2.1Software

Software is not a traditional research subject in linguistics. Generally, linguists group it with the information sciences or see it as a tool for scientific work or as an infrastructural element of linguistic research (e. g., in corpus linguistics or the digital humanities). There is immense difficulty in defining what software is at all (cf. for example Mackenzie 2006 and Chun 2011). Typically, reference is made to the difference between hardware and software, but even the question of the relationship between the source code of a program, an executable file, and a program when it is running touches on complex questions about the semiotic status of software, about its materiality and historicity, that cannot be answered within a single discipline. Software’s code basis can be characterized as

“a labile, shifting nexus of relations, forms and practices. It regards software formally as a set of permutable distributions of agency between people, machines and contemporary symbolic environments carried as Code” (Mackenzie 2006, 19).

Cramer (2008, 173) determines that “software as a whole is not only the ‘code’ but a symbolic form involving cultural practices of its employment and appropriation”. For this reason, a linguistic treatment of software as a symbolic form may not rest at a reference to the particular formal, semantic, and pragmatic characteristics of programming language code by means of which program functionally is algorithmically guaranteed. It is not until the premise of the assumed linguistic neutrality of software is abandoned and the question is posed of its fundamental inseparability from verbal language that it is possible to understand software in its complex linguistic, cultural, and social character. Here, a contextualization of linguistic programming constructs in the field of technology-related languages for special purposes (cf. Wichter 1998) is only one possible linguistic approach.

One point of reference for the treatment of these aspects of software could be the recently developed discipline of software studies (cf. Manovich 2013; Fuller 2003a; Fuller 2008; Mackenzie 2006; and the journal Computational Culture. a journal of software studies at <http://computationalculture.net>). This consists of a set of theoretical and methodological approaches; however, the linguistic form of software has not frequently been dealt with in this field thus far (cf. for example Cramer 2008; Tedre/Eglash 2008; Galloway 2006; Mackenzie 2008; Seiler 2013a and 2015).

2.2Localization

The discourse on localization and the corresponding terminology has its origin in industry (the language industry). Any scientific treatment of software localization will be confronted with this origin and faced with the question of whether and how it can be dealt with in translation studies and applied linguistics (a question that is not always clearly formulated). On the website of the Globalization and Localization Association (GALA) (<http://gala-global.org>), the understanding of localization (also known by the numeronym l10n, as 10 is the number of letters between the first and final letters in the word “localization”) in the language industry is formulated as “the process of adapting a product or content to a specific locale or market”. A locale as a basic unit for localization is understood as a country or region and its corresponding language, and in a broader sense, as the whole of its particular cultural characteristics (e. g., French France vs. French Canada; cf. 2.3 for a discussion of the ideological content of this perspective). In industrial discourse, localization is part of a comprehensive process of generation and international sale of software, its documentation, and its online help, in which the quest for maximum profitability goes hand in hand with the highest possible degree of automatization (cf. Esselink 2000, for example). Here, translators are relatively weak actors in a work process dictated by others (e. g., producers, developers, and project managers). The focus is not on the translation, but on the product to be commercialized. The concept of internationalization, related to the concept of localization (and often represented analogously with l10n as i10n), refers to a procedure in which a product is designed in such a way that it will work with different languages and cultural conventions without requiring that the base code be redesigned (for example, the size of program windows will not conflict with the language-dependent length of character strings). Internationalization also plays a role on the level of documentation development. Here, it is important to establish that localization is a part of marketing strategies shaped for products, and not primarily for texts as language constructs (cf. the work in Reineke/Schmitz 2005, for example).

By contrast, in the discourse in translation studies, where software localization has become a firmly established subject, localization refers to all activities connected with the translation of software texts and the specific challenges with which translators are faced, which are closely tied to the particular features of the text type. Translation studies continues to deal with the role of translator activity in localization projects (cf. Dunne 2006) and occasionally dedicates itself to the social context of localization, among other things (cf. Cronin 2003 and 2013, for example). In doing so, it focuses on the tension between the refined translation skills conveyed in the academic arena and the very limited opportunities to implement these skills in industrial localization processes. Therefore, it cannot help but raise questions about the limits of technology in carrying out translation.

