RETHINKING QUOTING IN WRITTEN JOURNALISM: AN INTERTEXTUAL CHAIN FROM AN INTERVIEW INTO QUOTATIONS

This paper conceptualises quotations in a journalistic article as an intertextual chain and presents an array of core practices that journalists adopt in terms of quoting practices. The analysis is based on data from stimulated recall sessions with several informant-journalists. The “stimuli” originate from the recordings of journalistic interviews conducted by the informants as well as on the articles based on those interviews. We propose a three-part model of the recontextualisation of quotations : decontextualisation practices focus on selecting a suitable piece of information from the interview, while contextualisation practices influence the positioning of the quoted material in the final article and textualisation practices pertain to deletions, changes and insertions in the quoted material itself.


Introduction
This paper analyses the practice of quoting in written journalism from a media linguistic point of view. 1 As a sub-discipline of applied linguistics, media linguistics focuses on language use in journalistic media. In terms of theory, media linguistics analyses data from media settings to solve research problems raised by linguistics itself (e.g. Perrin 2013: part A). Whereas quoting is a central and frequently used linguistic action in general, the actual quoting practices adopted in some particular domain vary substantially and always serve some specific intent (e.g. Ekström 2006). In this paper, the main focus is to explore the transformational relationship between journalistic interviews and the quotations based on those interviews. I conceptualise this relationship as an intertextual chain, in which texts transform into other texts "in regular and predictable ways" (Fairclough 1992: 130). My main objective is to clarify this "chaining" process in the specific context of quoting in written journalism by presenting the following research question: What are the quoting practices that journalists both use and can recognise from their own work processes? While some research on actual quoting practices has been conducted (e.g. Haapanen in press;Lehrer 1989;Johnson Barella 2005), there is nonetheless a "pressing need" to examine quoting practices more closely (Nylund 2006b: 151).
In terms of practice, media linguistics aims to clarify problems in media practice by using linguistic tools. One such practical dimension in this paper pertains to the striking contradiction between the "official" perception of quoting practices and what actually occurs in real-life. By "official", I refer to media terminology handbooks and journalism guidebooks, which refer to quotations as more or less verbatim repetitions of an original utterance -and while the form needs to be altered slightly, the meaning must be maintained (for example, see Zelizer & Allan 2010: 31;Brooks et al. 2002: 72-86;Goldstein 2009: 232;Kramer & Call 2007: 107-109). However, in practice, the relation between spoken utterances and written quotations is highly complex, as will be demonstrated in the next section.
To explain the quoting practices that journalists consciously adopt, I needed to trace the journalists' mental activities. Such a study of thought processes and intentions has been criticised as mere "introspection" and "too impressionistic to verify processes" (e.g. Smagorinsky 1994: ix-x).
However, most flaws in reliability and validity can be overcome by combining different data and methods (for instance, see Flick et al 2004;Beaufort 2008). A multi-method approach has been successfully demonstrated in the landmark project in the field of media linguistics, the Idée Suisse (for overview, see Perrin in this volume; Perrin 2013), which combines ethnographic observation and verbal inquiry as well as a handful of other methods. While ethnographic observation captures the activity but leaves it to the researchers to interpret the motive for the activity, verbal inquiry allows access to the mental reflections of the people under investigation, but the results might then be affected by their selfawareness. To sum up, varying methods compensate for each other's flaws. These types of methodologically complex research designs have led to promising results in media linguistics (e.g. Vandendaele, De Cuypere & Van Praet 2015).
The research design of this paper that will be introduced in the following section takes advantage of my empirically grounded model of journalistic quoting, which offers not only a holistic but also structured perception of the phenomenon at hand. From this premise, I then exploit a stimulated recall method that is designed to increase the informants' awareness of their actual work practices. Thus, this method offers a more authentic picture of journalistic quoting and the goals steering it than, for example, semi-structured interviews or even more simplistic research designs such as common surveys. My paper is organised as follows: section 2 presents the background of my research framework, and sections 3 and 4 respectively introduce data gathering and data analysis. Section 5 presents the results and section 6 is a concluding discussion on these findings.

