Genre-related language change: Discourse- and corpus-linguistic perspectives on Austrian German 1970–2010

Abstract The motivation and diffusion of language change have been modelled and discussed in frequently conflicting terms, often focused on an isolated set of features rather than more integrative sociolinguistic concerns. We present a case study of language change in Austrian German along a broad range of lexical, syntactic as well as textual features, approached through a corpus based on genres situated in the pertinent fields of news reporting, education and business. Based on our results, we argue that drawing on ‘genre’ as socially-situated, interactive and goal-oriented patterns of language use provides both a conceptual and empirical framework that may help address some of the more prominent issues in modelling language change: as a concept, it provides a frame within which to grasp the social changes driving language change; as empirical focus, it guides data selection and allows us to describe and explicate complex and seemingly contradictory diffusion patterns.


Introduction
This paper 1 takes an integrated discourse-analytic and sociolinguistic perspective at what Weinreich, Labov and Herzog (1968) famously called the "actuation" problem: the motivation or driving forces behind linguistic change and its complex processes of diffusion. Conceptually, we focus on relevant attempts to model the processes and mechanisms of change in Discourse Studies, in particular with respect to integrating the concept of 'genre' along the lines of von Polenz's (1999) seminal work on the history of German. While recent sociolinguistic studies relate language change to individual genres, sets of genres (e.g. within a specific social field) or write German language history as a history of genres (Steger 1998, Linke 2003, we suggest that genres should be seen as integral mediating constructs between social change and language change. We elaborate on what genre-related factors might determine the diffusion of linguistic change and how best to describe this interplay by drawing on Austrian German data spanning 40 years (1970 to 2010). This case study includes quantitative and qualitative analyses of a broad set of linguistic features in written genres from social fields that previous research has shown to be highly significant for language change in terms of impact and as indicators of ongoing change (e.g. von Polenz 1999): mass media, education and business. Our selection of genres in these fields is guided by several considerations: continued existence since at least 1970, established and formally taught, well-researched, publicly accessible, written, fact-rather than opinion-oriented, and planned, formalised text structure. While some of the selection criteria were necessarily also pragmatic due to limited funding, they assured maximum comparability. The genres thus chosen for the study were the Erörterung, i.e. a type of essay written for final school exams in Austria (Matura); Agenturmeldungen, i.e. news agency reports on different topical domains (such as sports, culture, foreign policy, and economy); Zeitungsberichte, i.e. newspaper news reports (from both quality and tabloid papers); and Geschäftsberichte, i.e. annual management reports of internationally established Austrian companies. Incidentally, all four genres are salient in von Polenz's (1999: 5-6, 38, 79, 88) account of German language change in the 20 th century.
Despite required comparability (von Polenz 1999: 370), there remain, of course, notable differences between these genres to be considered. Since our aim, per se, is not to compare the development of specific genres but to relate potential language changes across genres (and social fields), those differences are not only relevant but of key interest to our work. While our scope is limited to written genres, the analytical approach is methodologically broad. We believe that an integrated quantitative and qualitative approach (using corpus linguistics, text linguistics and discourse analysis) allows grasping changes of language use across genres (as one of the features analysed might change in one genre but be constant in another genre).
Ultimately, this makes relating social change to both macro-tendencies (as identified by, e.g., Fairclough 1992) and micro-linguistic indicators on a textual level (Krzyżanowski and Wodak 2008) a more comprehensive and nuanced task.
Documenting broad, quantifiable shifts in the patterns of language use, we suggest 'genre' as a conceptual frame for discussing the forces driving overall linguistic change. We argue that one cannot fully account for such change without acknowledging the role of genres in social structures (Thibault 2003), working as stable patterns of purposeful, goal-oriented interaction, thus linking recognisable changes in language use to social factors, specific audience-related factors, changes in the media and economic landscape, and the institutional constraints of the respective field (Bax 2011).
While discursive change has been demonstrated in specific instances such as the globalisation and consumerisation of media discourse (e.g., Fairclough 2006;Blommaert 2005;Archetti 2004), affecting genres of news reporting (e.g., Cottle 2011;Joye 2009), business genres (e.g., Flowerdew and Wan 2010) and education (e.g., L'Hote 2010) in very specific ways, these phenomena have rarely been related to simultaneous trends in dissimilar domains and genres. For the German-language context, von Polenz's history of German, in particular the third volume (von Polenz 1999) is exceptional in bringing together countless studies, giving rise to an integrated perspective on language change that, on the descriptive side, informs what we argue for conceptually here. In light of this larger perspective, we pose the following research questions in our empirical case study on Austrian German from 1970 to 2010: -Out of a broad range of linguistic features, which have changed most notably?
-Across the range of selected genres situated in the fields of media, education and business, which of the changes are overall, field-specific or genre-specific? -Which of the changes are parallel, delayed, specific or counter-directional? -Which known contextual and text-inherent factors might explain the results?

