T HE D IASYSTEMATIC S TATUS OF THE D IATOPIC A

This paper is anchored in the tradition of variationist linguistics of Germanic and Romance studies in Europe. In variationist linguistics, the dimension of ‘space’ is essential. According to Flydal (1952:245) this dimension is one of the “extrastructuralismes” which, together with the structures of language, forms the “architecture de langue”. The other important extrastructural dimension mentioned by Flydal is diastratic, i.e. social variation. Coseriu takes over these two notions from Flydal, adding a third dimension: the diaphasic variation (1969:148 ss.). A fourth dimension proposed by Koch & Oesterreicher (1990), opposing the spoken vs written conception, implying the distinction between communicative immediacy or distance (the diamesic variation), was, however, not included by Coseriu. A number of publications and articles have recently questioned the theoretical basis of the three or four dimensions taken over from Flydal by Coseriu and further elaborated in the tradition of variationist linguistics in Romance and Germanic studies. My contribution focuses on the diatopic level, but intends to show, by means of a number of case studies, that this level cannot be considered independently of other variation factors.


Introduction
The present paper is anchored in the tradition of variationist linguistics of Germanic and Romance studies.This tradition is acknowledged especially in German, Italian, and French universities and has been an important scientific paradigm since the second half of the twentieth century.Already Weinreich, Labov & Herzog criticised the then canonical classification of free inter-and intra-speaker variation and demanded more scholarly attention to the "orderly heterogeneity" (Weinreich, Labov & Herzog 1968:100) within language varieties.However, the sociolinguistic approach to language variation favours the analysis of the social context, whereas the variationist tradition of Germanic and Romance studies, initiated by Coseriu and his former students, analyses variation as an element of subsystems of the architecture of language.
In the study of Romance languages, dialectology has held a privileged position for centuries.Therefore, it is natural that the dimension of 'space' is one of the so-called "extrastructuralismes" which, together with the structures of language, form the "architecture of language", according to Flydal (1952:245).The other important extrastructural dimension is diastratic, i.e. social variation, as Flydal notes: Structure et extrastructuralismes forment un ensemble que [...] nous appellerons ici l'architecture d'ensemble de la langue ou simplement l'architecture de langue, en entendant par architecture non pas la disposition architectonique des parties d'un tout, mais un tout systématique formé de parties solidaires (1952:245 -italics in the original).
Coseriu adopts the two dimensions, diatopy and diastracy, adding a third dimension or axis: the diaphasic; he refers to the three as dialect, social level, and style (1969:148 ss.)To these three axes, a fourth was proposed by Koch & Oesterreicher, opposing the spoken versus written conception (the diamesic variation), which should not be confused with the medium of communication: speech versus writing.Instead, it is anchored in the difference between the communication of proximity versus the communication of distance.This dimension was not considered by Coseriu as a separate axis, because he included it in the diaphasic dimension (Dufter 2018:66).
Important questions on these dimensions have been raised in a number of publications and articles, questioning the theoretical basis of the three or four dimensions taken over from Flydal by Coseriu and further elaborated in variationist linguistics, including the following, partly overlapping, questions: 1. Do the diasystematic dimensions have equal importance in the socalled architecture of language?2. Are they universal?3. Is their status permanent or subject to change?
And most importantly for the present paper: Lene Schøsler Scandinavian Studies in Language, 14(2), 2023 (55-72) 57 4. What is the relation between the dimensions: are they mutually dependent or independent? 5. Can they possibly be derived from or related to a superordinate principle?
A theoretical rethinking of the diasystem is necessary in order to understand the inherent variability of language in the light of recent research, as observed by a number of linguists working on variation, for example the authors of a thematic volume entitled Repenser la variation linguistiques (Glessgen et al. 2018). 1 In the present paper, I intend to propose an integrated view of the different axes of variation.
