Alterity marking and enhancing accessibility in lexical borrowing: meta-information techniques in the use of incipient anglicisms in French and Italian

3.1 The interactional dimension of the use of the techniques

In order to resystematise communicative techniques that are frequently used together with lexical borrowings and to gain more insight into their motivation and communicative benefits, I propose to adopt a usage-based perspective that permits us to re-examine the different categories and phenomena. A first aspect of crucial importance here is the interactional dimension of the use of the techniques. By analysing linguistic phenomena from the perspective of their use in communicative events between a speaker/writer and a hearer/reader (or groups of speakers/writers and hearers/readers), both production and reception, coding and decoding, are taken into account. It needs to be stressed that although production is primarily speaker-based and reception hearer-based, both acts involve both participants. As has been highlighted in pragmatic research since Grice (1975), speakers when formulating their utterances, anticipate the hearers’ interpretations – based on, e.g., what the speakers assume to be their linguistic and extra-linguistic knowledge, communicative interests, goals, etc. –, and conversely the hearers, when interpreting the speaker’s utterance, take into account their assumptions about the speaker’s linguistic and extra-linguistic knowledge, communicative interests, goals, etc. Thus, in the coding stage, the speaker anticipates the decoding stage by the hearer, and conversely, in the decoding stage, the hearer integrates assumptions about the coding stage by the speaker (e.g., communicative intentions, invited inferences, etc.). In both cases, however, we are dealing with assumptions that need not necessarily correspond to the real facts, i.e. it needs to be acknowledged that linguistic analyses will necessarily remain to a certain extent hypothetical, and the task will thus be to formulate sufficiently plausible analyses based on abstractions over individual speaker/hearer personalities.

It could be objected that analyses of newspaper articles such as the ones that will be proposed in Sections 4 and 5, are in contradiction with the usage-based approach argued for since the focus is only on the texts while the speaker/writer and hearer/reader are not taken into account. A response to this objection could be that in the case of newspaper articles, we are dealing with written communication in communicative distance, where the stage of production and the stage of reception occur in temporal and spatial distance and where there is a one-to-many communication with a relatively large and heterogeneous group of addressees (cf. Koch and Oesterreicher 2012). In scenarios of this kind, the texts are objectively given and directly accessible, whereas the stage of production and the stage of reception of the newspaper articles is typically not accessible to linguistic analyses (except when an experimental design is created). Still, production and reception are both reflected in the way the newspaper articles are formulated: it can be assumed that when formulating the text, the speaker/writer will anticipate how ‘average readers’ or groups of ‘typical’ readers (cf. the choice of a corpus of newspaper articles in widely diffused general national newspapers) will interpret it, taking into account their linguistic and extra-linguistic knowledge as well as the preceding co-text. For instance, when using an anglicism that has already been defined or translated in the text, the speaker may presume that no further meta-information techniques need to be used to ensure understanding for the later occurrences. The focus on communicative techniques in the entire texts taken as communicative units thus also raises new research questions about the ways in which their use may evolve in the course of the texts. These questions will be addressed in the following subsection as well as in the analyses in Sections 4 and 5.

3.2 The textual dimension

As the discussion in Section 2 has shown, Pflanz (2014) analyses the use of alterity markers in the immediate co-text of the borrowed items, generally in the same sentence. However, when we focus on communicative events between speakers and hearers, it seems reasonable to consider the entire texts as communicative units. For written communication, as in the case of newspaper articles, this means that the entire texts in which the borrowed items are embedded need to be taken into account. If this research focus is adopted, we can observe that beyond particular instances of, e.g., flagging, there are many other textual elements that may contribute to facilitate the use of anglicisms. We typically find various more or less salient links across the text established by the use of coreferential or semantically related expressions, and these expressions can also provide information on the use and meaning of the anglicisms. The notions of textual cohesion and coherence as they have been established in text linguistics, thus appear to be directly relevant to the analysis of meta-information techniques for lexical borrowings.

Such a focus on the entire texts as communicative units also implies a radical shift with respect to standard corpus linguistic analyses, in which typically only the immediate co-text is taken into account. The larger focus on entire texts appears to be particularly important for the technique that has been discussed as ‘explain X’ and characterised by the borrowing being defined or described (Pflanz 2014). From a text-linguistic perspective, the question arises what is to be included here exactly: should we count only definitions in the strict sense or also accept other types of contextual information that can inform the hearer/reader about certain aspects of the meaning of a lexical borrowing? I assume that it makes sense, as a first step, to adopt a broad approach and include all types of information that is semantically related to the borrowing and may thus facilitate its interpretation.

A criterion that permits us to identify the relevant information is provided by the inventory of basic semantic relations established by Blank (1997; cf. also Koch 2004): contiguity, metaphorical similarity, taxonomic relations (subordination, superordination, cotaxonomic similarity) as well as conceptual and cotaxonomic contrast. Among these relations, contiguity has been shown to be most important for both the lexicon and grammar. For borrowings, we can therefore assume that their understanding can be facilitated by providing information about the relations between the relevant semantic frames and the elements contained in these frames, as well as between different elements of the frames. Following the cognitive semantic approach of Blank and Koch, I therefore propose, in order to highlight more explicitly the theoretical semantic foundations of this textual technique, to relabel the relevant communicative technique as frame information.