Discourse, language, and social change
Critical Discourse Studies (CDS) research on social and political change was initiated by Norman Fairclough's seminal study Discourse and Social Change (1992). Although revisited (Fairclough, 2007) and modified with concepts such as "semiosis" (Fairclough et al. 2004), the overall conception of the relation between discourse and social change remains stable in its key premises. As Fairclough puts it, "many of these social changes do not just involve language, but are constituted to a significant extent by changes in language practices" (1992: 6). He identifies the important discursive changes as technologisation, democratisation, conversationalisation, and commodification of discourse, which have since been repeatedly traced in specific national contexts (Krzyżanowski and Wodak 2008;Flowerdew 2011;Flowerdew and Leon 2010;Wodak and Fairclough 2010). 2 While drawing on Fairclough's "minimum conditions" (1992: 8) for a CDS approach to "language and social change", we pursue a multidimensional conceptualisation of the relationships between discursive and social change in which the analysis of text properties is related systematically to social properties of discursive events as instances of social practice. While such relationships may seem self-evident (especially to those inside the respective speech community), the main task of a systematic, discourse-analytic approach is to deconstruct the complex and nonlinear interplay between changes in text, genre, discourse and social practices.
This view coincides with Kress's proposal for a "social theory of language" that describes social structure as "a web of overlapping and cross-cutting structural factors " (1989: 448). Inasmuch as the "kinds of texts produced are effects of and determined by the contingent social practices and meanings of the structures" (Kress 1989: 448)a sociolinguistic condition that manifests in the existence as well as change of genreswe should expect social changes to affect different genres differently as they are situated in different fields and relate to different affordances.
While the question whether the linguistic changes affecting German constitute some form of decline or deterioration has remained popular in the German-language context, linguistic studies on language change remain sparse, and studies on genre change as a rule focus on a singular genre and thus do not link to a broader perspective on language change. Linguistic change in terms of decades rather than centuries has been shown to be strongly (and often immediately) affected by social or technological changes, triggering specific changes from one form to the next. Linguistic innovation, in other words, must be motivated to take effect; the affordances of communication, by contrast, restrict language change in terms of speed and scope, enforcing a degree of consistency within the language community (Trudgill 1972).
Automatic or mechanistic diffusion models have focused on modelling language change in formal patterns, often disregarding social motivation, e.g., in the form of s-shaped curves (Baily 1973: 77), the wave model (Britain 2010: 148), the cascade diffusion model (Chambers and Trudgill 1980: 192), the hierarchical effect (Britain 2002: 623) or the upstream model (Britain 2010: 148), which describes innovations running counter to expected, usually broader developments. In contrast, other studies convincingly show the impact of sociological factors such as peer group-membership, ethnicity and social class on linguistic innovation (e.g., Cheshire et al. 2008). Since Douglas Biber's seminal work on historical change of the English language, changes in style or register have also been conceptually linked to genres (e.g., Biber and Finegan 1989;Biber et al. 2007). The Discourse-Historical Approach (DHA) provides a conceptual frame for explicating the contextual conditions of change, which both constrain and enable change, some working very broadly, others more specific to a field or even genre, some working in parallel and some in opposite directions. While this does not invalidate the other models or challenge the patterns they accurately describe, it may complement them in a useful and productive way.

Genre, discourse and change
Research in Systemic Functional Linguistics and Discourse Studies has extensively discussed the concept of genre (e.g., Bax 2011: 47-57). The topological description of genres uses the notion of types or prototypes, linking phenomena (texts) to patterns or cores which they more or less closely realise and represent (Paltridge 1995: 395). But genres "do not specify the lexicogrammatical resources of word, phrase, clause and so on. Instead, they specify the typical ways in which these are combined and deployed so as to enact the typical semiotic action formations of a given community" (Thibault 2003: 44, italics  Genres mediate between discourse and text as follows: as recurring patterns of language use, associated with socially defined participants, genres realise goal-oriented, purposeful action; they have developed (and change) to fulfil functions within specific contexts; they are located within and constrained by social fields (including institutions and legal provisions); and, finally, genres as patterns of language use manifest linguistic features including register, syntax, textual structure and forms of address as well as material properties. It is these features that we focus on in our linguistic analyses.
Given their status as stable-enough-to-fulfil-their-social-functions patterns of language use, genres are subject to change inasmuch as the social realities in which they are embedded change. We must therefore also address concerns of a historical perspective of genre analysis: genres may decline in use and disappear as the social functions they serve become obsolete or as they are superseded by new genres; or indeed, hybrid genres may develop due to new technological possibilities (such as letters written as e-mail). As many of their (linguistic and material) features change, they may become strikingly dissimilar from historical instances of the same genre. Even as prototypes, genres are determined by structures outside language and thus "have a continuity that is pragmatic rather than linguistic" (Paltridge 1995: 395). With respect to our data, we establish continuity of the text instances with the respective prototype based on criteria for genre variability (Scollon 2000): -structural and thematic progression, -forms of address, meta-communication and intertextuality, and -lexical and syntactical choices.
Variability within and across genres raises important questions about the uniformity of language change, especially with respect to nonlinear or counter-directional changes (see discussion below).
In the German research context, a parallel concept to genre is that of Textsorte (also related are Texttyp, Textmuster and Gattung). In the study of Sprachgeschichte (i.e. the history of the German language, with a historical focus) and Sprachwandel (i.e. language change with a focus on more recent developments), this analogous concept and related aspects such as communicative affordances and situational context have been acknowledged since at least four decades (Schenker 1977, elaborated in Steger 1998, von Polenz 1999. Because such studies either approach language history (or change) as genre history (or change), or focus on a singular genre, and thus do not conceptually and empirically relate genre-specific changes in language use to broader trends, i.e. across genres or fields, a strongly pragmatic history of the German language which theoretically and conceptually integrates genre remains to be written (Linke 2003: 34). A milestone in this development is Peter von Polenz's work on recent changes in German language use.
By including many studies that focus on genre-related changes, von Polenz's wide-ranging survey and discussion of research into language change indicates a sociopragmatic perspective that successfully links social change to communicative affordances termed "neue Textsortenstile", i.e. new genre styles (von Polenz 1999: 3). It is at this level that von Polenz situates changes in language use that have been empirically documented within genres, describing overlaps, contradictions and phase shifts between such changes with respect to individual genres or domains (von Polenz 1999: 5, 370). However, Polenz tends to downplay such changes as weaker forms of language development, such as the spread, continuation, specialisation, differentiation or popularisation of already existing forms rather than innovation; true to his system-centred view of language change, he therefore does not regard them as language change (e.g. von Polenz 1999: 338, 353;or Weinrich 1984: 99, who argues that changes in sentence length should be regarded in terms of a history of style rather than language). In our discourse-oriented terms, however, such massive change in language use does constitute significant change in language as social practice.