The question of whether the translation of software texts should be referred to as “translation” or “localization” has not yet been clearly answered. Mazur (2007) asks about the possibility of a new discipline called localization studies, the contribution of which is a testament to the dilemma resulting from an insufficient differentiation between scientific and industrial discourse on localization. Cf. Parra (2000), Pym (2004), and Dunne (2013) for further discussion and differentiation of terminology.

One special case in software localization is the translation of web pages, sometimes referred to as “website localization” (cf. Sandrini 2008 and Jiménez-Crespo 2013). This is appropriate in light of the fact that webpages are now frequently generated by web applications. Websites are particularly complex because linguistic decisions are often written into the software of such web applications, and linguistic material generated both automatically by software and by real users is integrated on the textual level (screen) (cf. Eisenlauer 2013).

2.3Software localization as a subject in linguistics

Academic studies on localization dedicate themselves primarily to technical questions or to the complex of educational content in translation programs of study. Investigations of the relationship between language and culture rooted in culture studies that venture beyond questions of how to format dates, numbers, and units of measure remain marginal (on this, cf. Bouffard/Caignon 2006, 807). The preferred point of departure from a place of linguistic interest in software localization is the question of the design of multilingual communication in the digital sphere, as well as that of the role of software in a linguistic globalization dynamic in which English plays a major part on one hand, and in which other languages are nevertheless of great importance on the other hand; e. g., in marketing processes. Moreover, software localization also presents a challenge with respect to linguistic theory that centers on the way in which language functions and on the constitution of linguistic meaning in software in a socio-technical context, as well as on alternatives to the industrial model of the language industry, which shapes its use of language resources in the realm of software according to economic profitability criteria.

In the following, several facets of software localization that are of linguistic relevance will be discussed. These have already been researched and provide additional points of departure. Here, three research categories can be determined: textuality and multimodality, sociolinguistics and language policy, and linguistic ideology.

Textuality and multimodality: Today, practically all media texts exhibit a software layer. Approaches in scripturality or written communication research exist that seek to consider the textual level of software, especially the level of the programming language (cf. Raible 1999 and Herbst 2004, for example). However, again and again, software finds itself at the center of non-linguistic debates that discuss the question of whether it represents a text or a technical artifact. This debate is tied to important legal decisions on whether software falls under copyright law or patent law (cf. Seiler 2013a, 378).

Linguistics concentrates on two levels of textuality in software: that of the source code, and that of the user interface. The source code, created in accordance with the rules of programming languages, contains algorithms that determine the functionality of the program. Beyond the programming language commands, source codes are also given elements of verbal language that include single-language character strings meant to guarantee interaction with the user (for other elements in source code that are not of the relevant programming language, cf. Seiler 2013a, 382).

A program that can be run is generated from source code (made to be machine readable or interpreted by special software) and the user can interact directly with this program. In many cases, this interaction runs via graphical user interfaces (GUI). These exhibit a different form of textuality than the base source code due to their multimodality. In them, verbal elements are occasionally combined with visual and acoustic elements. Here, the different semiotic resources are integrated in a process of meaning compression (cf. Baldry/Thibault 2006 and Cañuelo Sarrión/Seiler 2008, 50). These are texts with hypertextual and hypermedial characteristics, as they are constructed nonlinearly and enable different read and interaction pathways. Beyond studies about websites created by web applications, investigations of the textual composition of GUIs have hardly played a role in media linguistics thus far. But here, too, questions of hypertextuality and interactivity typically play a larger role than the character of software-generated interfaces.