Journalistic articles and quotations
In journalism, an umbrella term for texts such as news, profiles and factfocused articles is journalistic article. In terms of family resemblance, they all share some particular characteristics. Firstly, they are produced by journalists, distributed as a printed compilation (although electronic platforms have diversified this situation) and financed by subscription fees, advertising revenues and/or by some interest group. Secondly, journalistic articles can be described in terms of their consumption; by reading them, the audience seeks information, benefit and entertainment (Tammi 2016). Furthermore, the frequent use of quotations is undoubtedly one recognisable characteristic of most written journalistic articles (e.g. Gibson & Hester 2000). In brief, a journalistic article is shaped by a relatively stable set of conventions that is associated with a socially ratified type of activity (Fairclough 1992).
As for the origin of quotations in journalistic articles, quotations are almost solely based on oral interviews. Indeed, a major part of an article content is often based on interviews, although these recontextualised elements are only occasionally manifested as direct or indirect quoting (Haapanen in press 2016b;forthcoming). Therefore, a journalistic interview is clearly (one of) the conventionalized premise(s) for information gathering in journalistic work (Ekström 2006), and it can be regarded as a separate, self-motivated genre: The oral, (most commonly) face-to-face interview is performed in a somewhat conventional way and, despite the variations in execution, it has an explicit purpose -most obviously, to gather information for an article − as well as a fixed structure and predetermined participant roles.
As the fundamental source for the journalistic article and its quotations is the journalistic interview, the relation between the interview and the article is an important research problem. In this study, this relationship is conceptualised as an intertextual chain. This term, coined by Fairclough (1992: 130-133), refers to the transformational relation between texts. During this "chaining", a particular type of text is transformed into another type of text "in regular and predictable ways" (ibid. 130). Norman Fairclough exemplifies this procedure by citing the chain which links press releases with news, or medical consultations with medical records. This paper focuses on these conventional ways that guide the process of formulating quotations in written journalism.
To describe this "chaining" process in detail, I have adopted a theoretical model of journalistic quoting (see Haapanen forthcoming), which is based on the notion of recontextualisation as defined by Per Linell. He defines recontextualisation as "the dynamic transfer-and-transformation of something from one discourse/text-in-context […] to another" (Linell 1998: 154). Following , this model structures the process of "transfer-andtransformation" from an interview to quotations into three functions. It is important to note that this model is a simplification, and regardless of its chronological arrangement as presented in the following paragraphs, the functions are not so clearly defined and organised in reality, as they can overlap and occur simultaneously.
The first function is called decontextualisation. This refers to the process involving the journalist selecting and extracting the segments of the interview discourse s/he decides to exploit as a quotation in the journalistic article. In practice -and contrary to the guidance of manuals -journalists may combine utterances from two or more places in an interview into one single quotation, or merely quote the content of the original text with new wording, such as turning points of the narration.
The second function is contextualisation, which refers to a journalist positioning the selected discourse into the article -often, if not always, into a co-text that is different from the original. During this function, the material to be quoted is monologised: the interactive turn exchange as well as the presence and influence of the journalist in the original oral discourse is predominantly obscured in the article and the quotations are presented as the interviewee's independent, continuous and spontaneous speech (e.g. Haapanen submitted;Ekström 2001;Nylund 2006a).
The third function is textualisation. This function refers to a journalist modifying both textual and linguistic form and meaning. In addition, s/he might make deletions, changes and insertions in the quoted material. However, the quality and quantity of these actions can range from minor revisions to substantial alterations. Moreover, journalists also formulate quotations that do not seem to be based on the journalistic interview they allegedly come from. This paper adopts the three-part model of recontextualisation as a framework for gathering and analysing the data, which will be covered in the next section.