Case study: Austrian German
Our study of changes in Austrian German between 1970 and 2010 used a range of linguistic categories to identify changes across three fields, i.e., news reporting, economics and education.
In compiling our corpus we chose salient genres from these fields, assuming that social change would first and foremost affect specific genres (Bax 2011;Hundt and Mair 1999). We take this perspective to be a valuable addition to existing models of language change that emphasise 'outer layers' such as pragmatics, lexicon and orthography, and 'inner layers' such as phonology, morphology and syntax (Nübling et al. 2006: 3). We conducted our case study to test the assumption that patterns of language change occur in specific social fields and genres, often taking different directions at the same time, according to the specifics of the genre (Krzyżanowski and Wodak 2008; see also Wodak 2013; Krzyzanowski 2013).

Austrian German as standard variant
Our corpus comprises Austrian German, a standard variant of German that differs from the German of Switzerland and the Federal Republic of Germany. This is relevant with respect to questions of cultural capital and differentiation insofar as Austrian German is highly significant for Austrian national identity (Muhr et al. 1995;de Cillia and Wodak 2006). Using marked Austrian German, e.g., lexical choices, has become a popular strategy of politicians in speeches and interviews, local and multinational businesses in marketing and advertising, the media and school materials for teaching. It offers a form of prestige or cultural capital in distinction from the German of Germany, much the same way that the variant of English spoken in Scotland does in distinction from that of England (Unger 2013). As Austrian German is well documented in its phonological, morphological, lexical and syntactic specifics (de Cillia and Ransmayr 2014; Muhr and Sellner 2006;Zeman 2009;Muhr and Schrodt 1997;Ammon et al. 2004;Markhardt 2005), we refer to these works in questions of norms and patterns of language use. 3 Our conceptual focus on genre and the empirical focus on specific genres relates to previous research into language change in the German-language context. Focusing on specific social fields, media or genres (e.g., Denkler 2008 on change in the "language of advertising"), social strata or age groups (e.g., Henne 2006) as well as on specific linguistic features (e.g., Zimmer 1997 on lexical change) has proven a practical approach from a diachronic perspective on the genre-discourse connection.
Among such studies, media and particularly newspaper reporting have received much attention since the 1990s. Götzenbrucker (1990) has highlighted the impact of technological changes on language use in newspaper reporting, which is constantly under economic and social pressure: reduced paper size and more restrictive layout guidelines correlate significantly with reduced syntactical complexity and limited syntactical variation. A more integrative approach is exemplified in Schrodt's (1995) critical engagement with discussions on language change.
Among numerous insights into the concurrent language changes taking place in German, Schrodt offers an astute analysis of the interplay between specific features of language change, genre as a framework of understanding language use, and social change.
The reduction of sentence length and complexity has been described as a feature of the conversationalisation of discourse, colloquial speech replacing literature as stylistic model (Fairclough and Mauranen 1997). In German, this development has been linked to the declining use of some forms of the Konjunktiv mood, the genitive suffix -s for nouns (replaced with the analytic form von) as well as lexical changes: the increasing use and morphological assimilation of English loanwords, the declining use of 'elevated language' (Bildungssprache) identified mostly with Latin-, Greek-and, in Austrian German, French-derived words, as well as the increasing use of originally scientific terminology for its connotations, irrespective of its denotation (Roth 2009: 80). This strategy is seen as the pursuit of cultural capital (and hence prestige) as well as justification, occurring in politics, the media, business and student essays (Schrodt 1995: 267).