In the process of localization, single-language elements selected for translation are extracted from the source code and gathered in language files that are then used by translators. Thus, translators do not work with the texts themselves, but rather with such extracted chains of strings, using them to create new language files that can then be selected and integrated into the program. It is clear that this kind of decontextualization of linguistic material is problematic in the translation process.

Sociolinguistics and language policy: Beyond the semiotic characteristic of the text, software exhibits a discursive side through which new forms of linguistic performativity have been developed throughout the history of media. Due to its linguistic specifications and communicative potential, software is a medium of the linguistic socialization of its users, and of social control (cf., e. g., O’Hagan 2011; Mackenzie 2006). Software is a complex conflict area in which, among others, language, identity, and economic interests collide (cf. Kelly-Holmes 2013). Multilingualism in the digital sphere is at the center of a sociolinguistic interest in software localization in that this type of multilingualism is both partly constructed by software and filtered by it. Localization is an important way to preserve and promote multilingualism in the digital sphere. The question of which language to translate software into involves economic interests as well as those of the state and civil society. Further, this question is always of sociolinguistic relevance and a part of linguistic political strategies. Griffiths et al. (1994, 10s.) examine the field of language acquisition and determine which advantages localization has on the levels of society, education policy, and economics. Djité (2008) compares the societal significance of software localization for African languages with that of Christian missions.

Language ideology: Due to the general lack of visibility and comprehensibility of internal software processes for the average user, many users have the tendency to see their current constellation as unchangeable and to use default settings even in the presence of flexible user preferences. Software specifications and recommendations have significant ideological potential in that our

“interactions with software have disciplined us, created certain expectations about cause and effect, offered us pleasure and power that we believe should be transferable elsewhere. The notion of software has crept into our critical vocabulary in mostly uninterrogated ways” (Chun 2004, 47).

The relevance of software and software localization to ideology is visible on several levels. In terms of textuality, the ideological functions of interfaces should be noted here; via metaphoric processes, among others, these suggest ideas about how computers work (cf. the desktop metaphor, for example; on this, Fuller 2003b, among others). Furthermore, a silent identification of software with hardware promoted through trade chains when they offer computers with pre-installed operating systems without listing the price of hardware and software separately is also ideological. When problems arise, this can lead users to consider a piece of hardware broken even when the issue is with the software alone. Furthermore, we see a differentiation of operating system cultures, especially in the realms of Windows, Mac, Linux, and mobile devices, which are connected with social, political, and economic strategies and are, or can be, extremely ideologically influenced in that they each generate different forms of sociality, lifestyle, and both instrumental and affective interaction with technology (cf. Seiler 2013a, 387). The corresponding monolingual discourses on software are accessible to models of discourse analysis.

The central concept of the locale is relevant to ideology (cf. 2.2), first because decisions must be made here about the often hotly debated denominations of languages and varieties, and also in those cases where the locale is understood as a simple correlation between language and country that reproduces a territorializing, national-philological perspective in the digital sphere. In a broader sense, Dunne (2013, 5924) defines the locale as “local market requirements”. Thus, the aim of culturally adapting software promoted in localization literature is shaped by a market discourse that evinces an extremely reduced understanding of language and communication, and consequently leads to the idea that it is impossible to differentiate between market and culture. Budin’s definition is similarly industry-based:

“A ‘locale’ is a virtual rather than physical location, where a group of people share certain cultural and linguistic conventions in a consistent way so that the localization industry is able to identify the locale and distinguish it from other, maybe neighboring, locales” (Budin 22006, 290).

The business-motived essentialization of language-country links categorizes web users and can make it more difficult, for example, to display websites in languages other than French to a user in France, “and thus curtails and even eradicates ‘unwanted’ linguistic practices in the form of autochthonous and allochthonous alternatives, encouraging instead uniformization and conformity with the equation of national group with language” (Kelly-Holmes 2013, 138s.). What Kelly-Holmes establishes in the special case of website localization is also true for the linguistic design of software in general, which through localization becomes a “space of multiple languages, rather than a space of multilingualism” (ibid., 144).