Data and methodology
Determining the answers to my research question required data and methods that capture the practitioners' chain of thought, writing strategies and intentions. A typical option for my data gathering would have been to conduct an inquiry that was based on interviews. However, this approach involves indirect access to mental processes, as the data would be based solely on the informants' explanations regarding what they were thinking and what they were both willing and able to share with the researcher. (Grésillon & Perrin 2014) This challenge materialised in my earlier research (Haapanen 2011) when I conducted a series of semi-structured interviews with experienced journalists and journalism educators inquiring about their perceptions on quoting. The answers I received mostly repeated the idealistic standpoints expressed in journalism guidebooks and textbooks (see also Mitchell & Rosenstiel 2000). However, my own experience as a journalist made me realise the possible discrepancy between what informants claimed they did and what they actually did in practice.
It therefore became apparent to me that adopting only a conventional retrospective protocol, such as a semi-structured interview, would not work. Instead, a retrospective protocol needs to be linked concretely to the parts of the intertextual chain, that is, an interview and (the writing of) an article, to provide a better insight into concrete and actual work practices. Therefore, to overcome these potential problems in validity, I adopted the method of Stimulated Recall (SR) 2 . Traditionally, an SR begins with videotaping a selected person at work, after which the person is asked to view and comment on the video. The method is designed to increase the person's awareness of his/her performance and thus to reconstruct the trains of thought the person had while working. Due to the stimulus (the videotape), the method also prompts informants to "discuss processes and interactions that they otherwise might have neglected" (Smagorinsky 1994: xv) 3 .
In general, SR is a flexible tool for various research frameworks (e.g. DiPardo 1994). In my application of the method, I used the transcript of the recording of the original journalistic interview 4 and the published article as stimuli for the reconstruction of the quoting process. Additionally, two aspects of my research design served to particularly enhance its reliability. Firstly, I guaranteed the anonymity of my informant-journalists, and this encouraged them to speak honestly about a sensitive topic. Secondly, I clearly introduced my own history as a journalist at the beginning of each SR session. Thus, the informant-journalists were aware that I was familiar with the various strategies that journalists used in quoting, and as a consequence, the informants were more likely to comment also on those quoting practices that contradicted the prevalent guidelines. Let us now turn to review the step-by-step procedures of the SR, which begins with two preparatory phases that are followed by the two cycles of the actual SR.

Preparatory phase
Before an actual SR session, two preparatory phases were completed. During the first phase, I requested that sixteen Finnish print media journalists (who had from 3 to 20 years of work experience) record one or two of their journalistic interviews for me. At this point, I did not divulge the exact nature of what I was 2 The SR has been most frequently used in the analysis of learning processes, interpersonal skills and decision-making in the field of educational, medical/clinical and second-language research (for overview, see Lyle 2003), but it has also been applied to media research (e.g. Rautkorpi 2011). 3 see also Dempsey (2010: 350-351). 4 Videotaping might have distorted the original interaction between the journalist and the interviewee. The audio recording, in contrast, worked flawlessly, as it is commonly used in the field of journalism. studying. A rough transcript was then prepared for each of these recordings and these transcripts constitute Stimulus 1.
During the second preparatory phase, I collected the articles based on the aforementioned journalistic interviews. The types of these articles as well as the platforms of publication vary considerably. Indeed, the articles can be described as news articles, fact-focused articles, interview articles and profiles 5 . They were published in newspapers, magazines, customer magazines ("B-to-C magazines") as well as in web-publications. These articles comprised Stimulus 2. I then located the specific passages of text in Stimulus 1 that the quotations in the published articles were based on, and transcribed these passages in detail. Stimulus 1 and 2 are presented in Table 1. The example is an excerpt from an article published in a Finnish business magazine. The topic of the article is the career and company of an interviewee who is an immigrant from China. Both the journalistic interview and the article were originally in Finnish, but they are translated here into English 6 .

Stimulus 1: Transcription of the interview Stimulus 2: Published article
Before the section below, the interviewee and the journalist reflected on the importance of language skills and local education for immigrants.

Interviewee:
although I was so good… I can say that I was a diploma engineer and everything… education and career and… at the peak [of my career] in China that time

Journalist: yeah
Interviewee: but when I came here I am… a zero… At Midsummer of 1994, Wang's life changed completely when she arrived to a deserted Helsinki with her husband. "In China I was a successful diploma engineer 7 , here I was nothing. It was hard to accept."