The starting point of my paper is diatopy, with exemplifications mainly from the history of French syntax.I intend to show that diatopy should not be considered independently of other variation factors.Although my exemplifications primarily concern the diatopic axis, I intend to discuss all variation axes and claim (cf. the questions listed above) that diasystematic factors are subject to change, that the dimensions are mutually dependent, and that they can be considered related to a superordinate principle.
The paper is organised in the following way: section 2 presents the traditional understanding of the variation axes; in section 3 I will consider the questions 3 and 4 raised in this Introduction, i.e. the question whether the diatopic axis is permanent or subject to change, and the question whether diatopy can be investigated independently of the other axes.Section 4 discuss the remaining three questions (1, 2, and 5).My conclusion is found in section 5.

The generally accepted presentation of the DIA-axes
Variationist linguistics is certainly not one homogeneous paradigm, but a common denomination of research anchored in the investigation of the interplay between language external and language internal factors, with reference to the diasystematic axes.2A widely accepted, but simplified, presentation of these axes, which will be subject to discussion in the following sections, affirms that there is, in principle, a fundamental difference between the diachronic and the diatopic axes, on the one hand, and the diaphasic and the diamesic axes on the other hand (see Table 1).The difference is rooted in the observation that diatopic and diastratic variation is conceived as variation found between different speakers, i.e. interpersonal variations or "varieties according to users", whereas the diaphasic and also the diamesic variation is found with the same speaker, so they are intrapersonal variations or "varieties according to use". 3able 1. Diasystematic parameters of variation, interpretation by Gadet (2007Gadet ( /2003:23):23), reproduced by Kragh and Lindschouw (2013:8) The idea behind the distinction between inter-and intrapersonal variation is that dialectal and social4 variation are permanent features, in contrast to stylistic and diamesic variations which can be chosen by speakers according to their communicative needs.However, in societies with social mobility, the diastratic level can be conceived as intrapersonal.It should be noted that Gadet (2018:58) rightly specifies that not just any speaker has access to the intrapersonal variation; indeed, higher social position and education provide speakers with a larger range of variation possibilities than socially less favoured speakers.
This generally accepted presentation of parameters of variation fails to include another important aspect of variationist linguistics: the study of text traditions, which is especially intense in Romance studies in Germany and Switzerland (see for example Kabatek 2013).I cannot elaborate on this aspect in the present paper, but I will briefly return to it in section 4. The first part of my research on the diatopic axis is intended to examine whether the traditional assumption concerning its permanent nature is correct, and whether this axis should be investigated independently from the other axes, in order to answer questions 3 and 4 posed in my introduction.

The status of the diatopic axis I
In this section, I will provide a number of examples (mainly from the history of French, in order to answer the questions 3 and 4 asked in the Introduction, i.e. question 3: Is the status of the diatopic axis permanent or subject to change?(see section 3.1) and question 4: Can diatopy be investigated independently of the other axes?(see section 3.2).

Discussion of question 3
My intention here is to show the change of status of diatopy in the course of history, by means of the development of the negation in French.This is a phenomenon well studied by many researchers investigating the determining factors for this change: is it extra-linguistic, intra-linguistic, or both taken together?The starting point was formulated by Otto Jespersen (Jespersen 1917), hence the name "Jespersen cycle".According to him, there is the following evolution (the elements in brackets are optional): Latin non dico 'I don't say' > very Old French (je) ne dis > Old and Middle French (je) ne dis (pas / mie / point etc.) > Modern Standard French (High Standard) je ne dis pas > Modern French (Low Standard) je dis pas.