The importance of frame information for text structure can be demonstrated for the borrowing shoffice introduced in the Italian sample text I0, which has already been cited in the introduction. Across the Italian newspaper article, there are numerous expressions that contribute to inform the reader about the meaning of the new expression shoffice and the new concept it designates. For example, we find references to the hyperonymous concepts home office and office, and to various concepts that belong to these conceptual frames and are linked to them and to each other via relations of contiguity, e.g., work, garden shed, furnishing, table, chair, nature. In addition, we also find metaphorically related concepts (nomadism) and relations of contrast (in – out), as well as references to the novelty and innovativeness of the concept of shoffice:

The text thus creates a network of semantically related expressions that can help the addressee to infer the meaning of shoffice, if needed (cf. Figure 1, where this is illustrated for selected concepts, focusing on concepts related by contiguity and belonging to the frames garden, office and trend, which appear to be most important here; in addition to the concepts, I also indicate the expressions as they appear in the Italian text).^[7]

Figure 1:

Frame information on shoffice based on contiguity relations within the frames garden, office and trend as provided in newspaper article I0.

The example shows the importance and ubiquity of semantically related information, which is key to textual coherence. For concrete text analyses, however, especially if quantitative aspects are to be integrated, the question arises how the relevant cases of frame information can be reasonably restricted, broken into relevant information units, and counted. In the analyses presented in Section 4, I will adopt a narrower heuristic based on the criterion of contextual coreference, and limit the frame information taken into account in the analyses accordingly. It is important to note that the coreference may only be given in the concrete utterance context (i.e., it is not only based on the lexical meaning of the expressions but also on their interpretation in the concrete utterance). In addition, I propose to include here not only contextual near-synonyms and hypernyms, which can be identified by the formula ‘X is Y’ (e.g., “un ufficio tutto per sé”, “un ottimale temporary office sotto il sole”: shoffice is un ufficio tutto per sé/shoffice is un ottimale temporary office sotto il sole), but also predicative structures describing the referent, which can be identified by the formulas ‘(Being) X means that Y’ or ‘X has the characteristic Y’ (e.g. “il capanno da giardino è diventato nuova postazione di lavoro”: shoffice means that il capanno da giardino è diventato nuova postazione di lavoro). In contrast, expressions such as “tavoli e sedie” or “il verde” referring to contiguous concepts such as tables, chairs, green, etc. will not be counted in the analyses.

3.3 Types of meta-information techniques: flagging, metalinguistic comments, and frame information

The previous discussion has already suggested that various strategies can be employed when using new lexical borrowings in the RL in order to facilitate understanding. Given the strong differences between the strategies, however, the question arises whether they can be adequately described as alterity-marking techniques. At first glance, this term appears to adequately describe the basic function of flagging. However, a closer look reveals that the techniques that can be ranged under this concept exhibit different degrees of alterity marking. Whereas quotation marks conventionally express that a stretch of discourse uttered by another speaker is inserted into the utterance, i.e. they conventionally signal alterity, this interpretation represents only an invited inference in the case of the use of italics or bold print, the conventional meaning of italics and bold print being restricted to the mere function of highlighting.^[8] As to the other techniques, they do not only aim to mark the relevant expressions but also provide some additional information that can contribute to facilitate the comprehension and use of the novel expressions in the RL. It therefore seems more adequate to speak of ‘meta-information techniques’ because the techniques all convey some type of additional information that refers to the linguistic expressions themselves.

To systematise different subtypes of techniques, we can then refer to semiotic criteria, asking how the relevant meta-information is conveyed by the speaker and communicated to the hearer. Two aspects appear to be fundamental here: intensional richness, and the question of whether the information is explicit/overtly presented or implicit. Based on these criteria, we can distinguish three basic types of meta-information techniques that can accompany the use of novel lexical borrowings (and lexical innovations in general). These can be defined as follows:^[9]

1. Flagging, where the relevant linguistic expression is explicitly marked by prosodic or typographic means (italics, quotation marks, pauses, etc.), bringing the addressee’s attention to the expression itself but without providing any explicit metalinguistic information about it; the marking is thus realised by linguistic devices below the lexical or morphological level, e.g., “Les relecteurs «woke»” [«woke» proofreaders];

2. Metalinguistic comments, consisting in meaningful expressions^[10] that accompany the relevant linguistic expression and provide explicit metalinguistic information about it, e.g., “ha cominciato a chiamarsi woke (dal verbo svegliare)” [started to call itself woke (from the verb to wake up)] (with additional flagging for woke);

3. Frame information, consisting in information contained in the co-text that is semantically related to the meaning of the relevant linguistic expression and can therefore contribute to illuminate its meaning and usage but without the presence of explicit metalinguistic expressions, e.g., “Tendenza Shoffice, un ufficio tutto per sé” [The shoffice trend, an office of one’s own].

Flagging^[11] and metalinguistic comments roughly correspond to the techniques ‘show X’ and ‘comment on X’ as defined by Pflanz (2014), whereas frame information strongly overlaps with ‘explain X’. Depending on the presence or absence of explicit metalinguistic expressions, cases of ‘translate X’ may correspond to either metalinguistic comments or frame information.

The three techniques correspond to fundamentally different ways in which the relevant items are coded and decoded by speakers in production and perception. Whereas flagging can be characterised as being explicit and intensionally poor, metalinguistic comments are explicit and intensionally rich, while frame information is implicit and intensionally rich.