Journalist:
was it a hard situation to accept Interviewee: yes yes it was In the actual stimulated recall sessions that generated my data, I met each of the informant-journalists individually. I conducted SR sessions with 11 out of the 16 informants because at this point, their answers appeared to be "saturated" in that new aspects no longer seemed to arise. Each of these eleven informants has his/her personal identification number (I-XI), and whenever the excerpts from these SR sessions are mentioned, the informant's number is located at the end of the data excerpt.
The SR session began with enquiry of the informant's biographical information (such as education, work history, and received guidance on quoting, if any) and this was followed by two cycles of close reading of the transcript and the published article. These cycles were designed to reveal the practices of each of the quotation-making functions: decontextualisation, contextualisation as well as textualisation. I shall describe these two cycles below.

First and second cycle of the SR
During the first cycle of the SR, the informant-journalist and I as a researcher examined each of the quotations in the published article separately. The aim of this first cycle of the SR was to determine the practices and motives concerning the first and second function of the process of quoting, namely decontextualisation and contextualisation. In practice, I asked the informant-journalist 8 to explain, firstly, her reason for selecting this particular text segment or content to be quoted, and secondly, why the quotation was positioned in this particular place in the article. Below are excerpts of the journalists' answers regarding (a) decontextualisation and (b) contextualisation in the data excerpts presented in Table 1. a) Researcher: Why did you decide to use quotations in the first place and why did you select this particular segment to be quoted? Informant-journalist: "Because this is a profile story where the interviewee is talking about themselves, of course the story will contain quotes. Here the interviewee presents opinions and also emotional matters, which she was allowed to say herself in the story". (I) b) Researcher: Why was the quotation positioned in this particular place? Informant-journalist: "First of all, what's practical about magazine writing is that you don't have to chronologically follow the course of the interview, as you can construct the storyline while writing the article afterwards. Additionally, the quotes and the body text need to form a kind of natural exchange, so that the text proceeds smoothly". (I) During the second cycle of the SR, the informant-journalist and I closely read the published article and the transcript of the journalistic interview. The main objective of this SR cycle was to determine the practices and intentions for textualisation of the quoted discourse. In practice, the informants were asked the following: How would you describe your process of quoting in this particular case? Why did you edit the quoted material in this way? What influenced the process?
Below, there is an answer (c) describing the quoting practice of textualisation between the data excerpts presented in Table 1 9 . c) Informant-journalist: "The point is that reality is what it all is based on -so that nothing is like coloured. But the fact is that these modifications happen all the time because quotations should be in proper standard Finnishreadable and smooth, not kinda clunky". (I) The answers concerning each of the three functions of the quoting model were handled and analysed separately. All the SR sessions were tape-recorded and, as preparation for analysis, transcribed.

Analysis
The analysis of this research focused on the data which consist of the verbal protocols related to the quoting practices. In this analysis, I applied the two-cycle procedure proposed by Saldaña (2009). The first cycle identifies similar quoting processes, which emerged from the data, and then labels them with a common code. The second cycle identifies the similarities within these codes and organises similar codes into categories. These categories are the quoting practices that my informant-journalists use and can recognise from their work process. The coding and categorisation took place according to each of the three functions of recontextualisation. The analysis was conducted with ATLAS.ti. The two cycles of coding are described below and summarised in Table 2 with illustrative coding  and categorisation examples. Firstly, I demarcated all the individual segments where the informantjournalists somehow comment on their process of quoting. This diverse set of segments was subsequently coded with focused conceptualisations of the content of the segments. In the names of the codes, following Saldaña (2009), I used gerunds ("-ing" words) to help define what was actually occurring in each data segment. The segments with shared features, as those in terms of topic, purpose, goal, or practice, were assigned the same code.
After the data were coded, I categorised and reorganised the codes according to the similarity of the patterns of quoting process to eventually reach a compact number of conceptual categories (pattern coding, according to Saldaña 2009: 152). Since the categories are assigned conceptual names, which are more abstract than those given for the codes, the categories have the power to unite those conceptualisations pertaining to the same phenomenon.
I applied three conditions to enhance the credibility of my research. These conditions, which are elaborated on below, were to maintain the relevance of the analysis and keep its focus on the practices the informant-journalists have actually performed. Furthermore, these conditions were to prevent the analysis from becoming negatively influenced by the informant-journalists' second guessing what the researcher wanted to hear and from renewing the idealistic jargon of verbatim quoting (see Observer's Paradox in Labov 1972).
Condition 1. My analysis only considered those work practices that the informant-journalists actually recognised by themselves. In other words, I did not analyse practices that were not mentioned by the informant-journalists in SR interviews, even though they could have been plausibly inferred from the journalistic interviews, from their published articles, or from the research literature.
Condition 2. I considered only those sections that I was able to "confirm". That is, when the informant-journalist states "I took this filler word out" and points to an expletive that exists in the interview transcript, but not in the article, this phrase in the SR transcript was assigned the process code of <REMOVING EXPLETIVES>. However, I disregarded the sections where the informant-journalist imagined what she might have done in some hypothetical situation (for example, with the informant-journalist: "The demand for faithful word-for-word repetition could be important in those situations where the matter at hand is questionable, as in political matters, when someone denies something or justifies it. But this [= the article under scrutiny in this SR] isn't that kind of story".).  Table 2 comprises the analytical procedure of the data presented in section 3. I first identified and coded segments from the stimulated recall sessions. Then, I categorised these codes into groups of thematically and complementarily related codes. The last row of the table shows all the parallel categories in the same function of the recontextualisation process. I will elaborate on these categories in detail in the next section.