The data
Our corpus was drawn from three social fields generally seen as prime arenas of language change in recent history (von Polenz 1999). Of these, business is perhaps the most straightforward: wide-ranging developments such as globalisation and short-term effects such as the economic crisis of 2008 have clearly impacted language use in this field. The genre Geschäftsbericht, i.e. 'management report', was selected because it has been extensively studied in many language contexts and provides a high-profile public self-representation of business (Ditlevsen 2002;Ebert 2002). It closely relates to the development of other genres in business, especially in the industrial workplace (von Polenz 1999: 6).
In the field of mass media, the two subfields 'newspaper' and 'news agency' were included precisely because of their proximity (indeed dependence) but different outlook (as businesses, as forms of journalism with different target audiences). In terms of genre, we included Zeitungsbericht, i.e. 'news report', and Agenturmeldung, i.e. 'news agency report', each from five editorial fields in order to better differentiate what is often amalgamated as 'newspaper language' but constitutes differentiated branches of journalism (von Polenz 1999: 88): National and International Politics, Economics, Culture and Arts, and Local News. Massive changes in both technology and public readership make these genres pertinent sites of language change (von Polenz 1999: 79), keeping in mind that national/regional and quality/tabloid media may differ notably with respect to language use (von Polenz 1999: 353).
In the domain of education, the Erörterung, a type of factual argumentative essay, is arguably the longest-standing didactic genre in Austrian secondary schools (von Polenz 1999: 38) and practiced by students throughout their education until their A-levels. Its definition for teaching has not changed in the decades spanned by our study, although the curriculum and approaches to teaching have (from text-centred to problem-based, communicative approaches, Struger and Witschel 2013). Although not a public and professional genre in the same way of the genres described above, students writing such essays for their A-levels have acquired years of expertise; being a highly influential didactic genre, it presents an intriguing opportunity to study language change.
The composite corpus of 332,969 running words of text thus comprises the following subcorpora: management reports by two large industrial companies in Austria for 1970 (labelled  Corpus compilation and preparation had to proceed differently for each subcorpus: The texts for MR1970 were obtained from of the company archives of voestalpine and verbund as paper originals, scanned and run through OCR, then checked for errors and saved as both pdf and txt files. Texts for MR2010 were obtained as pdf files and also converted to txt files. The two businesses were picked as the largest industrial companies in continued existence in Austria. The student essays for both SE1970 and SE2010 were obtained as colour-photocopies of the archived paper originals at three schools (two in Vienna, one in Graz, a convenience sample based on access, preservation of essays in archives, and willingness to cooperate). They were then anonymised and transcribed, keeping all errors and noting any unclear hand-writing but disregarding the teachers' corrections and comments. The newspaper news reports for NR1970 and NR2010 were obtained as paper originals from the National Library archives, scanned and run through OCR, then checked for errors and saved as both pdf and txt files. News agency news reports for NAR1970 and NAR1970 were provided in digital format by the Austrian Press Agency (APA) and stored as txt files. Both newspaper and news agency news reports were random-sampled from the fields of National and International Politics, Economics, Culture and Arts, and Local News.
WordSmith Tools (Scott 2013) were used to retrieve some essential statistics, such as corpus size and words/sentence, for all subcorpora. However, all qualitative analysis presented in the following was carried out using MAXQDA annotation software for tagging and retrieving data by categories.

Categories: Aspects of genre-related language change
In order to identify both continuities and changes in language use across genres, we selected a broad range of criteria. This choice was informed by results of previous research on German language change outlined above, keeping in mind that notable changes in a given category for one genre need not translate to similar changes in that category for other genres: sentence length and complexity may have decreased in newspapers, but that does not mean they did in news agency reports or school essays; loanword usage may be stable in news agency reports, but might have in-or decreased in other genres; reader orientation may have changed dramatically in business genres, but that does not imply it did in news reports. The following categories and criteria therefore combine likely indicators of language change in one or more of the genres studies, not despite but precisely because they may not reveal similar developments across genres.

Address and reader orientation
Forms of address are characteristic of many genres and have been shown to change rapidly, e.g., in business genres (Eisenegger and Wehmeier 2010) and tabloid newspapers (Virchow 2008). They establish and maintain author-reader relationships by means of direct, second or third person address, reader-inclusive 'we' and group labels, as well as metacommunication (e.g., advance organisers) and authors' self-reference (whether by first person, name or metonymically). Reader orientation also relates to more implicit aspects such as tacit knowledge, manifest in the use of register and, more specifically, terminology (Wodak 2011).
For instance, the explication of terminology or background knowledge on the one hand, and the presumptive use of abbreviations on the other hand, were identified. Changes of presupposed knowledge of the target audience should therefore prove reliable indicators of genre change.

Text structure
Text structures have demonstrably developed in conjunction with social change in businessrelated genres (Ebert 2002) as well as media genres (Hess-Lüttich et al. 1996). Structural phenomena considered in the analysis are titles and headings, section length and structure, paragraph length and structure, and the use of metacommunication. Given the importance of text length for other linguistic features, structure must be assumed to be highly affected by changes in the social context determining genre features.

Lexical choices
Lexical choices have been shown to be very susceptible to language change, in particular in news media (Otto 2009, Heyne andVollmer 2016) and business (Messina 2015, Moss 2009).
To obtain more specific results and relate them to genre change, we analyse the use of: jemanden an der Nase herumführen, i.e. 'to lead someone by the nose') and idiomatic phrases with limited or strong collocational patterns (e.g., eine Rede halten, i.e. 'to give a speech'). 4

Syntax
Closely related to text structure, syntax on the sentence level was analysed in terms of sentence length, syntactical complexity and its variation. These features of German language use have been shown to be changing in general and in particular in newspapers (von Polenz 1999: 353-355, 505). In addition to the rudimentary indicator of words/sentence, we used the following numerical scale of syntactic complexity (Götzenbrucker's 1990

Results
The following sections present the most important results of the detailed analysis for each of the four subcorpora. Given the roughly similar size of subcorpora compared to each other, e.g.
MR1970 and MR 2010 (see Table 1 above), all results are given in raw frequencies.
Additionally, percentages are given for diachronic comparisons to indicate change.