Furthermore, concerning the category of flagging, I would like to argue that in addition to the techniques that have been mentioned in previous research, new possibilities opened up by technological progress, most importantly in digital communication, should also be taken into account. That is, I include here all kinds of techniques that can visually or acoustically highlight particular expressions against the rest of the utterance. For the newspaper articles investigated in this paper, this concerns above all the use of hyperlinks. Even though the primary function of hyperlinks is not to highlight the expressions, they are often indicated by a different font colour and/or by underlining and a changed display of the cursor, so that a highlighting element is also present here. Before turning to the analyses of French and Italian newspaper articles to investigate the applicability of the revised classification and the relevance of these techniques, I would now like to address the question why speakers should feel a need to use meta-information techniques at all.

3.4 Reduced accessibility of lexical borrowings at the levels of form and content

Previous research on lexical borrowing and loanword integration has highlighted the fact that loanwords are typically characterised by specific features that distinguish them from the native lexicon and from other types of lexical innovation. Adopting a usage-based perspective, these features can be reinterpreted in the light of the concept of accessibility. I understand accessibility as an inherently gradual and dynamic concept that expresses how easily speakers can retrieve particular linguistic items or structures from their linguistic knowledge in text production, and how easily addressees can link linguistic items or structures they encounter to their existing linguistic knowledge.^[12] In this respect, we can distinguish between aspects of form and content, and describe typical features of reduced accessibility that can be observed in lexical borrowings for both dimensions.

Before commenting on specific aspects of reduced accessibility, I would like to note that, in accordance with the usage-based approach argued for, accessibility is first of all an individual phenomenon that occurs in single usage events. Accessibility is closely linked to the notion of cognitive effort, i.e., the relative ease with which a speaker or hearer can process particular linguistic items. The accessibility of a certain linguistic item will therefore depend on the individual speakers’ linguistic and extra-linguistic knowledge as well as on co-textual and contextual factors (e.g., emotional state, time pressure, activation level of different languages for plurilingual speakers, etc.). When investigating specific examples of lexical borrowings, however, it is necessary to abstract away from individual speakers and usage events, and the remarks that follow thus comment on aspects that can be assumed to be relevant for “typical” speaker groups in the RL. The remarks on specific examples cannot claim to cover the entire speech community, however, and more fine-grained distinctions would be necessary to integrate different speaker groups within the RL depending on factors such as the degree of knowledge of the SL, etc.

Moreover, the additional processing effort assumed for linguistic items with a reduced accessibility should not be equated with a difficulty that is perceived as such by the individual speaker. For example, in cases of strong exposure to the SL as is the case for English in France and Italy, words containing non-native graphemes such as <k> may not be felt to be difficult or in need of explanation; still, it can be assumed that these items are felt to be to some extent marked and different from the core forms of the RL (cf., e.g., pragmatic choices of spelling variants containing <k> in computer-mediated communication, cf. Italian <che> vs. <ke>). In some cases, there are secondary codes introduced via borrowings that have become strongly established in the RL and that can even be productively expanded to other RL items.^[13] Here as well, I suppose that the mere existence of a peripheral code in addition to the core code can be seen as causing some extra cognitive effort because the classification of the relevant linguistic items as belonging to a peripheral system represents an additional step in processing these items.

A first major domain where reduced accessibility can be observed concerns form-related aspects. These can be linked to previous research on the structural foreignness of borrowed items, and to the extensive research on loanword integration. Processes of loanword integration consist in structural changes in lexical borrowings where non-native features contained in the SL realisation of the borrowing are replaced with features that correspond to the structures of the RL system, i.e. loanword integration is characterised by a non-conformity of the relevant structures to the SL form, and a conformity to the RL system (cf. Winter-Froemel 2008a, 2011). In the absence of loanword integration, or in cases of incomplete loanword integration, additional rules (e.g., pronunciation or spelling rules) need to be learned together with the items, which requires an additional effort in production and perception.

Structural foreignness can concern the levels of pronunciation and spelling as well as the morphological level. For the former two levels, it is important to stress that it is not only the individual phonic and graphic segments that are potentially concerned, but also phonotactics and graphotactics as well as correspondence rules between phonic and graphic segments, i.e. the question whether the way in which the borrowing is written and pronounced corresponds to the orthographic rules of the RL system. For example, for French baby [bebi] (cf. PR 2022), no structural foreignness at the level of the phonic and graphic units can be observed. The phonic segments [b], [e], [i] correspond to elements of the French phonological system: /b/, /e/, /i/; the graphic segments <b>, <a>, <y> are also used in many native words of French and represent graphemes of French. Still, the lexical item baby does not entirely follow the rules of French, which would require either a different spelling (e.g. [bebi] → <bébi>) or a different pronunciation (<baby> → [babi]).^[14]

Form-related aspects of reduced accessibility are frequent for recent lexical borrowings from English into French and Italian. The importance of this dimension of alterity, however, strongly depends on the respective SL and RL, their genealogical and typological distance, their recourse to joint patterns of lexical innovation (e.g., based on roots from Greek and Latin) as well as on conventions of loanword integration.^[15] It is thus perfectly possible that lexical borrowings do not or no longer exhibit any structural foreignness when they are imported into a particular RL (cf. also the example of the anglicism minutes in French, which does not exhibit structural foreignness at the levels of spelling and morphology, see Section 4.2 below). An interesting example in this respect is the French expression mission impossible: when this expression appears in a French text, it could in principle be interpreted either as a code-switch (cf. English mission impossible), or as a loanword imported from English, or as a calque.