The extract from an SR session in which the informantjournalist (I) comments on her process of quoting in this particular function
Researcher: …why did you select this particular segment to be quoted? Informant-journalist: "Here the interviewee presents opinions and also emotional matters, which she was allowed to say herself in the story". (I)

Other categories (in this function)
The total number of codes combined in the category is in brackets.

Results
This section presents my findings on the quoting practices that the informantjournalists acknowledged. I recognise that the field of journalism is too broad and heterogeneous for any amount of qualitative data to exhaustively cover the quoting practices in written journalism. However, due to the diversity of the informant-journalists included in my study, I suggest that these results can, in fact, be generalised in a broader context.
In the stimulated recall sessions (section 3) as well as the data analysis (section 4), the process of quoting was divided into three functions based on my theoretical propositions (Haapanen forthcoming). Accordingly, I identified nine quoting practices and I will present them in this section classified under the particular functions of quoting in which they occur. These nine practices dynamically create the intertextual chain between the journalistic interviews and the quotations that occur in the journalistic articles.
To obtain an overview of quoting that transcends any individual quoting practices, I attempted to connect the categories axially in order to detect a core category (Saldaña 2009: 163-167). A core category is composed of all the products of the analysis condensed into a few words, and thus will "explain variation as well as the main point made by the data" (Strauss & Corbin 1998: 147). In other words, a core category needs to cover, and have an explanatory relevance for, all the practices revealed in this research, and therefore it needs to be a sufficiently abstract concept so that all the categories can be linked to it. The core category of this research is the overarching objective of <EXECUTING THE OBJECTIVE(S) OF THE ARTICLE OVER THE DEMAND FOR "DIRECTNESS">. In practical words, journalists aim primarily to make a good story with good quotes, not to transform spoken utterances into a written format as verbatim as possible, nor to maintain the exact meaning of these utterances.
It is important that the "objective of the article over the demand for 'directness'" be understood here as constituting an ideal preparation of an article in terms of content as well as the manner of representation. In other words, a journalist receives an assignment describing the task that she must complete, and this task serves as the objective the journalist strives to achieve when searching for sources, selecting interviewees, outlining the text, and finally writing an article. During this process of completing the task, she also formulates quotations by adopting the practices that will be presented in this section. However, it is worth noting that this same task, which will be realised as an article, might also serve other objectives for the other stakeholders in publishing, such as chief editors, media managers, and publishers as well as the target audience. This aspect will be discussed further at the end of this article.