The change of negation in French is spectacular, it has been examined among others by Price (1962), whose study on the historian Froissart is particularly illuminating.In Froissart's Chronicle, begun around 1379-1381 and whose manuscript dates from the fifteenth century, Froissart takes up a previous Chronicle by Jean le Bel, integrating it into his own text (Price 1962:20-21).The two authors are both of Picardy origin and this dialect tends to prefer the particle of negation mie rather than pas or point.Froissart's books are especially interesting for a diachronic study because the use of the negation particle varies significantly from the first volume of his Chronicle, as shown by Price (1962).Price finds that Froissart uses pas in his first volume as often as mie(s).In the second volume, pas is more frequent than mie(s), and in the third, mie(s) has almost disappeared.In other words, thanks to the study of one single author, adapting and writing in one single textual genre (the historical narrative), using one scriptawhich is rooted in the Picard dialect5 -and of three volumes of texts which undoubtedly were addressed to the same audience, we are able to follow the diachronic change of the negation particles, which in fact consists of a progression of a Central French (francien) dialect form (pas) which spreads at the expense of the original dialect form of Picardy (mie).This diachronic change is equivalent to the extension of a given (Central French) variant (pas), in diatopy.This is therefore an excellent case for the investigation of the correlation of the factors of time and space.
We have seen that the Central French (francien) negation particle (pas) prevails at the expense of the Picardy form (mie). Here, another phenomenon comes into play: the status of the diatopic form -in other words, diastracy.In fact, during the fourteenth to fifteenth centuries, the prestige of the Picardy dialect declined compared to Central French (francien), which explains why the French variant pas replaced the Picardy form mie.This observation necessitates a rethinking of the status of diatopy during the history of French and of the relation between diatopy and diastracy.I will provide three short illustrations intended to cast light on the status and the relation between these two axes, from Old French, from Classical French, and from Modern English.

Discussion of question 4
The present subsection is intended to illustrate my point concerning the interdependence between diachrony, diatopy, and diastracy.In the early Middle Ages, the dialect was considered by the speakers/writers to be a trait closely linked to his/her identity.The implications of the existence of a hierarchy between the diatopic varieties, as clearly seen in the texts, have traditionally been neglected.There is an abundance of evidence to support my claim concerning a hierarchy; e.g. the following quotations, both from the twelfth century: (1) the excuse used by a nun, who is aware that her Anglo-Norman variant is less appreciated than other variants of high prestige (e.g. the Central French variant or francien) which is the variant of the author of (2).
( My second illustration stems from the so-called ‛classic' French period.In the seventeenth and eighteenth centuries, the Grammarians (Remarqueurs) undertake the huge work of describing and codifying the French language with the intention of making it prestigious (‛clear' and ‛logical') by imitating the use of the Court, of good authors, and of learned people.
Vaugelas' Preface remains emblematic: Le bon usage… C'est la façon de parler de la plus saine partie de la Cour, conformément à la façon d'escrire de la plus saine partie des Autheurs du temps ("Good usage … is the way of speaking of the healthiest part of the Court, in accordance with the way of writing of the healthiest part of the authors of our period", my translation).
At this time, diatopic varieties completely lost the prestige they would have had in the Middle Ages.The judgment made on dialects is negative and henceforth the word 'dialect' points to old-fashioned use, which should be avoided, according to the Grammarians.But if ‛dialectal' means ‛oldfashioned', this is a reinterpretation of the term diatopy, which can no longer be conceived as a variation according to the user (interpersonal variation) but, rather, interpreted in a sense which belongs to the variation according to use (intrapersonal variation).The use of the dialect, instead of being an inherent feature, thus becomes a choice of the speaker.
In such a configuration, the speaker can decide to get rid of his or her dialect but also, on the contrary, choose to claim it as a constitutive trait of his or her identity.Let us look closely at these two options: in a situation of strong standardisation of the language, as in the case of seventeenth-to nineteenth-century France, it is the first attitude that dominates, further reinforced by compulsory public education and, later, by electronic media (Le Dû 2000).Dialect usage is then interpreted as a low variant in terms of diastracy.The use of standard speaking is equivalent to a high level of prestige, which thus becomes a means of social progress.
There is, however, an entirely different -and historically more recent -attitude, represented by my third illustration, which is that of claiming a local identity through the explicit preservation of dialect features.This attitude was first brought to light by William Labov in his study of dialects at Martha's Vineyard (1962).Research on dialect features of the south of France, more precisely on the pronunciation of schwa [Ə], points in the same direction (cf.Armstrong & Unsworth 1999).The two authors have shown that the presence of this dialect trait is not only correlated with gender (greater presence in the pronunciation of men than of women), but also with the degree of the speakers' sense of local belonging.Thus, like the opposite attitude (the rejection of the dialect in favour of standard language), the identity claim of a local language becomes a choice made by the speaker.