Annual management report
While recognisable as a genre by its social functions, legal requirements and participants, the management report is a "genre set" (Nickerson and De Groot 2008: 326) in the sense of comprising various genres integrated in a relatively consistent manner (Frandsen et al. 1997).
In Austria, the annual management report is legally required of all capital companies: It must include last year's accounting report, a situation report and, in the case of stock companies, also a corporate governance report, and must be presented to the Board of Directors for verification (Austrian Commercial Code,(193)(194)(195)(196)(197)(198)(199)(200)221,(277)(278)(279)(280)(281). While large companies are required to make this report public, business practice internationally has moved well beyond this requirement, making the annual report a key genre through which companies address their of the Berlin-Brandenburg Academy of Sciences to assess dominant patterns and deviations. Phrases that could not be verified in a single case (but were evident in a standard phrasing) were judged to be deviations. stakeholders, including existing and potential shareholders, employees, and customers (Nickerson and De Groot 2008: 327). Internationally, relatively stable features of the management report have been found in forms of address, text structure, framing and material design as well as register and information density (Ebert 2002;Ditlevsen 2002).
The management reports in our corpus are of the Austrian voestalpine AG and the Austrian VERBUND AG. To obtain comparable data, we included only sections present in all cases, Allgemeiner Überblick 'general overview', Lagebericht 'situation report', Personal While there was no direct address of the readers in MR1970, we found 29 instances of second person formal address (Sie, Ihnen) in MR2010. These occurred mainly in contexts of text organisation and, less frequently, in letter-like embedded texts.
(1) You will find more details in the Corporate Governance Report on page 212.
(2) The strong brand verbund shall continue to give you, our customers, as well as our employees, orientation and inspire confidence. (MR2010) In addition, various sections such as the 'Report of the Board' in MR2010 explicitly address a specific segment of the general readership: (3) Dear shareholders, especially in times of change, one thing is of greatest importancea clear strategy. (MR2010) In MR1970, these sections begin with immediately stating factual information: (4) The Austrian economy of the year 1970 exhibited the typical features of the late stages of an economic upturn, which incidentally was the strongest since the end of the war. The real growth of the national productsimilar to 1969was nearly 7% (MR1970) Authorship is performed differently in 1970 and 2010: While MR1970 contains references to a collective identity linked to the companies in question, these are limited to several instances (24) of a vague and undifferentiated 'we' that may be taken to include the entire company.
Considering that in 1970 the reports addressed primarily the companies' employees and business partners, this 'we' has to be able to include the employees (reader-inclusive 'we') in detailing the achievements of the year, but it also has to speak for the company to business partners. In contrast, MR2010 shows a more complex authorship 'we', indicating more specific entities alongside vague references to the company collective (for a much higher total of 154 instances, an increase of 43%). We found numerous personal addresses throughout the report, using 'we', 'our company' and constructions like 'together with our employees, customers and stockholders', suggesting a 'we'-group consisting of the Board and the top ranks of management.
(8) We are convinced that renewable energy sources will dominate the future. (11) is a remarkable example of the polyphonic fabric of the 2010 reports: it is the conclusion to a letter ostensibly written by the accounting firm who audited the corporation and the 'we' it employs is that of a collective external to verbund. However, this is clearly not the actual audit report but a statement crafted precisely for inclusion in the management report: Its addressees are not the members of the Board, nor verbund itself, but the readers of the management report.
Some of these statements also contain a third-person self-reference: while MR1970 contained no individual first person reference, MR2010 includes several uses of 'I'. These occur in the context of testimonials given by selected employees invited to speak with authority, but they speak about rather than for the company.  However, MR2010 also includes many instances of explanation and elaboration within the main sections of the reports, using typical phrases like das heißt 'that means' which are absent from MR1970.
(15) If the credit rating or the rating does not meet the required standardsthat is, if no investment grade rating is achieved -, business operations is only done with sufficient guarantees (e.g. pre-payment, bank guarantees, patronage declarations).
In summary, the frequency and degree to which technical terms and abbreviations are elaborated and explained indexes the tacit knowledge expected of the reports' assumed readership as well as the care taken to ensure comprehensibility for that readership.

Text structure
The structure of the reports has changed drastically, incorporating a greater number of subgenres while also losing one: an obituary section. While the reports of 1970 were relatively cohesive texts featuring a single Überblick 'summary' section in advance, the 2010 reports are highly redundant. Materially speaking, one report includes a small booklet attached to the inside of the cover, summarising the report and duplicating the glossary, contact details and editorial.
While we did not analyse the multimodal aspect of the reports in detail, only noting salient aspects such as pointed out in Section 5.1.1 above, this material change is clearly relevant to the change of text structure. The first pages of the same report present 'Highlights' of the full text that follows. At the highest level of text organisation, i.e. the 'genre set', the 2010 reports are structured into several main sections and subsections, including numerous text organisational devices, previews and summaries. Such complex structures accommodate highly selective reading practices corresponding to the heterogeneous target audience in 2010, its distinct segments unlikely to be interested in the report in its entirety.
Explicit intratextual references found in MR1970 are rare (9 instances total) and include advance organisers and recapitulation: In sum, MR2010 shows more frequent and more elaborate intratextual references; however, these are never combined with an authorial 'we' into performatives as in MR1970.

Register
MR2010 shows a 95% increase in use of loanwords of Greek or Latin origin (4,825 instances) compared to MR1970 (2,480 instances), especially verbs and adjectives such as differiert, divisional, akquiriert and präferiert (i.e. 'differ', 'divisional', 'acquire' and 'prefer'). These highly marked lexical choices can be linked to elevated, prestigious style in German, but might also be due to the use of management jargon steeped in English terminology. This change, then, may manifest a strong emphasis on representation in contemporary management reports. Add to this numerous instances of English terminology (972) The following excerpt, detailing the company's ongoing projects, illustrates these lexical differences.

(21) Future investment and acquisition projects are evaluated following the companywide uniform criteria and models. For the individual segments, different capital costs are drawn on, which vary by region. Crucial to the decision are valueoriented criteria. They are intended to allow capital market-oriented yields for the shareholders, increase market value and guarantee a sustainable performance for the company. (MR2010)
In the German original, this short paragraph contains numerous loans and compounds.