Content-related aspects of accessibility can be analysed making use of cognitive-semantic approaches (Blank 2003; Koch 2004). These approaches emphasise that together with formal relations, there are basic semantic relations that link lexical items to other items of the lexicon, thus creating motivational networks, or more exactly, possible motivational networks. Following the usage-based framework, it needs to be acknowledged that the motivational relations may be perceived differently within the speech community; in this sense, they rather express a motivatability of the relevant items, i.e. a potential of being related to other items in the lexicon.

In order to characterise the relative accessibility of lexical borrowings, it is helpful to compare them to other types of lexical innovation, i.e. word formation and semantic change. Fundamental differences emerge here: lexical innovations arising from semantic change or word formation are typically motivated by the items from which they are coined (e.g., in the SL English, the meanings of the polysemous item woke are semantically related, so that the meaning ‘alert to racial or social discrimination and injustice’ is motivated by the original meaning of ‘awake’; similarly, the compound panic buying is motivated by its components panic and buy(ing) via relations of taxonomic subordination and contiguity – panic buying is a subtype of buying that arises out of a feeling of panic; cf. OED 2007. For lexical borrowings that are based on analogical innovation (cf. the traditional categories of calques and semantic borrowings), there is also a motivatability: for instance, for French souris, coined by imitation of the English model of mouse ‘animal mouse’ → ‘computer mouse’, the new meaning ‘computer mouse’ is also linked to the original meaning of French souris ‘animal mouse’, so that the new item is motivated in the RL as well. For loanwords, i.e. the other major group of lexical borrowings, which is based on importation of foreign items, however, no such motivatability is present in principle since only the isolated expressions without their SL motivational networks are borrowed (see, e.g., Italian mouse, which cannot be related to other items of the Italian lexicon). This lack of motivatability or intransparency has been mentioned as a potential advantage of loanwords, leading to new expressions that are unambiguous in the RL (cf. Winter-Froemel 2008b).

The feature of intransparency needs to be nuanced in several respects, however. As borrowing implies two (or more) languages, the analysis of the motivatability or semantic transparency in the RL needs to be complemented by transparency/motivatability with respect to the SL and potentially, other languages, i.e. there is also an interlinguistic dimension of transparency (cf. Winter-Froemel 2008b). Many recent borrowings from English do not only concern a particular RL but also reflect more general processes of cross-linguistic influences at an international level, often leading to internationalisms, i.e. to items that are diffused in different RLs and language families and remain sufficiently recognisable for non-native speakers (cf. Braun 1978).

Moreover, as already mentioned in the preceding subsection, full or partial transparency can arise from different sources, among others, the relative genealogical and typological closeness, traditions of language contact between the SL and the RL, the existence of similar traditions of lexical innovation based on common cultural adstrates, etc. Depending on the SL knowledge of the individual RL speaker, borrowed items can be fully transparent, especially if the SL items are highly frequent or represent words that are typically encountered early in foreign language acquisition (e.g., English eye, hand, cat, come, back, second, next, etc.) or appear in internationally diffused movie titles, song texts, etc.

New motivational networks can also be created when various elements from a word family or word formations based on the same elements are borrowed. For example, the loanword cat eye can also be linked to previous borrowings such as eyeliner, and when interpreting the expression baby bump in French, it can be assumed that the RL speaker can relate this element to the item baby, which is well established in French. In some cases, however, formal similarity may also suggest misleading semantic links that may hinder the comprehension of the borrowings in the RL (cf. the traditional category of ‘false friends’). For instance, the first component of the French anglicism motion gaming is not directly linked to French motion ‘(formal) request’.

Concerning semantic and pragmatic aspects of accessibility, Pflanz (2014: 178–189) also observes that the use of particular marking techniques may depend on whether there are RL equivalents available to express the relevant concept or not (cf. the discussion of cases where ‘the WORD and the THING [are] foreign to the writer’ in Section 2). This aspect can be linked to the distinction between catachrestic and non-catachrestic borrowings (Onysko and Winter-Froemel 2011; Winter-Froemel 2011), where the former are defined as introducing a new concept alongside the new expression (e.g., Italian shoffice), and the latter are characterised by the presence of a native equivalent in the RL (e.g., French topless alongside seins nus). For both groups, we can assume a reduced accessibility, but one arising for different reasons: for catachrestic borrowings, a new expression and a new concept need to be simultaneously integrated into the linguistic and extra-linguistic knowledge of the RL speakers. For non-catachrestic borrowings, in contrast, it is their relative markedness compared to the established near-synonyms that defines their reduced accessibility.

Finally, it should be noted that form-related and content-related aspects of accessibility are closely linked. This can be illustrated by cases of loanword integration highlighting morphological boundaries and units. In previous research, it has been observed that for anglicisms introduced into French (cf. Mathieu-Colas 1994; Koláříková 2020) and Italian (Winter-Froemel 2018), a hyphen can be additionally introduced. Interestingly, this tendency concerns both borrowings that are used with compound spelling in English (e.g., English burnout → French burn-out, cf. Koláříková [2020: 41], English eyeliner → Italian eye-liner) and borrowings with separate spelling in the SL (e.g., English smoked meat → French smoked-meat, cf. Koláříková [2020: 41], English blue jeans → Italian blue-jeans, Winter-Froemel [2018: 90], cf. also Klajn [1972: 94–95]). From the perspective of RL addressees, hyphenation enhances accessibility: on the one hand, the hyphen “indicates the togetherness of the elements functioning as a structural unit, thus offering a benefit compared to separate spelling” (Winter-Froemel 2018: 90), and on the other, it

signals the compositional structure of the items borrowed – possibly facilitating a partial semantic transparency for the RL speakers with respect to highly frequent SL elements (e.g., English eye […]) or formally close RL equivalents (English blue – Italian blu […]). It thus breaks the long strings of segments of the SL items into two subsegments of shorter length which can be processed more easily (Winter-Froemel 2018: 90).