Practices of decontextualisation
The duration of the journalistic interviews comprising my data varied between a few minutes and almost two hours. In each case, only selected parts of the interview are used as direct quotations in the article. These decontextualisation practices were classified into three categories. a) Constructing the persona of the interviewee.
Journalists tend to quote the opinions, insights, viewpoints and other personal perceptions that describe the interviewee's mindset. In addition, utterances that reflect how the interviewee structures her thoughts and her reactions towards a subject matter are likewise quotable, as are utterances that describe the interviewee's delivery and distinctive manner of speaking. Informant-journalist: "I wanted the quote to be that difficult to understand, because I find someone speaking in such a complicated manner terribly interesting." (II) Informant-journalist: "If someone has an interesting or individualistic way of saying things, they are great to replicate in the quotations". (III) Informant-journalist: "I wanted to place that item in to the quote because it was so peculiar". (IV) b) Disclaiming responsibility.
The source of information is indicated by a quotation, or more precisely, the reporting clause. Journalists quote utterances on subject matters that cannot be easily verified, and thus they protect themselves by transferring the responsibility for the factual content to the interviewee. Quotations marks also confirm − or rather aim at creating an illusion − that word choices and other linguistic details in the quotation are originally from the interviewee.
Researcher: If those issues weren't in a direct quotation, might they lack credibility? Informant-journalist: "Exactly, that's because they are expressed in such a unique way". (V) Informant-journalist: "Usually when a start-up entrepreneur speaks about her future and goals, I want to write the statements as direct quotations. That way the entrepreneur is personally responsible for what they said". (VI) Informant-journalist: "Because it's impossible to check this, I can't get that info from anywhere". [The informant-journalist tries to account for why she quoted directly.] (VII) c) Adding plausibility to the article A quotation that is attributed to an expert-interviewee strengthens her presence in the article (which is an intrinsic value in human-centred articles), and this in turn gives further credence to the veracity of the factual content covered in the article. Indeed, informant-journalists generally perceive quotations as having particular significance in relation to the body text of the article, and it therefore makes sense to place important and interesting content into a quotation. Furthermore, any content presented in the form of a quotation gains added value. Informant-journalist: "They were such strong and extreme opinions and I wanted to emphasise that, so I placed them in a quotation. In my opinion, when something is in a quotation it has more weight". (II) Informant-journalist: "It gains more credibility when the person says it themselves. It might be a little milder way to use 'according to Ms X…' or 'in Ms X's opinion…'". (VII)

Practices of contextualisation.
Journalistic articles are not accounts of the course of the journalistic interviews they are based on. Therefore, the decontextualised segments cannot be mechanically transferred to their "right places" in the article. Instead, this contextualisation process requires deliberate and conscious decision-making. a) Constructing the narration A journalistic article is not an account of the course of a journalistic interview, but an independent text entity. For this reason, informant-journalists place the quoted content in the article in a way that best fits and contributes to the storyline.
[Informant-journalist explained the merging of several sections from the interview into one single quotation.] "Here I'm also justifying this with that need for succinctness. Since we talk about that same thing in both sections and combining them doesn't alter either statement, then I think it's okay to do that". (VIII) Informant-journalist: "In a profile article like this, where the whole point is to characterise that person and bring up interesting things about her, then I think it's okay to change the order [of topics covered in the interview] more freely". (II).

b) Pacing the structure
The quotations and the body text need to alternate in a smooth and natural way. Journalists seem to share the common principle that they will not incorporate two quotations in succession without having at least a reporting clause between them. Furthermore, relatively short quotations are preferred and long quotations are avoided, although there are variations on formal preferences among journalists (cf. the first data extract below). Another observation is that the length of the final quotations does not need to correlate with the length of the original stretches of talk on which the quotation is based. In other words, however long the original stretches of text might be, they are truncated into relatively short quotations. Informant-journalist: "Generally I like long quotations, so that you let the interviewee really say and describe her way of thinking". (IX) Informant-journalist: "I'm trying to create an impression that me and the interviewee sort of tell this story together. I say something and the interviewee comments it. This dialogue moves the story forward, and it's also a kind of rhythm thing". (II) Informant-journalist: "That first quote came pretty late. Usually I like to bring in the interviewee a little closer to the beginning of the article". (IX).