These illustrations can easily be multiplied: however, they suffice, I believe, to confirm the idea that diatopy may change from an interpersonal to an intrapersonal feature, depending, of course, on the nature of the society in question.
If indeed belonging to a specific place, characterised by a specific way of speaking, remains the original defining meaning of the term diatopy, this is not a geographic or spatial dimension (‛space', in Cresswell 2004) but a linguistic entity in its own right.Diatopy is therefore subject to the effects of linguistic awareness, interpretation by speakers, and identity constructions: Space […] has been seen in distinction to place as a realm without meaning -as a 'fact of life' which, like time, produces the basic coordinates for human life.When humans invest meaning in a portion of space and then become attached to it in some way (naming is one such way) it becomes a place.(Cresswell 2004:20) Recent research in sociolinguistics is oriented in this line of thinkingtowards the reinterpretation of dialect features as a social construction with an identity goal -by, among others, Juillard (2016:97), who defines the term 'sociolinguistic space' as follows: J'entends par 'espace sociolinguistique' une notion qui, d'une part, tient compte tout à la fois des lieux géographiques et/ou socio-symboliques, des situations de communication, des réseaux, des activités et des types de relations interpersonnelles, ainsi que des variétés, langues ou usages, et traits disponibles comme ressources, et les relie dans une non-dualité, et qui d'autre part, implique toujours qu'un espace donné soit relié à d'autres espaces sociolinguistiques, proches ou distants, potentiels, latents ou manifestés.
("By ‛sociolinguistic space' I mean a notion which, on the one hand, takes into account both geographic and/or socio-symbolic locations, communication situations, networks, activities and types of interpersonal relationships, as well as varieties, languages or uses, and traits available as resources, and connects them in a non-duality, and which, on the other hand, always implies that a given space is connected to other sociolinguistic spaces, close or distant, potential, latent, or manifested."my translation) One of the causes presented to explain this evolution is the great mobility of the modern world, having repercussions in the anxiety felt by certain speakers between the global and the local (cf.Auer, Schmidt & Lameli 2010:XI).But the metalinguistic and identity dimension of diatopy has perhaps always been intrinsic to it, since there is no such thing as a completely homogeneous language without variation.
Summing up, with reference to questions 3 and 4 raised in the introduction, my own research, confirmed by that of other scholars, shows that diatopy is not permanent in the sense of being an interpersonal variation.For at least a number of speakers, dialectal features are an option. 6Moreover, this option does not only imply anchoring in the linguistic space, it may also imply what is traditionally considered diastratic and/or diaphasic features: most importantly, markers of individual or group identity.Consequently, diatopy cannot be separated from the other axes -they are mutually dependent.I will explore this point further in the following section.

The status of the diatopic axis II
In the Introduction, three additional questions were asked on the status of the diasystematic dimensions, i.e., question 1: Do the diasystematic dimensions have equal importance?question 2: Are they universal?and question 5: Can they possibly be derived from or related to a superordinate principle?These questions are related and will therefore be discussed together in the following.
In accordance with Koch and Oesterreicher's line of thinking, I believe that the diatopic, diastratic, and diaphasic axes can be interpreted as three continua which correspond to three parameters: space (strongly or weakly marked), linguistic prestige = diastracy (high or low), and context specificity = diaphacy (high or low). 7The three axes should be interpreted as integrated in the architecture of variation headed by the distinction between the two ends of the conceptual scale: language of proximity (Sprache der Nähe) and language of distance (Sprache der Distanz), see Figure 1 (Koch 2003:105): In concrete terms, this implies that strong diatopic features can be interpreted as diastratic marked features of low prestige, which in turn can be interpreted as diaphasic marked features of low style, and typical for the language of proximity.In contrast, week diatopic features can be interpreted as diastratic marked features of high prestige, which in turn can be interpreted as diaphasic marked features of high style, and typical for the language of distance.However, and this is important, such transfer from one axis to another presupposes the existence of a linguistic standard or norm.Indeed, the opposition at level 4, strong or weak diatopy, is measured in contrast to an accepted norm (see, e.g., most recently Oesterreicher & Koch 2016, section 8:1).