The development of more advanced high-level programming languages and especially their success in online technologies has led to a certain democratization of programming and, subsequently, of translation This development is most clearly expressed in the field of free and open source software (FOSS), the development of which is generally community-based and carried out by volunteers, frequently laypersons, and therefore takes other paths than those in industry. However, for quite some time now, hybrid forms of software production have also existed, in that FOSS has come to be used more and more in the corporate world, and in that large software companies, for example, also pay FOSS developers or use the unpaid labor of volunteers via crowdsourcing (McDonough Dolmaya 2011; cf. also ↗21 Crowdsourcing Translation). A central factor is the free insight into, modifiability of, and transferability of software that is assured by different (more or less industry-friendly) licenses, which is much more comprehensive that the mere fact that it is available free of charge (<https://gnu.org/philosophy/free-sw>). The free availability of FOSS source code and the openness of file formats also imply free availability for translations. In this way, it is different from proprietary software, the source text of which is kept secret. Thus, the localization of FOSS represents another to purely market-oriented access to localization in that it allows, among other things, for the decision to localize in less common languages as well – languages that would not be profitable according to market considerations. For linguistics, the freedom and openness of software means free access to source code and to software development processes, including the negotiation of single-language decisions.

FOSS now plays an increasingly important role in translator education. Among other sources, information on this is available from Díaz Fouces/García González (2009), in the edition 2011:9 (Programari lliure i traducció) of the Revista Tradumàtica, as well as in Sandrini (2011), which includes a description of the project initiated by Peter Sandrini, Tuxtrans. Linux for translators (<http://www.uibk.ac.at/tuxtrans/), a collection of free translation tools.

3.1General remarks

The marginality of the linguistic treatment of software and its localization is also visible in work on multilingualism in Romance-speaking world. Here, we only see initial research-based approaches, and no overview of the language conditions in the field of software. Up to this point, linguistics research has largely depended on the evaluation of literature from translation studies in the context of specific linguistic research interests. Beyond this, we only have what we can glean from the data provided by software producers, public debate, and the discourse of all relevant actors. One useful bibliographical instrument for software localization is BITRA. Bibliografía de interpretación y traducción (<http://aplicacionesua.cpd.ua.es/tra_int/usu/buscar.asp>).

Due to the great heterogeneity of the Romance-speaking world, questions of software localization in different areas are relevant. For example, from a political perspective in Europe, a central concern is to preserve Europe’s language diversity, including in the realm of technology. This is the goal of Meta-Net, a network of European research institutes with the objective of constructing a technological basis for the multilingual European information society (<http://www.meta-net.eu>). In developing countries, by contrast, the foremost questions are those of how to overcome the digital divide, which is always reproduced when digital technologies are only available in the widely used languages of the former colonial powers. Here, the software in a native language also acquires a special significance for writing, literacy, and education. For the speakers of less frequently used, marginalized, or dominated languages, the fact that the computer “speaks one’s own language” is certainly of great symbolic value.

In general, it can be said that much is being localized into French, Spanish, and Italian. Different operating systems and the important applications used by the masses are available in all three languages; the numerous contact languages in the Romance-speaking realm paint a much more differentiated picture.

In August 2015, according to the Microsoft Language Portal, Windows 10 was localized into 116 languages and language varieties (<https://www.microsoft.com/Language/en-US/Terminology.aspx>). In the Romance-speaking sphere, this includes Catalan, French, French (Canada), Galician, Italian, Portuguese (Brazil), Portuguese (Portugal), Romanian, Romansch, Spanish, Spanish (Mexico), and Valencian.

