.0/