Syntax
At the same time, sentence length and syntactical complexity have increased between MR1970 and MR2010. While the length shows a slight increase, complexity has risen more notably.  Example (22) is striking in its simple syntax, showing the typical subject-verb-object constructions in MR1970, while (23) demonstrates the very frequent use of subordinated and coordinated clauses in MR2010. In summary, it seems that by 2010, higher syntactical complexityin conjunction with elevated register and material production valueis used to index the high prestige of the management reports.

German A-level essays
The argumentative essays that Austrian students write for their A-levels in German differ from the other genres included in our study: they do not realise a professional but a didactic genre; they are written on the spot, by hand, and under pressures of both time and assessment.
Deviations from norms and dominant patterns of language use are therefore likely to be more frequent. We included texts of all five grading levels used in Austria, ranging from excellent (1) to failed (5), to represent a normal distribution at an average of (3).

Address and reader orientation
The form of essay included in our study, the German Erörterung, is taught as a factual and impersonal genrewithout any form of direct address to the reader. Any such forms of address would likely be considered errors by grading teachers.
Accordingly, we only found very few cases of direct reader address in either SE1970 (1) or SE2010 (3). While the single instance in 1970 was in formal phrase using third person (and not marked as a mistake), the latter three cases used the polite second person (and were marked as errors).

(24) The kind reader may excuse this. (SE1970)
(25) -as you already know -(SE1970) In addition, we found reader-inclusive uses of 'we' in both corpora, doubling from 3 cases (all in one essay) to 6 (spread through 3 essays), all occurring in directive or generalizing phrases like the following. Due to the didactic nature of the genre and the exam situation in which our data were written, it is not surprising that both corpora included very little explicit handling of jargon in the way of explanation or elaboration. Both SE1970 and SE2010 contained some instances (11 and 14, an increase by 27%) of the phrase das heißt 'that means', announcing an explanation of a previously used word or phrase.

Text structure
The Erörterung is taught with an argumentative text structure in mind, opening with a question, issue or thesis statement, leading through a series of arguments and counter-arguments (usually drawing on a source text and the student's own knowledge) before arriving at a conclusion. In our analysis, we therefore focused on the argumentation laid out in the texts.

Syntax
While average sentence length of the student increased somewhat between 1970 and 2010, syntactic complexity did not. However, the outliers of both values did increase, thus also increasing variance within the sample. This is accompanied by an increase in grammatical norm deviations (with the exception of texts receiving the best grades, see Figure 6). Since the texts in SE1970 and SE2010 were graded, we also tested for correlations between the linguistic features analysed and the teachers' grading. Notably, individual texts' syntactic complexity correlates significantly with the grades the students received, with best and worst graded essays showing the highest average syntactic complexity. We also observed an increase of analytic forms and tense/mood constructions, generally seen as characteristic of informal or spoken German, as well as violations of the prescriptive sequence-of-tense rules of formal written German. In (30), the phrasing of von der eigenen, while grammatically correct, is shunned in formal written German and closely associated with spoken and/or informal language use. In (31), the second clause violates the sequence of tenses required by chronology and presents the incident ruining progress as having happened before one had begun to make progress.

Register
Striking changes can be identified in the student writers' use of register: The number of loanwords used in SE1970 (3,203) has decreased by 49% in SE2010 (1,642). The loanwords are still predominantly Greek-or Latin-derived words characteristic of elevated style, although the number of English-derived loanwords has increased by 1,057% from 14 to 162.
In comparison to 1970, the student essays written in 2010 also show a 28% decrease in the use of phraseologisms from 385 in SE1970 to 277 in SE2010, while deviations from prevalent collocational patterns within these phraseologisms have increased by 153% from 36 to 91. Such deviations include creative extensions to other semantic fieldse.g., from Ziel 'aim' to Grenze 'limit'and changes of word type or syntactic relatione.g., from adjective to verb. While some might be considered deliberate alterations, they would be regarded as errors in the school context. In both of the above examples, the verb component of a fixed phrase was replaced by a close cognate which, however, is not a frequent collocate in the respective phrase. This might be due to interference or mismatching (e.g. the phrase es liegt an jedem or mistranslating from the English phrase 'to give a speech').

Newspaper reports
The two newspapers from which we sampled news reports, Krone and Presse, have been nationwide publications since before 1970 and represent the tabloid and quality press in Austria, allowing an additional perspective on changes in language and genre. 6 While we found similar trends (e.g., an increase of English loanwords), overall language use in the quality paper has come to approximate that of the tabloid paper in terms of several respects.
The interrelated parameters of text length and sentence length as well as syntactic complexity were significantly lower for the tabloid than the quality paper in 1970 (by 49% and 51% respectively). This difference was most pronounced in Local News, Culture and Arts, and International Politics and least in Economics, Sports and National Politics. While the text length in the tabloid has not changed significantly in 2010, the quality newspaper's articles have lost an average of 25% in length. This trend is resisted by articles in Culture and Arts in the quality paper, which remain nearly as long as in 1970.