3.5 Basic functions of meta-information techniques: alterity marking and enhancing accessibility

Based on the previous reflections on the reduced accessibility of lexical borrowings, it can be assumed that in addition to alterity marking, i.e. indicating that the expressions originate in the discourse of another speaker or speaker group/speech community, further communicative aspects also play a key role in the use of meta-information techniques. Thus, I propose to add the function of enhancing the items’ accessibility as a second basic aspect that motivates the use of meta-information techniques. While the notion of alterity marking foregrounds the process of coding and thus the perspective of the speaker, the notion of accessibility primarily focuses on decoding and thus on the perspective of the hearer. Following the interactional approach argued for in Section 3.1, however, both speaker and hearer represent essential participants of communication, and thus, both notions take into account both perspectives.

The notion of accessibility, complementing alterity, permits us to reconsider the three groups of meta-information techniques with respect to their communicative functions. Alterity marking and accessibility enhancing appear as two key functions, and an interplay of these functions can be assumed, with one or the other function predominating depending on the communicative context. Moreover, certain correlations between the three main groups of techniques and the two functions can be assumed. For flagging, the function of alterity marking appears to predominate; depending on the concrete communicative context, however, it can also invite the addressees to focus their attention on the relevant expression in order to prepare themselves for an additional cognitive effort caused by a “difficult” pronunciation, spelling, or simply a new expression that is not yet strongly entrenched in their mental lexicon. For metalinguistic comments, Pflanz (2014: 166–168) already proposed to distinguish axiological and didactical aspects. In my view, these two options are closely linked to alterity marking (indicating that the expression belongs to the discourse of others, with an implicit distinction made with respect to the speaker’s discourse, and often a negative evaluation of the discourse of the other being suggested), and accessibility enhancing, respectively. Finally, for frame information, the function of enhancing accessibility appears to clearly predominate (but with the function of alterity marking being implicitly present in the background as well, e.g., when the frame information focuses on the foreign origin of the expressions). The three basic types of markers can thus pragmatically be characterised as shown in Figure 2. It should be stressed, however, that the correlations indicated in Figure 2 should only be considered as general tendencies that can be actualised to different degrees in concrete utterances depending on the communicative context.

Figure 2:

Meta-information techniques between alterity marking and accessibility enhancing.

The classification thus reconceptualises the meta-information techniques, emphasising more strongly their cooperative functions in speaker-hearer interaction. The interplay of the two key functions can also be directly linked to the findings of Klosa-Kückelhaus and Wolfer (2020). In their survey on the use and evolution of anglicisms in German compared to German neologisms, they observe a higher overall frequency of flags for borrowed than for German neologisms, and a decrease of the frequency of flags for both groups as the neologisms become more diffused and frequent. Both findings can be linked to the interplay between alterity marking and accessibility enhancing: the overall decrease of the use of flags can be interpreted as reflecting the fact that the neologisms become more easily accessible so that the function of enhancing accessibility loses in importance. At the same time, the stronger use of flags for neologisms of non-native origin reflects, on the one hand, the fact that loanwords exhibit a reduced semantic accessibility (typically being semantically intransparent in the RL), and on the other, that speakers may feel a communicative need to signal the foreign origin of the items in the sense of alterity marking.

The following sections will further develop these reflections by investigating the use of recent anglicisms in French and Italian newspaper articles, linking analyses of the items’ reduced accessibility to observations on the use of meta-information techniques in the texts.

4.1 Methodology

In order to analyse the use of the techniques discussed for recent anglicisms in French and Italian, this section will present a primarily qualitative survey based on ten articles (F1–F10, I1–I10) published in widely diffused general national newspapers in France and Italy (Le Figaro/Madame Figaro, Le Monde, Libération, and La Repubblica, La Stampa). The length of the newspaper articles varies between 319 and 1,643 words for the French texts and 390 and 1,472 words for the Italian texts. The articles were published between March and June 2022 in the online versions of the newspapers and were accessible free of cost. For each language, I selected ten articles where a recent anglicism in the headline or in the lead paragraph is accompanied by at least one of the techniques discussed above, that is, flagging, metalinguistic comments, or frame information. The survey thus focuses on articles where a strong use of meta-information techniques can be expected, in order to reveal how they are used and combined in the course of the articles, which are understood as communicative units exchanged between the text producer and the addressee. A list of the articles is given in the Supplementary Material.

