Language Norm and Usage Change in Catalan Discourse Markers: The Case of Contrastive Connectives

Language norms are dynamic conventions that change over time. In the case of Catalan, the 20th century represents a critical codification period. In this paper, the author discusses the influence of internal and external factors on the evolution of four connectives mainly used in formal (written) communication (nogensmenys ‘nonetheless’, emperò ‘but’, no obstant (això) ‘nevertheless’ and tanmateix ‘however’), which have experienced remarkable change during the 20th century. Three sources of information are considered: dictionaries, grammars, and corpus examples (from Corpus Textual Informatitzat de la Llengua Catalana) in the period 1904–2013. The review of dictionaries and grammars shows that descriptions and norms change over time. The corpus reveals the impact of prescription on the use of the connectives in formal registers. The basic research question asks: to what extent do linguistic and socio-cultural factors have an influence on change in the field of connectives and how do norms and formal use interact? The analysis shows that external factors (i.e., bilingual context and codification processes) in correlation with internal ones (i.e., paradigmatic relations) affect both linguistic change and norm change. The evolution of the four connectives analyzed is the result of a changing language norm forbidding or encouraging the use of certain forms, almost always triggered by an intent to compensate for the influence of the dominant language, Spanish.


Introduction
Language norms are dynamic conventions that change over time. As Amato et al. (2018, p. 8260) point out, Broadly speaking, a convention is a pattern of behavior shared throughout a community and can be defined as the outcome that everyone expects in interactions that allow two or more equivalent actions (e.g., shaking hands or bowing to greet someone).
Norms change because of language use and also as a result of various external factors. The driving forces of norm change can be either formal institutions, such as language academies, or informal institutions, such as mass media, but change can also be unregulated. Amato et al. (2018) propose an evolutionary modelling approach to norm change in Spanish and English from 1800 to 2008. After analyzing the evolution of a group of variants, (mostly spelling variants) they conclude that the changes driven by institutions tend to be quick, especially in the case of formal institutions: "the old convention is rapidly abandoned in favor of the new one" (Amato et al. 2018, p. 8261). Informal institutions also favor a quick change but not as much as formal institutions, and the process is not homogenous for all cases of the same norm. When the driving force is not an institution, the transition is slower, and there is "some asymmetry between the two forms, such as the presence of a small fraction of individuals committed to one of the two alternatives" (Amato et al. 2018, p. 8261).
In all cases, norms and formal use interact. On the one hand, explicit prescriptions ban some forms and constructions while promoting others, and this has a clear influence As Table 1 shows, Catalan and Spanish exhibit a similar paradigm of opposition parenthetical connectives, but there are also some interesting differences that account for some of the changes in the norm about the connectives under analysis. Almost all Catalan opposition connectives have a direct lexically related counterpart in Spanish. Only però, as a parenthetical connective (and the variant per (ai)xò), 2 tanmateix, and nogensmenys do not have a direct or verbatim correspondence in Spanish. As for Spanish, only sin embargo lacks a direct lexically related counterpart in Catalan. 3 Most parenthetical connectives are grammaticalized units derived from pre-existing lexical units that have experienced semantic change and/or category change (see, e.g., Traugott and Dasher 2002;Heine 2013). However, this is not the only way that a connective appears in a language. As Pons Pons Bordería (2008) and Garachana (2019b) have argued for esto es ('this is') in Spanish and no obstant això ('notwithstanding') in Catalan, respectively; loans and calques are frequent sources for discourse markers. 4 During the 20th century, Catalan experienced key changes in a general context of diglossia regarding Spanish. Periods of revival have followed periods in which Catalan was forbidden and almost erased from formal use, even though informal use was general. The end of Franco's dictatorship opened a new scenario where Catalan could be used in formal contexts, including public communication, and could be taught at schools. However, Catalan has strived to survive, especially in formal communication, in competition with Spanish as the dominant language in a more or less diglossic situation, depending on the territory and period.
Connectives, especially those used in formal (written) contexts as the ones analyzed here, have experienced a direct influence from the Spanish system, because until recently, most writers were educated exclusively in Spanish or mostly in Spanish. As a reaction to this diglossic situation, the standardization of Catalan has also led to banning some forms, which is considered as a result of interference.

Historical Background: The Codification of Catalan
Catalan is a recently standardized Romance language-the process started during the first third of the 20th century and consolidated by the last third of the century. The changes related to the standardization process of Catalan and the influence of Spanish are key elements to be considered when analyzing the interaction between prescription and use. Spanish was-and still is-the dominant language in most Catalan-speaking territories. Over the decades, Spanish was almost exclusively used in writing by most of the population and was the only option in education and other formal communication environments. The pressure of Spanish over Catalan, its lexical, syntactic, and phonetic interference, and the growing insecurity of the speakers as to what is genuinely Catalan, cannot be ignored as key factors influencing Catalan norm and formal use. Thus, the description and prescription included in dictionaries, reference grammars, and normative documents have special importance in order to analyze the data.