Address and reader orientation
While direct address it is not a staple of newspaper reporting, i.e. in factual genres, but more common in op/eds, columns or glosses, previous work has shown that tabloids may directly address their readers, for instance through headlines or inclusive phrasings (e.g. Virchow 2008). 6 In the Austrian population of approximately 8.5 million, the Krone has approximately 3 million readers, the Presse 350,000 (Media-Analyse 2015). The former is known for its crude and sensationalist, politically outspoken journalism; the latter maintains a conservative outlook and high journalistic standards.
Not counting quoted speech, e.g. interviews, NR1970 contains a total of 6 instances of an inclusive or generalised 'we', while NR2010 includes double that number at 12. Similarly, authorship is not commonly made explicit in news reporting apart from the name or shorthand of the journalist(s) given below the textthese were not included in the analysis.
However, references to the newspaper itself or the desk or editorial team (Redaktion) were found in NR1970 (18) and, increased by 50%, in NR2010 (42). These occurred in the main body of the texts.
(36) The administrative police of Bregenz, too, confirmed to the "Presse" that the activity had been neither registered nor permitted, it should "simply not have been allowed to happen". (NR2010) (37) That is how diverse the coalition's reactions were to a report in the "Presse" this Tuesday. (NR2010) The newspaper news reports seem to avoid jargon rather than explain it, but both NR1970 and NR2010 contain several instances of explications of subject-specific terms or background knowledge, usually provided in parentheses.
(38) Dörflinger "is worried" that many in Styria (=Austria) "still see themselves as the center of the Earth" and thus overlook that "highly educated elites have long been at work" in the former Eastern Block. (NR2010)

(39) […]"Pandora's Box" (the title of Wedekind's second drama) […] (NR2010)
Example (38) is notable for its unusual form, presenting a partly nonverbal equation in brackets, and is moreover factually inaccurate: Styria is part of Austria, but Styria does not equal Austria.
It is entirely unclear why the author may felt it necessary or helpful to provide this clue to the readers of an Austrian newspaper. (39), in contrast, seems more traditional in providing background knowledge on Wedekind's second Lulu drama; however, the explanation undercuts the play on words which seems to have originally motivated the use of 'Pandora's Box'.

Text structure
The structure of news reports in our corpus has changed considerably between 1970 and 2010, as is evident from the increased use of subheadings and crossheadings (averages per text are up 350% from 0.4 to 1.8), a 34% decrease in paragraph length (average words per paragraph are down from 97 to 64) and strongly increased use of cross-referencing within the text, to other texts in the same edition, or to earlier reporting in the same paper (0 to 28 instances).
(40) However, it is criticised that the healthcare support money will not be increased next year (see report above). (NR2010) It must be noted that some of these trends affect the tabloid and quality paper differently, as we will exemplify with respect to syntax.

Syntax
In 1970, the average sentence length in the quality newspaper was 8 words above that of the tabloid (43%), corresponding to a similar difference in text length. This was most pronounced in Local News, Culture and Arts, and International Politics and least in Economics, Sports, and National Politics. Syntactic complexity yields an even stronger measure of the two newspapers' difference in 1970, with sentences in Krone only half as complex as in Presse (see Figure 7). Articles in Local News, International Politics, and Culture and Arts provide striking differences (see Table   1). In 2010, however, average syntactic complexity in the quality paper is almost at the same level as in the tabloid, showing a 34% decrease from 4.7 to 3.1 (see Figure 7). Syntactic complexity in Local News, National Politics, Sports, and Economics is virtually indistinguishable between quality and tabloid paper; as expected, articles in Culture and Arts remain complex and, to a lesser degree, so do those in International Politics (see Table 1).
Figure 7: Words/sentence and degree of syntactic complexity in Presse und Krone

Register
Latin-and Greek-based loanword usage was arguably the most striking linguistic difference between the two newspapers in 1970: the tabloid contained only a third (652)   In summary, the quality newspaper has lost many of its linguistic distinctions, showing a trend towards simplification that affects both newspapers but affects the quality paper more strongly. In terms of genre, the reports in both newspapers indicate several trendsa striking decrease in loanword usage and elaborate register as well as a decrease of syntactical complexitywhich are, however, offset by contrary developments in Culture and Arts.

News agency reports
As the main press agency in Austria, APA explicitly embraces its role as a Leitmedium, a medium followed by most others, and recognises its responsibility with respect to language use. 7 This is evidenced by its explicit stance on language policy (e.g., regarding Austrian German) and its guidelines for language use for its authors and editors. APA compiles reports and authors original reports. Our analysis includes original APA reports in Economics, Culture and Arts, National Politics, and Local News.

Address and reader orientation
The news agency news reports contained in our corpora for 1970 and 2010 strictly eschew all direct address to their readers in the form of second person Sie 'you' as well as inclusive Wir if not also amiss in judging the GDP to be a less known abbreviation than Wifo and HIS (two Austrian economic research institutes). In summary, NAR2010 predominantly shows news agency reports that are careful not to presuppose much background knowledge of institutions, technical terms and abbreviations.

Text structure
Average text length has increased by approximately 29% between 1970 (195 words/text) and 2010 (252 words/text). More significantly, the average number of sentences per text has increased from 8 to 13 (63%), while average sentence length has decreased from 24 to 19 words per sentence (21%). In short, APA news reports are now more strongly segmented into smaller syntactic units of information. In 1970, the reports in all four fields show similar text length (maximum variance of 15%); in 2010, however, the reports in Culture and Arts run to more than 200% of the other texts.

Syntax
Average sentence length in 1970 is highest in Culture and Arts as well as National Politics, approximating 26 words/sentence. In 2010, sentence length has decreased in every field, most clearly in National Politics (by 36% from 26 to 18) and least in Culture and Arts (by 13% from 26 to 23). The segmentation of information indicated by Figure 9 is indicative of a general conversationalisation and economisation of discourse, and of a substantial change in the genre (possibly news reporting in general). As it is accompanied by a decrease in syntactic complexity, complex content-relations are more commonly broken up into simpler and smaller units. of Austrian German (usually on the lexical level) have also increased, including the diminutive suffixes -erl and -dl, phonological contraction in the prefix ge-to g-, as well as lexical items such as Bub and basteln (i.e. 'lad' and 'tinker').