The analysis first addresses the reduced accessibility of the twenty anglicisms investigated (see Section 4.2), and then proceeds to the use of the different meta-information techniques in the newspaper articles, asking how the techniques are combined and whether their use evolves within the articles, and if so, how. Does their usage show certain patterns in the course of the texts, e.g., in the sense of certain techniques being predominantly used in earlier or later parts of the newspaper articles, etc.? Concerning the identification of relevant frame information, the criteria outlined in Section 3.2 are adopted, i.e., contextual near-synonyms and hypernyms, which can be identified by the formula ‘X is Y’, and descriptions identified by the formulas ‘(Being) X means that Y’ or ‘X has the characteristic Y’ are taken into account. As this type of technique is precisely characterised by the fact that no explicit link to the relevant borrowings is established, the identification of textual elements that fulfil this function is based on a qualitative semantic analysis of the contextual meanings of the relevant expressions. In addition, I will comment on the occurrence of other anglicisms and other types of marked lexical items in the articles. The group of other anglicisms includes words or phrases with structural features of non-nativeness (spelling, pronunciation, morphology), proper names of English origin, and code-switches (e.g., anglicisms or stretches of English text appearing in song titles, magazine titles, etc.) as well as further RL innovations based on borrowed items (e.g., French tennistique, Italian manageriale).

In the discussion of the use and interplay of the different techniques, I will focus on the anglicisms with the highest numbers of occurrences, which were accompanied particularly often by one of the techniques; for the anglicisms and the techniques, overall counts will be given. To investigate the use of the different types of meta-information techniques in the course of the articles, the techniques will be categorised according to their distribution in different parts of the articles, distinguishing between the headline, the lead paragraph and the text body, which fulfil different communicative functions with respect to attracting the readers’ attention and providing information (as most articles do not contain a clearly identifiable conclusion part, no additional category is applied here). This will provide insight into the use of the techniques following their linear order in the text and permit us to determine whether the anglicisms are flagged throughout the individual articles or only highlighted at their first occurrence; I will therefore also indicate unflagged occurrences when discussing the results. Sample analyses of two of the newspaper articles are given in the Supplementary Material; for each language, the shortest text was selected (F2/I3).

4.2 Results: reduced accessibility

An overview of the reduced accessibility of the anglicisms investigated in the present survey (for French: baby bump, cat eye, come-back, fashion victims, minidrone, minutes, motion gaming, Next Gen, panic buying, topless; for Italian: algospeak, blockchain, cyber-operations, freelance, gamer, gender gap, running, second hand, smart working, workout) is given in Table 1 below (on borrowings of English deverbal nominalisations with the suffix -ing into Italian, see also Meinschaefer, this issue). Concerning formal aspects, we can find various patterns at the levels of spelling, pronunciation, and morphology that imply a reduced accessibility. As to non-native graphic units, we find for the French anglicisms <sh> in fashion victims, <ck> in come-back, <uy> in panic buying, and for the Italian anglicisms <ck> and <ai> in blockchain, <ee> in freelance, <ea> and <k> in algospeak, <w> and <ou> in workout, <y> in cyber-operations, <w> in smart working.^[16] All these graphic patterns do not correspond to the core graphic systems of French and Italian and can be assumed to require additional processing efforts and thus be potentially perceived as difficult by RL speakers (cf. Meisenburg 1996).

Concerning the pronunciation of the anglicisms by RL speakers, a coexistence of different variants needs to be assumed, e.g. depending on degrees of knowledge of the SL and context variables. Moreover, the smooth transitions between code-switching (characterised by a pronunciation that remains very close or identical to the SL patterns) and lexical borrowing (characterised by different degrees of loanword integration; cf. Poplack [1980]) need to be mentioned. As the newspaper articles investigated represent cases of communication in the graphic code, a full analysis of the relevant pronunciation variants cannot be given based on the data. I will thus only highlight major phenomena at the interface of the graphic and the phonic level; the comments that follow are not intended to cover all pronunciation variants that can be observed in the RL communities.

Non-native correspondences between phonic and graphic units^[17] include in French the correspondences <a> ↔ [e] and <u> ↔ [œ] in baby bump, <a> ↔ [ɛ]/[e] in fashion victims, cat eye, come back and motion gaming, <o> ↔ [a] in come-back, <ti> ↔ [ʃ] in motion gaming, <g> ↔ [dʒ] in Next Gen and the pronunciation of <eye> as [aj] in cat eye, and in Italian the correspondences <c> ↔ [(t)s] in freelance, <c> ↔ [s] in cyber-operations, <a> ↔ [ɛ] in gender gap, and <u> ↔ [a] in running, and the pronunciation of <ai> as [ej] in blockchain. However, strong variation in pronunciation needs to be assumed, especially given the recency and incomplete conventionalisation of the loanwords in the RL. Depending on a more or less SL-oriented pronunciation and less or stronger loanword integration, further aspects of alterity and reduced accessibility could thus be mentioned, or some of the features listed might disappear when loanword variants with a stronger degree of integration become conventionalised.

Further structural features of foreignness arise at the phonotactic and graphotactic as well as the morphological level because the loanwords contain patterns that do not correspond to the structures of the RL system. This concerns loanwords with word-final consonants or consonant clusters such as [mp] and [ms] in French baby bump and fashion victims, and the word final consonants and consonant clusters in Italian blockchain, second hand, gender gap, algospeak, workout, cyber-operations, smart working, running (in this respect, it is interesting to note that especially for Italian, pronunciation variants with an additional epenthetic vowel following the final consonant are largely diffused, cf. Schweickard [1998]). In some cases, an interplay of phonetic and morphological aspects can be observed, e.g., if the plural marker <-s> is phonically realised in French (e.g., fashion victims), this also represents a marked feature even if the correspondence <s> ↔ [s] itself is not marked in French.