Let me return to the illustrations presented in section 3.In the case of negation particles, a change was observed in Froissart's Chronicle from the use of the particle mie, typical of Picardy in the first volume, towards the later, more generally accepted use of Central French pas in the following volumes, mirroring the change of prestige of the dialects: the northern Picardy dialect losing prestige in favour of francien.In other words, even in societies without one single norm, as was the case in Medieval French, the relative prestige of individual dialects had an impact on the choice of variants.
Another study on negation particles in charters from Luxemburg from the thirteenth century (Völker 2007) confirms the importance of the different status of dialects and their relevance for the distribution of variants.Völker distinguishes three social groups of addressees: A) minor nobility and clergy and town-dwellers, B) high nobility and high clergy, and C) the King's chancery.The latter is located in Île-de-France, in contrast to the former, located in Luxemburg or the nearby region.In the first two social groups, the particle mie, typical of this dialectal area, is predominant (77.8% and 75.9%, respectively), whereas it has a low frequency (22.2%) in charters of group C. In contrast, pas is the most frequent negation particle (55.6 %) in the charters addressed to the chancery, but much less frequent in A and B (11.1 % and 10.3 %).These figures show that the local negation particle is chosen for local communication, whereas the central French form is chosen when addressing the chancery located in the dialect favouring the use of pas.This implies that scribes were conscious of different dialectal standards and that they adapted their language use according to these standards.In other words, in languages without one single standard, texts are adapted to conform to the standard of the addressee.Referring to Figure 1, the local dialect, here that of Luxemburg, is avoided only in the charters addressed to the royal chancery of Île-de-France, in order to adapt to the language of distance, in diatopic, diastratic, and diaphasic terms.
Some scholars (e.g.Dufter & Stark 2003;Selig 2011:113-114, 119;Dufter 2018) have claimed that the 'diamodel' is incompatible with the Koch & Oesterreicher model, the latter being universal in nature and as such does not allow integration into the model of the diasystem provided for the variation of a concrete language.8However, as repeatedly argued by Koch and Oesterreicher most thoroughly in their posthumous article (Oesterreicher & Koch 2016:47 ss.), this criticism is not relevant, and I hope to have presented further evidence in the preceding sections in favour of the integrated model.
A final point of criticism to be mentioned here concerns the alleged homogeneity of the diamodel, even inside a social group (see, e.g., Dufter & Stark 2003: 86 ss.).Against this point, the freedom of choice of the individual speaker has repeatedly been put forward by the defenders of the model.An example mentioned both by Dufter & Stark and commented by Koch (2002) and Oesterreicher & Koch (2016) is the unexpected presence of the passé simple in Modern French speaking and writing, for example in newspapers.Another striking case of the sudden reappearance of an old form is the case of moult, originally an adverb or mass noun, replaced by the end of the Middle Ages by the adverb très and beaucoup, respectively.In Modern French, moult has acquired the new function of determiner and is, for the time being, most frequent in the register of journalism, mainly written. 9o, for diachrony, old vs. new forms are strongly or weakly marked and of high or low prestige.In other words, space, prestige, context, and coexisting old and new forms, i.e. diachrony, are relevant parameters for each individual utterance.Thus, any utterance is placed on each of the axes.This is consistent with the variationist tradition: a dialect is a variety of the sociolectal type (= a group of speakers), with a strong diatopic marking (= low range in space).A sociolect is also a variety of sociolectal type (= a group of speakers), characterised by a well-defined linguistic prestige (which can be weak or strong).Finally, a style or register is a variety with a strong diaphasic marking, having linguistic prestige, high or low, and variable diatopic marking.