For the sake of brevity, only some highlights in the codification process of Catalan will be provided. 5 Catalan, once a consolidated language of culture, lost areas of formal use due to political developments leading to a progressive integration of its territories under Spanish rule. After a period called "Decadència" ('Decadence'), referred to decay of use in formal-mainly written-registers, the 19th century opens up a new perspective under the nationalist ideology of Romanticism. Catalan was very much alive in informal contexts-most of the population was Catalan monolingual-but the language was almost absent from 'high' culture. The Romanticism spirit boosted a revival of Catalan culture and language in formal uses generally known as "Renaixença" (literally, 'Re-birth'), which made it possible to start the codification of Catalan as a modern language by the end of the 19th century and the first third of the 20th century. The main landmarks in this critical period are the following (Rafanell 2020a(Rafanell , 2020b: These landmarks in the codification of Catalan occurred simultaneously with a significant increase in formal uses and printed material. By the 1930s, Catalan not only had a corpus of normative reference documents, mainly created by Pompeu Fabra, the 'Architect of Modern Catalan' (see Costa Carreras 2009 and also IEC-Institut d'Estudis Catalans 2020), but also gained legal recognition and cultural vitality, as Rafanell points out: With the Republican revolution in April 1931, the Catalan language obtained the legal recognition that it had been lacking for more than two centuries. The Statute of Autonomy voted for by the Catalan people established it as the only official language for internal purposes, although the text finally approved by the Spanish Parliament would see it share official status with Spanish. (Rafanell 2020a, p. 540) . . . by 1930 there were 27 newspapers being published in correct and corrected Catalan. There were also two radio stations broadcasting in the language of the country. (Rafanell 2020a, p. 540) Rafanell (2020a, p. 541) reports on Fabra's summary of the achievements regarding language codification during the first third of the 20th century: Although the refinement 'is still unfinished' ("encara està inacabada") and its dissemination 'was just beginning' ("tot just comença"), Fabra concluded his speech in front of the most conservative audience of Catalanism by saying 'today we can be glad to have achieved not only this orthographic unity, but also that linguistic unity which has made possible to elevate Catalan to the official Language of Catalonia [ . . . ]. Fabra's words were pronounced in 1934. The Spanish civil war (1936-1939) and Franco's dictatorship (1939-1975 put a hold on this successful codification process, even though the bases for a standardized language were solid and speakers and cultural agents kept using Catalan to the extent that it was possible in a state that forbade the public use of languages other than Spanish. Franco's death in 1975 and the advent of democracy started a new period in which Catalan regained co-official language status, which resulted in the introduction of Catalan in teaching, use extension, and the actual beginning of the standardization process. In this period, among other important facts, the newspaper Avui was founded (1976), and three regional Autonomous Laws declaring the co-officiality of Catalan, among other changes towards self-government, were approved (Catalonia in 1979;Valencian Country in 1982;Balearic Islands in 1983). As a result of the new legal status, Linguistic Normalization Acts were passed (Catalonia in 1983;Valencian Country in 1983;Balearic Islands in 1986), and TV and Radio emissions started in Catalonia (TV3, 1983) and in Valencia (RTVV, 1989), which was a milestone in the standardization process.
Since 1990, Catalan has further expanded its formal uses and the body of prescriptive documents has regularly been increased and updated. In 1995, the new official dictionary, Diccionari de la llengua catalana (DIEC second edition, 2007), was published. Updating of the official grammar took longer. In 2016, the new grammar (Gramàtica de la llengua catalana, GIEC) and orthography (Ortografia catalana, OIEC) were published. Finally, two versions of the new grammar (GIEC) appeared: First, Gramàtica essencial de la llengua catalana, an online version of the reference grammar (GEIEC, 2018, geiec.iec.cat access on 25 September 2021), and, second, a reduced, printed version (Gramàtica bàsica de la llengua catalana, GBU, 2019) aimed at a less-specialized audience, especially schools.

The Origin and Evolution of Catalan Contrastive Connectives: Previous Contributions
The evolution of connectives from Latin into Catalan is the result of different mechanisms, such as phonetic adaptation, meaning change or a grammaticalization process implying both meaning and grammatical change, as shown in Cuenca (1992Cuenca ( -1993. In that paper, the general evolution of connectives is illustrated through the analysis of the contrastive markers: mas 'but', ans 'on the contrary', però 'non-exclusive but', sinó 'exclusive but,' and encara que 'although'. This account only relies on internal factors of change and explores how meaning change takes place through the conventionalization of a new (metaphorical) meaning that coexists, at least for a period of time, with the old meaning, and eventually triggers category change (from lexical word or phrase to conjunction or parenthetical connective).