Another feature of alterity and reduced accessibility is the structure of compounds conserving the SL order of modifier-head, which differs from the Romance order head-modifier (see French baby bump, cat eye, fashion victims, minidrone, motion gaming, panic buying, Italian algospeak, blockchain, gender gap, cyber-operations). This feature is important for compounds that are fully or partially semantically transparent (cf. English victim/French victim, English panic/French panique, English operation/Italian operazione, and for internationally widely diffused components such as baby, fashion, cyber, eye, gaming, gender), whereas it can be assumed that the SL morphological structure will not be perceived as a feature of alterity when the borrowings are entirely non-transparent (this could be the case, e.g., for the anglicism blockchain in Italian, at least for part of the RL speech community).

Concerning content-related aspects of (reduced) accessibility, we can first observe that various anglicisms can be motivatable for RL speakers based on other RL items, including cognates and items that have previously been borrowed and are nowadays widely diffused in the RL (e.g., French baby bump → baby, cat eye → eyeliner, fashion victims → victime, minidrone → mini and drone, panic buying → panique, topless → top; Italian: blockchain → blocco, cyber-operations → ciber-/cibernetico and operazione, second hand → secondo). Because in these cases, a relative motivatability is given and accessibility is not reduced but enhanced, a ‘minus’ is indicated in the respective cells in Table 1.^[18] For French minutes and motion gaming as well as for freelance, however, relations to native items (French minute, motion, Italian lanciare) could be interpreted as being rather misleading, in the sense of suggesting an interpretation that does not apply to the meaning of the anglicism. In these cases, a reduced accessibility with respect to motivatability in the RL can be assumed. In addition, for some of the anglicisms, an international motivatability can also be assumed (see the ‘minus’ indications in Table 1); this is the case when the items are widely diffused in international communication and in different language families, such as for French come-back, topless, and Italian blockchain, freelance, gamer, gender gap, second hand or workout (e.g., all of them are attested in the German dictionary [Duden 2022] as well).

Concerning the catachrestic or non-catachrestic nature of the borrowings, both categories are represented in the data. The following items can be interpreted as being catachrestic, i.e., both the expression and the concept are new, and therefore, a reduced accessibility can be assumed: French cat eye, minidrone, motion gaming, Italian algospeak, blockchain. For the other items, in contrast, there are established RL equivalents so that the anglicisms can be interpreted as being relatively marked and thus less accessible, compared to these (near-)synonyms: French baby bump ∼ ventre de bébé/ventre de femme enceinte, come-back ∼ retour, fashion victims ∼ victimes de la mode, minutes ∼ procès-verbal/compte-rendu, Next Gen ∼ de prochaine génération, panic buying ∼ achat panique, topless ∼ seins nus, Italian cyber-operations ∼ operazioni informatiche/operazioni cibernetiche, freelance ∼ libero professionista, gamer ∼ giocatore, gender gap ∼ divario di genere/divario tra i sessi, running ∼ corsa, second hand ∼ (di) seconda mano, smart working ∼ lavoro intelligente, workout ∼ allenamento.

Finally, for French come-back and Italian cyber-operations,^[19] it can be assumed that the accessibility of these items is increased by hyphenation. Otherwise, hyphenation is not strongly attested, and the other compounds and syntactic compounds (French baby bump, cat eye, fashion victims, minidrone, motion gaming, Next Gen, panic buying, topless, Italian algospeak, blockchain, freelance, gender gap, second hand, smart working, workout) thus appear to be less accessible.

4.3 Results: meta-information techniques

A first observation is that the French and Italian newspaper articles contain meta-information techniques that are similar to the ones observed by Pflanz (2014) for German newspapers (see the results shown in Table 2 below). When applying the categories to the texts, the general usefulness of the revised categorisation and the distinction of three basic techniques – flagging (F), metalinguistic comments (C), and frame information (FI) – was confirmed. All the techniques are attested, with frame information being most frequent (representing 213 of 256 cases, i.e. 83.2 %), followed by flagging (31 cases, 12.1 %), whereas metalinguistic comments appear to be relatively marginal (12 cases, 4.7 %). The strategy of flagging is illustrated in (5)–(7), showing instances of flagging with the use of double quotation marks, italics, and bold print, respectively.

Metalinguistic comments are illustrated in (8) and (9). (8) also illustrates multiple uses of marking techniques (here, there is additional flagging with the use of double quotation marks and a hyperlink), as already observed by Pflanz (2014). Only in three of the French texts and two of the Italian ones, metalinguistic comments are found at all, and the analysis suggests that the use of this technique may depend on further factors such as text length (the three French texts containing metalinguistic comments are precisely the longest texts in the French corpus) and content (in the two Italian newspaper articles where metalinguistic comments are found, the headlines explicitly announce the intention to explain a new concept to the readers).

Finally, frame information occurs in all texts, and it is also often combined with the other techniques (see e.g., the descriptions “un mercato che nel nostro Paese vale […]” in [7], “un libro mastro sotto forma di database” and “è una catena (infinita) di blocchi” in [9]).

The semiotic criteria proposed to identify flagging and metalinguistic comments, i.e., their explicitness and informational poorness/richness, proved to be straightforwardly applicable. For frame information, in addition to strictly co-referential expressions, the textual analysis suggests that further concepts linked to the meaning of the anglicisms by relations of contiguity also help facilitate their understanding. Moreover, textual structures, e.g. identical or similar collocates appearing together with anglicisms and coreferential expressions, turn out to be highly relevant (see the discussion in Section 5). Finally, concerning the overall importance of meta-information techniques in the newspaper articles, the analyses show that their frequency of use varies (see the ratios between 0.41 and 3.13, and 0.98 and 3.93 techniques per 100 words for the French and Italian articles, respectively).