Practices of textualisation
Quotations need to meet and fulfil their functions in the narration of the article (see Haapanen 2011). This often requires substantial modifications to the textual and linguistic form of the quoted text. However, it is important to note that when the original discourse fulfils the target that the journalist has set for this particular quotation-in-the-making, it can be quoted in its original linguistic form without modification. a) Standardising the linguistic form.
Quotations are modified into standard language on a routine basis. This requires both deleting the "disfluencies" caused by the on-line nature of spoken language (such as colloquial words, re-starts, self-corrections and expletives) and simplifying clause structures that reflect the oral origin in their fragmental shape.
Informant-journalist: "I don't feel the need to make my interviewees appear stupid in these stories [published in a regular column], so I want to portray them saying grammatically correct things in a comprehensible way". (VI) Informant-journalist: "We've talked in the office about editing quotations and our boss has given guidelines that the same grammatical rules that apply to the body text also apply to the quotations. This means that we don't have to leave colloquialisms into the quote unless leaving them in serves some specific purpose". (VIII). b) Intentionally including vernacular aspects.
Deviations from universal standardisation are acceptable as a "flavour" (II), but these deviations need to serve some specific function in the storyline. Most often they concern a single informal word or phrase and thus they usually characterise the interviewee by indicating his/her original word choice or enunciation.
Informant-journalist: "The story is about a Syrian man, who lives in South-eastern Finland. He used the word 'mie', which showed how well he's assimilated into Finland. So I included it once in the quotation". [Mie is a dialectical variant of the pronoun minä 'I'.] (I) Informant-journalist: "In a TV interview, you can see and hear the whole thing, questions and answers and the way they talk and explain things. All this gets easily left out in print articles, and that's why I think writing the quotation as verbatim as possible is the only way to inject a genuine feel to the story. Though of course some alterations and editing are unavoidable". (III) The aforementioned quoting practices (a) Standardising the linguistic form and (b) Intentionally including vernacular aspects are all-embracing in my data. However, these strategies do not account for a considerable number of modifications. These modifications involve deletions and insertions as well as changes in the order of elements ranging from a suffix or word, to a phrase or a longer stretch of text. It is noteworthy that insertions can also include some linguistic elements that do not exist in the journalistic interview. To explain these modifications, the informant-journalists mentioned the following two practices. c) Clarifying the original message.
Deletions, insertions and other modifications are used to clarify and condense the original message that the interviewee disseminates through quotation (or to be more accurate, the journalists clarify their own interpretation of the message). The following are the informant-journalists' observations on this: Informant-journalist: "You know that people talk in a way that you understand when you speak with them, but then if you write that down, it can no longer be understood by anyone who wasn't present at the time; so you have to write it so that the reader can understand what's being said". (I) Informant-journalist: "The content has to be accurate, but you can modify the text grammatically so that it's more readable and understandable". (VII) Informant-journalist: "I think that when I first wrote that quote, I had a kind of feeling that there might be a risk of misunderstanding or incomprehension. So I added that 'I went to bed', it clarifies the whole thing -and it's probably true". (V). d) Sharpening the function of the quotation.
Essentially, the modifications are made so that the quotations fulfil their function in the storyline. In other words, certain material is selected (decontextualised) from the interview and positioned (contextualised) into the storyline for some specific reason; and the quotation is modified (textualised) in a way that best fulfils this task.
One emphatic quotation includes the Finnish word helvetti, 'hell', as a curse word. However, in the very section on which the quotation was based, the interviewee did not use the word hell or any other curse word − although it should be mentioned that the interviewee cursed profusely during the interview in general. During the SR session, the informant-journalist (V) told the researcher that the function of the insertion was to add "a couple of hammer blows to the end of that quotation". Informant-journalist: "We have really tight space restrictions and at the same time, we have to cover major issues from multiple angles. So that's why I think it's justifiable to make those kinds of changes [= deleting some words and repetitions which are caused by the process-likeness of spoken discourse], since they don't alter the meaning in any way". (VIII) Informant-journalist: "The interview section for this quote was pretty informal and disjointed, more like random chatting, but I had to pack it tightly in to the article. Therefore, all the colloquial meandering had to be taken out, and so did all the incomplete thoughts and ambiguous expressions". (X) Informant-journalist: "This ain't a true direct quotation, more like a five-minute rant which I then condensed into a two-line quote, and I think it's okay to do something like that because that's in line with stories like this and because I had sent the story to her for checking [prior to publication] − which is common practice with these stories − and she was okay with the quote". (VI) Researcher: The body text leading up to the quotation is different from the question to which the quote is given as an answer. I mean, the interpretational context has changed.
Informant-journalist: "That's true. When I wrote the story I slightly altered the theme we discussed in the interview and then I placed this quote under the new theme". (III).