The language settings of Apple’s OS X Yosemite supports the following Romance languages: French, Spanish, Spanish (Mexico), Italian, Portuguese (Brazil), Portuguese (Portugal), Catalan, and Romanian. The following languages and varieties are partially supported: French (Canada), French (Switzerland), Galician, Haitian Creole, Corsican, Neapolitan, Occitan, Sicilian, Spanish (Latin America), and Walloon. Latin is also represented here.

Take, as a selected example of FOSS, Debian, a very prominent Linux distribution. Debian tracks multiple statistics on the state of translations. These also include the translations of Debian’s websites, documentation, and the installer (currently available in 86 languages), which can be seen at <https://www.debian.org/international/l10n>. The distribution contains language files in 308 languages and varieties with translations in widely varying degrees of completion (<https://www.debian.org/international/l10n/po/rank>). Among the Romance languages, the following are included here: French (with the varieties from France, Canada, Belgium, Switzerland, and Luxembourg), Spanish (with the varieties from Spain, Argentina, Columbia, Mexico, Ecuador, Peru, Chile, Venezuela, Nicaragua, Panama, Honduras, Guatemala, El Salvador, Uruguay, Costa Rica, the Dominican Republic, Puerto Rico, Paraguay, and Laos), Italian (with the varieties from Italy and Switzerland), Portuguese (with the varieties from Brazil and Portugal), Catalan (with the varieties from Catalonia, Valencia, France, Italy, and Andorra), Galician, Asturian, Occitan, Romanian, Friulian, Corsican, Sardinian, Haitian, Romansh, and Piedmontese.

It should be noted that the data gathered here on the three operating systems are difficult to compare and only apply to these base systems, and not to applications. However, they do provide indication that, in general, more languages and language varieties are encompassed in a free operating system, and that the smaller Romance languages are given significantly more room than with Windows and Apple’s OS X. Conflicts may arise where supported languages cease to be supported between versions. For example, Catalonia showed resistance when Catalan became unavailable at first following the jump to OS X (<http://www.macworld.co.uk/news/mac/catalonians-demand-catalan-support-mac-os-x-17033>).

All large-scale software producers and many FOSS localization projects have their own glossaries and style guides for Romance languages that have a normative effect. Here, the design of internal linguistic normative processes can vary greatly. Between operating systems and prominent programs as well, we see processes of linguistic leveling (adaptations, adjustments) both within individual languages and between languages. These processes have hardly been researched in the field of the Romance languages to date.

3.2Francophony

The localization of software into French receives special attention in the realm of activities in institutionalized Francophony, which includes the design of the relationship to endogenous languages, particularly in sub-Saharan Africa (here, cf. the survey in Gerbault in 2010 as well as Djité 2008 and Lipou 2007, 56). For Guédon (2004), the localization of software into French is an important opportunity to strengthen the position of this language over English, which he characterizes metaphorically as a kind of linguistic Microsoft (cf. ibid., 53). He presents the use and development of FOSS as a model to be developed for the production, circulation, and archiving of content in French. By contrast, Ben Henda (2004, 61) refers explicitly to technologically autonomous French alongside English, although he sees the software market as characterized by a linguistic imbalance in which other languages are technologically marginalized.

With reference to the linguistic construction of software, Bouffard/Caignon (2006) investigate the influence of localization on the varieties of French using the example of active elements in interfaces (menus, links, buttons, etc.) in the online presence of Mercedes-Benz in Belgium, Canada, France, Luxembourg, and Switzerland. The authors ask whether the lexical units used on these websites contribute to an international leveling effect for French, or whether they strengthen the independence of national varieties. They refer to variations even where international French is used; e. g., in the translation of the concept “send”: Belgium: “envoyer”, Canada: “soumettre”, France: “valider”, Luxembourg: “envoyer” (ibid., 813). Overall, their analysis paints a picture in which clear cultural distinctions are made in Francophone Canada with the innovative assistance of language, while in the francophone world, the practice of localization pushes toward a global uniformity – one that, however, does not take effect by means of an internationalization of French, but rather by means of Anglicization (cf. ibid., 820).