Summing up: the answer to the questions asked in the title of this section can be put in the following way: all dimensions are relevant for each utterance.They are related to a superordinate principle, which is the continuum of proximity and distance.But, importantly, this principle is individually organised in languages, not universally organised.
Interestingly, Raible (2019) arrives at a similar conclusion in his thorough comparison of two approaches to linguistic variation in different languages, that of Koch & Oesterreicher and that of Biber.In spite of their different starting points, i.e. a theoretical one, as presented in Figure 1, versus a corpus-based one, he concludes that they arrive at comparable results, cf.Raible's quotation from Biber & Conrad (2009:259): A synthesis of previous research on spoken and written registers10 shows three general distributional patterns: (1) linguistic features that are common in informational writing tend to be rare in the spoken registers, and vice versa; (2) spoken registers are surprisingly similar to one another in their typical linguistic characteristics, regardless of differences in communicative purpose, interactiveness, and pre-planning; but in contrast (3) written registers have a wide range of linguistic diversity.
This quotation leads us to take into consideration text types.Any utterance, spoken or written, belongs to a text type, which is more or less marked and responds to more or less defined and definable communicative configurations.From a diachronic point of view, the most recurrent communicative situations give rise to an elaboration of a text type, i.e. a ritualisation of the linguistic formatting, which can lead to the constitution of a tradition of discourse.In synchrony, the text types maintain an interactive relationship with the diasystematic axes -in particular, diaphacy.
A classification of text types is difficult, especially if we consider their diversification in modern times.For example, the type "scientific text" covers very diverse sciences and also reflects very diverse textual models within each discipline: monographs, journal articles, reviews, or even lectures.The type "school book" also covers all subjects taught in primary and secondary education, taking into account the age of the pupils, their level of training as well as the type of education.Nevertheless, any scientific text and any school book is immediately recognisable as such in its tradition of discourse (cf.Biber 1988;Biber & Conrad 2009).
Moreover, classification of text types is subject to strong changes over time.For example, the didactic text type was, from the outset in Antiquity and also in the Middle Ages, conceived as a dialogue between teacher and pupil -this model is no longer used.Another example is letters, a text type which since Antiquity included very stylish letters -those with high spatial reach, high prestige, and a high degree of specificity.But this type of rhetorical letters, often copied and distributed widely, declined in use after the nineteenth century.

Conclusion
The present paper has shown, I hope, that the traditional conception and classification of diasystematic axes, distinguishing interpersonal and intrapersonal variation, is not satisfactory for the analysis of modern language variation, and that it probably never was, as suggested by the illustrations presented in sections 3 and 4. According to the analysis proposed in section 4, the different diasystematic dimensions do not represent oppositions but a co-presence which characterises any utterance at any time.From this perspective, the diasystematic axes gain an explanatory power and provide a tool for the description and interpretation of language data.

Figure 1 :
Figure 1: The linguistic variation and the continuum of proximity and distance 1) Si joe l'ordre des cases ne gart / Ne ne juigne part a sa part / Certes n'en dei estre reprise / Ke nel puis faire en nule guise.
/ Qu'en latin est nominatif / Ço frai romanz acusatif./ Un faus franceis sai d'Angleterre / Ke ne l'alai ailurs quere./ Mais vus ku ailurs apris l'avez / La u mestier iert, l'amandez.(Prologue de la Vie d'Edouard le Confesseur, translation from the twelfth century of a Latin text into Anglo-Norman by a nun, quoted from Schøsler (1984:171)).Free translation into English: "If I do not respect the cases, and do not put together what should go together, I should not be blamed, because I cannot do better.What is in the nominative in Latin, I will make it accusative in French.I know bad French from England, because I never learnt it elsewhere.But you who have learnt it elsewhere, please repair [my grammar], whenever necessary."(2) Mis langages est bons, car en France fui nez.(Vie de S. Thomas Becket, v. 6161-5, quoted from Schøsler 1984:171.)"My language is fine, because I was born in France [i.e.Île de France]".