Discussion: Language and genre change
Our analyses reveal a complex picture of how language use has developed in several of the linguistic categories investigated: some changes extend across genres and fields, such as the decline of loanwords in newspaper reporting and school essays; other changes appear to be specific to individual fields and even genres, such as the complex text structure in management reports; and yet other trends point in opposite directions, such as decreased syntactic complexity in news agency reports vs. increased syntactic complexity in management reports. It is also noteworthy that closely connected newspaper reports and news agency reports developed partly in parallel (reduced syntactic complexity) and partly diverged (stable use of loanwords in news agency reports). Language change on the level of language use, if viewed through a genre lens, is thus anything but uniform. Moreover, we have seen that changes observed for one or some genres of a field (such as news reporting) cannot be assumed to be true for all related genres (as evidenced by the Arts and Culture section). Ultimately, this casts some serious doubt on whether trends like conversationalisation can be assumed without factoring in genre as a factor and whether such trends affect all aspects of language use even within a genre.
Language change, when observed on such a detailed level and a relatively short timescale, we therefore contend, is complex, multi-layered and asynchronous, and sometimes follows seemingly opposite trends. Inasmuch as genres are shaped by social contexts because they fulfil social functions, they are more immediately (and locally) affected by social change than the language use patterns of an entire speech community.
The genre of annual management report in our case study showed strong contextdependent change. In 1970, the voest and verbund were state-owned and union-dominated enterprises, facilitating a strong spirit of community and solidarity among employees (Abel 1995;Köstler 2002). While legal provisions required a minimal technical and financial account of business proceedings, the reports manifested the communal spirit of achievement and struggle, addressing the public in an almost perfunctory way while trusting in the shared technical knowledge of the workforce. By 2010, under pressures of globalisation and neoliberal economy, both companies had been privatised and taken to the stock market, their business activities diversifying and becoming increasingly international. The legal requirements meanwhile have become more detailed and the workforce less closely knit.
The disparate groups addressed in the annual management reports of 2010 manifest in the proliferation of forms of address, subtexts, and embedded genres. The loss of the previously internal focus, re-oriented towards a composite readership and purpose (banks and investors, end consumers), shows clearly in the reports' changed register. Where formerly evident in the use of technical terminology that implied shared knowledge (and prestige within the group) and an otherwise simple register and syntax, it now manifests in explanations of the technical terminology used, elevated register and complex syntax. We thus see larger, even global forces working through the economic, legal and social fields in the Austrian context in changing language use within the genre of management reports.
Regarding newspapers, news reporting showed stable genres overall and changes in language use inflected by a) the type of newspaper (quality vs. tabloid) and b) the field of reporting (Culture and Arts vs. Local News, International Politics, National Politics, Sports, and Economics). These changes are driven by several large-scale developments affecting newspapers globally, i.e., pressure from online media and the changed habits of computermediated communication, changing readership demographics and reception behaviours as well as specific economic conditions within the media landscape of Austria (Puppis et al. 2013).
Overall, this has led to a striking simplification of the lexico-grammatical characteristics of the quality newspaper, the tabloid remaining constant in many respects. Both seem to be affected, however, by the decreased use of loanwords and elevated stylewith the notable exception of the Culture and Arts section. Here, journalists have maintained a highly formal register and complex syntax, presumably catering to a specific readership. Thus, topic and readership seem to be significant with respect to genre and language change within newspapers, counteracting wider tendencies.
This contrasts sharply with the news agency reports in several respects: #numbers.
While these are largely produced for clients in the press and thus under similar pressures in terms of media reception, the Austria Press Agency follows a self-defined policy with respect to Austrian German and language use. Its reports have thus followed the general trend of highly simplified syntax but have maintained an elevated register. However, the tendency observed in the newspapers is also reflected in the news agency reports: Culture and Arts items in 2010 are not only dense with loanwords but also use complex sentence structures.
Unlike the other genres in our case study, the student argumentative essay is produced spontaneously by non-professional writers and taught as well as graded within the institutional setting of school. While the genre and its conventions have not changed between 1970 and 2010, the curricular framework within which it is taught has. Educational reforms in Austria now follow competences-based didactics, foregrounding the production of genres taken to be relevant for the students' future and backgrounding the previously emphasised reading, analysis, and discussion of classical German-language literature (see Saxalber and Wintersteiner 2012;Struger and Witschel 2013). This corresponds to our observation of a shift from literary to colloquial register as well as a decrease of elaborate forms and improvisations of loosely interpreted phraseologisms.
Thus, the genres we investigated, if understood as "social structurings of occasions of linguistic (inter)action" (Kress 1989: 450), provide empirical access to social fields as "domain[s] of struggle" (Kress 1989: 457) over meaning and language use. Such a conceptualisation allows us to see how broad trends such as conversationalisation may be inflected, halted or even reversed within specific genres. In conclusion, we emphasise the importance of genre as conceptual and analytical framework in studying language change in discourse studies and sociolinguistics, which goes well beyond studying genre change itself.
This view, nascent in many of von Polenz's (1999) arguments, is supported by the results of our case study on Austrian German. Due to its limited scope, this approach must be tested in larger corpora and further genres; spoken language should be necessarily included.