3.3Hispanophony

In comparison with French and Italian, the localization of software into Spanish has been better scientifically investigated (cf. Borrás Giner 2006, for example). Cañuelo Sarrión/Seiler (2008) deal with the localization of graphical user interfaces and with the translation of a program destined for the Argentine market from a theoretical perspective. The model of an español neutro oriented panhispanically and free from local and regional expressions that has been constructed and favored in audiovisual media out of financial interests, and the surrounding debate, are also visible in the realm of software. Mendiluce-Cabrera/Bermúdez-Bausela (2006, 451) mention a deal between Microsoft and the Real Academia Española in 1999 on the integration of the academy’s dictionary and the spelling and grammar check function in their own products. According to the authors, this makes Microsoft a technical language academy “because their choices of terms for the graphical user interface affects the terminology of other products that want to be compatible with those of Microsoft” (ibid., 451s.). In the context of this strategy, for example, the terminological difference between the Castilian “ordenador” and the Latin American “computadora” is bridged by the Anglicism “PC” and a clear attribution of grammatical gender is avoided by using the possessive, as in “su PC” (cf. ibid.). These forms may at first be perceived as unnatural by many users, but are soon accepted (cf. ibid., 453). Both authors clearly answer the question of the diatopic variation of Spanish in the realm of software with a reference to the implementation of an español neutro, but do not support their findings with empirical data as Bouffard/Caignon (2006) do for French. By contrast, many works by Jiménez-Crespo (e. g., 2010 and 2013) that focus on website localization (here, also cf. Diéguez Morales/Lazo Rodríguez 2011) are very well supported empirically.

In Spain, linguistically autonomous areas constitute a field in which a great deal of faith is placed in the economic, societal, and linguistic potential of FOSS within the connected social movement in terms of both policy and civil society (as in Latin America, often in a show of resistance to Microsoft as a linguistically political actor that seeks to monopolize certain decisions on localization). Relevant activities are registered at the Centro Nacional de Referencia de Aplicaciones de las TIC en fuentes abiertas (CENATIC) (<https://www.cenatic.es>). The autonomous regions have developed their own Linux distributions that are also localized into regional languages and that have been more or less successfully implemented in public administration and the education sector (e. g., LliureX in the Comunidad Valenciana). González-Barahona/Fernández-Sanguino Peña (2008) discuss the process of localizing Debian into Spanish. Aspects of the collaborative character of the localization of FOSS and the linguistic knowledge transfer between the actors involved (translators, users, programmers) are discussed in Seiler (2013b) using the example of the Spanish translator community of KDE (<https://www.kde.org>).

3.4Italophony

In comparison to French and Spanish, the scientific interest in localization into Italian is less prominent. The literature on this is very limited when one ignores texts that focus on the corresponding concepts, tools, and techniques in translator education. In this area, some of the most notable work is that by Sandrini which, among other things, deals with website localization and the potential of FOSS for localizing in the Italian context (e. g., Sandrini 2011 and 2014). So far, no publications have been made available on the role of software in the Italian sociolinguistic sphere or the pluricentricity of the Italian language within global Italophony.

The contribution of Cappelli (2007) is notable in the field of website localization; it is dedicated to the localization of websites in the tourism sector under consideration of a linguistic perspective. The author evaluates an Italian-English parallel corpus and a comparable English corpus of informative and promotional websites for Siena, performing a keyword analysis of the web content against the backdrop of considerations in search engine optimization.

Different project-oriented communities of translators are notable in the field of FOSS that create their own rules (e. g., cf. <http://tp.linux.it/buona_traduzione.html> or <http://l10n.kde.org>) and communicate via mailing lists, among other things (e. g., cf. <http://l10n.kde.org/team-infos.php?teamcode=it>). This type of internal list communication is a treasure trove for lay-linguist research and provides direct insight into language discussions and normative processes of Italian and other languages.