Conclusion
In the spirit of media linguistics, this paper has been guided by both theory and practice while it also claims to add back value to both of them. The main argument of this paper is that it is useful to conceptualise the relation between journalistic interviews and the quotations in journalistic articles as an intertextual chain. With respect to theory, this analysis classified the process of "chaining" into nine practices and presented each practice under the particular function of the process of recontextualisation it serves.
In decontextualisation, journalists aim at (1) constructing the persona of the interviewee, (2) disclaiming the responsibility for the content, and/or (3) adding plausibility to the article. In contextualisation, journalists aim at (4) constructing the narration and (5) pacing the structure. In textualisation, journalists aim at (6) standardising the linguistic form, although they occasionally (7) allow some vernacular aspects that serve a particular purpose in the storyline. Furthermore, journalists aim at (8) clarifying the original message and (9) sharpening the function of the quotation.
In order to acquire a broader view of quoting beyond any single practice, I connected the categories into the core category of <EXECUTING THE OBJECTIVE(S) OF THE ARTICLE OVER THE DEMAND FOR "DIRECTNESS">. This core category covers and has a high explanatory relevance for all the practices revealed in this research. In doing so, it indicates -contrary to the prevalent perception of journalism guidebooks -that quoting in written journalism does not primarily aim either at replicating the quoted text verbatim, or at maintaining its exact meaning. This finding is in line with the observations made about television news production: Both the interviews as well as the editing of the news stories are guided by the reporter's preliminary idea of what the emerging story should and could look like, rather than by the actual outcome of the interviews (Altheide 1974;Clayman 1995;Nylund 2003;2006a;Sand & Helland 1998;Kroon Lundell & Ekström 2010).
The core category <EXECUTING THE OBJECTIVE(S) OF THE ARTICLE OVER THE DEMAND FOR "DIRECTNESS"> also applies more widely to the process of making an article: For instance, the objectives of an article are negotiated before the gathering of information. This gathering, in turn, may lead to a revisit of the objectives; and these revisited objectives sometimes lead to different decisions than the original objectives. Finally, all these back and forth processes also affect the quotations.
Furthermore, the all-embracing core category does not appear out of thin air. One can presume that the fundamental motives for these objectives derive from factors such as the publishers' ideological values and purposes, the financial basis of publications, the needs and interests of the audience and, furthermore, the current journalistic culture and the societal context in which publishing takes place at large (see Kang 2007;Helle & Töyry 2009). To confirm all these assumptions, further research beyond the scope of the data and methods used in this article is required.
Another theoretical outcome of this study concerns a phenomenon that I call monologisation (Haapanen submitted). This notion refers to the common procedure of eliminating the interview situation's interactive turn exchange between an interviewee and a journalist and presenting the quoted interaction in the article as the interviewee's independent, continuous and spontaneous speech. However, this study indicates that journalists may not be aware of this procedure. In the stimulated recall sessions, my informant-journalists did not encapsulate and conceptualise this process of monologisation at all. This is a significant "blind spot" for professional journalists to have and therefore it deserves more profound consideration.
From the perspective of practice, this analysis has demonstrated that the definitions of journalism terminology handbooks and the guidance of guidebooks and textbooks are contradictory to the actual and deliberately executed quoting practices of journalists. Therefore, the dissemination of actual quoting practices is not only useful but also essential for practitioners and newcomers alike. It is also fair to the audience: a question of the truthfulness of quotations and journalism in general is, to a great extent, a question of the transparency of the principles of work practices 10 . The results of my study have already been disseminated in lectures and through participation in public discussion both within the field of journalism as well as among the readers. Furthermore, the stimulated recall sessions themselves raised awareness and created positive experiences among the informants who were involved in my study.
In essence, quotations are a common and effective means for a number of reasons in written journalism. In order to formulate quotations as a journalist and to interpret them as a reader, it is necessary to understand the common objectives adopted in quoting and the circumstances involved in quoting. Furthermore, besides being interesting in their own right, quoting practices are a window for perceiving and better understanding the complex and intrinsically contradictory activity of journalistic writing.