Another observation that emerges from the analysis is that flagging is more regularly used in French (27 cases of flagging) than in Italian (4 cases) in spite of the smaller corpus size of the French subcorpus (cf. the numbers given in the Supplementary Material). Among the French articles, flagging is completely absent only in F6, whereas in the Italian articles, flagging only appears in I2, I8 and I9. We mostly find double quotation marks and italics in the French texts, whereas bold print prevails in the Italian texts; for both languages, we also find occasional examples of flagging via hyperlinks. Moreover, flagging appears to be the normal case in French (27 flagged uses vs. 4 unflagged uses), whereas the situation is markedly different in Italian: the 53 unflagged uses clearly surpass the 4 flagged uses of the anglicisms. However, due to the small corpus size, these findings need to be carefully interpreted and backed up by further research.

As regards the use of the different techniques of meta-information in the course of the newspaper articles, several observations can be made. The selection of articles was based on the occurrence of at least one type of meta-information technique in the headline or the lead paragraph. It therefore appears interesting to note that except for F1, all the articles show the use of further techniques, most importantly, frame information, in all of the text sections. The analysis suggests that the use of the anglicisms as novel expressions in the headlines and lead paragraphs serves to attract the readers’ attention and interest, but the frame information that immediately follows them (or precedes them, cf. F4, F5, I9) shows that the writers also aim to ensure the comprehension of the potentially difficult items right from the beginning of the articles.

In the body of the text, we find numerous meta-information techniques, most importantly, frame information, whereas metalinguistic comments are less frequent but still attested (F7, F8, F10, I1, I5). The articles I1 and I5 are to some extent specific cases since their titles explicitly indicate their main aim i.e. to comment on the new realities (and expressions) in an accessible way (cf. the reference to the concept of grandmother, prototypically characterised by limited knowledge of techniques of information technology, in I1 “La blockchain spiegata a mia nonna” [The blockchain explained to my grandmother], and the colloquial way in which the question is formulated in the headline of I5 “Cos’è l’algospeak” [What is algospeak]).

From a qualitative and text-linguistic perspective, it is furthermore interesting to note that in most articles where flagging is used, it is consistently applied to all occurrences of the relevant anglicism (F1, F2, F7, F8, F9, F10). In some cases, however, flagging is only used for the first occurrences and then abandoned as the article proceeds (F3, F4). This can be interpreted as reflecting the writer’s assumption that thanks to the meta-information techniques used, the anglicism has been sufficiently introduced and made accessible so that no further flagging is needed later on in the article. Unflagged uses preceding flagged uses can also be found (F5, I1, I2, I8, I9). These uses could reflect the neological status of the anglicisms, which may lead to hesitations concerning the use of flagging. In addition, we can observe that lexical items expressing the concept of being new frequently cooccur with the anglicisms, e.g., “la nouvelle tendance make-up” [the new make-up trend] (F4), “Victimes de la mode: quels sont leurs nouveaux codes ?” [Victims of fashion: what are their new codes?] (F5), “nuova postazione di lavoro” [new workplace] (see the article on shoffice commented on in Section 3).

Concerning the occurrence of other anglicisms in the newspaper articles, the results show high numbers (see Table 3, where the respective numbers of types of other anglicisms – not tokens – are indicated, with different meanings and citation forms being counted separately).^[20] Especially in the French articles, proper nouns, e.g., names for institutions or products, appear to be of high importance (see Table 3, percentages rounded off). The three types of meta-information techniques are also used for other anglicisms in the texts, again with metalinguistic comments being less frequent than the other techniques, and flagging being highly established (flagging is missing in only 2 of the 20 articles). This confirms the overall importance of flagging and frame information. The Italian articles show a higher overall number of other anglicisms and a less frequent use of frame information to enhance their accessibility, e.g., the expressions Great Resignation; frame-rate; squat and round; reskilling, tecnostress and over working are not further explained in I3, I6, I7 and I9, respectively. A special case concerns hyperlinks, e.g., in F4, eye-liner and make-up are highlighted by hyperlinks that direct the reader to other articles on these topics. Moreover, explicit meta-information techniques are also found for other kinds of marked lexical items (e.g., borrowings from other languages such as French tapis roulant in I10, hybrid forms such as English-Spanish fashionistas in F5, or jargon or diasystematically marked expressions such as sape and fringues both meaning ‘duds, rags’ in F5).

When we focus on specific techniques of flagging, it appears that in French quotation marks and italics appear to be to some extent interchangeable as regards the function of alterity marking. At the same time, other principles such as the avoidance of italics in the headlines seem to also guide the use of these techniques (see F1 and F7, where the anglicisms are marked by quotation marks in the headlines, whereas italics are consistently used in the text of the articles). Further investigations would be necessary to confirm these observations for larger data samples. For Italian, some of the articles suggest subtle differences between different highlighting techniques; whereas italics are used for the Latin expression repetita iuvant in I8, the use of quotation marks or bold print prevails for anglicisms. In other articles, cases of hyper-marking are found, e.g., when quotation marks, bold print and italics are used for the expression Great Resignation and quotation marks and bold print are used for the anglicism contractor in I3.