Apart from this general study, the diachrony of Catalan connectives has remained unexplored until recently, a gap that Martínez (2017Martínez ( , 2018Martínez ( , 2019Martínez ( , 2020 and Garachana (2019a, in print) have contributed to fill in. Martínez (2017Martínez ( , 2019 describes the diachrony of noresmenys 'nothing less/nonetheless' and nogensmenys 'nonetheless'. The author shows the intertwined history of these two contrastive markers. Noresmenys had two values in Old Catalan: the additive and the concessive. The concessive meaning appeared by the end of the 15th century in contexts in which noresmenys occurred in correlation with concessive conjunctions (Martínez 2019, p. 86;Garachana 2019a, in print: Section 38.5.2.6). The concessive meaning replaced the additive meaning by the end of the 16th century. Noresmenys disappeared in the 17th century but reappeared later on for a certain period of time. However, Fabra banned its use in several normative works and proposed nogensmenys as an alternative (Martínez 2019, pp. 87-88).
. . . com és sabut, no res menys ha estat usat molt, en el català modern, com a conjunció adversativa, ús que nosaltres hem combatut i que sembla, en fi, comptar ja amb ben pocs partidaris, almenys entre els bons escriptors, que, com a rèplica a les conjuncions del tipus nientemeno, néanmoins, semblen ja preferir no gens menys a no res menys. (Fabra 1926;apud Mir and Solà 2011, p. 747) . . . as generally known, no res menys has been very much used in Modern Catalan as an adversative conjunction, which we have fought against, and now it seems to have very few supporters, at least among good writers, who, as a correspondence to conjunctions of the type nientemeno, néanmoins, already seem to prefer no gens menys rather than no res menys. Martínez (2019, p. 93) suggests a number of possible reasons why Fabra explicitly considered noresmenys as non-normative and prescribed the use of nogensmenys instead: (i) the simplification of the meaning of noresmenys to addition, ignoring that it can also be contrastive in Old Catalan, (ii) the tendency to avoid a calque of the Spanish expression nada menos 'nothing less'; (iii) the singularity of gens 'any/no', as a specific Catalan quantifier, as opposed to res (corresponding to Spanish nada 'nothing'), and (iv) the parallelism of nogensmenys with markers in French (néanmoins) and Italian (nondimeno/nientedimeno). 6 Garachana (2019a, in print: 38.5.2.1) briefly describes the evolution of emperò, a contrastive marker already used in the first document in Catalan, Les Homilies d'Organyà. The connective emperò was very frequent in the Middle Ages, but its use decreased since the Languages 2022, 7, 66 7 of 26 16th century in favor of the simpler form, però, which imposed itself in the 17th century, while emperò began to be considered as an archaic form. Garachana (2019a, in print: 38.5.2.4) offers an account of the use of the connective no obstant (això) 'notwithstanding (this)' until the 18th century. The author argues that no obstant (això) does not appear via a grammaticalization process in Catalan or in Romance but is the result of language contact and the influence of rhetorical norms affecting religious texts (Garachana 2019a, p. 150). Absolute present participle constructions with non obstante are documented since the 12th century in Medieval Latin, mostly in Scholastic texts. The construction spread by means of law texts, probably through Italian, which was very influential during the Renaissance. This hypothesis, which contradicts previous accounts of the form, 7 is supported by several facts: (a) No obstant constructions are first documented in 12th century texts in Latin. (b) Most Western Romance languages incorporate this marker almost simultaneously, by the second half of the 14th century, e.g., Spanish no obstante, Italian nonostante, French nonobstant, and Portuguese não obstante (Garachana 2019b, pp. 141-42). In the case of Catalan, the first attested occurrence is a 14th century translation of Boccaccio's Il Corbaccio. The fact that the Romance developments, except for Italian, do not follow the general phonetic evolution, implying the simplification of consonants (bs > s), but keep the Latin structure, is a strong argument for considering its introduction as a grammatical calque, as opposed to a grammaticalization process, which usually implies phonetic reduction and adaptation to the phonetic system of the target language. (c) Constructions with no obstant spread from theological texts to law texts-most lawyers were also clergymen-because they were a useful rhetorical device in argumentation and gave a 'Latin-like' flair to the text, which was fashionable in 14th and 15th century texts (Garachana 2019a, p. 156; in print: Section 38.5.2.4).
In conclusion, the origin of no obstant això is to be found in a calque from Medieval Latin in the context of the Italian Renaissance in the 14th century. 8 During the next two centuries, its use extended to law texts. In the 17th century, its use increased in frequency and the simplified variant lacking the demonstrative açò/això ('this') spread, a tendency that consolidated in the 18th century. Martínez (2018Martínez ( , 2019 reviews the history of the marker tanmateix ('however', literally 'so the same'), which derives from a former comparative quantifier ((ai) tant mateix 'so many/much amount') and dates back to the 14th century. Tanmateix developed its current use as a parenthetical connective in the 17th century, first with an additive value, referred to quality ('in the same way'), then also with a value of confirmation (agreement or surprise) in the 18th century, especially in informal dialogic texts, and, finally, a value of contrast emerged. The contrastive meaning disappeared at the end of the 18th century but reappeared by the end of the 19th century and became predominant only in the 20th century.
Per tant, a la darreria dels segles XIX-XX, dels dos valors vius de tanmateix [CONFIRMACIÓ] [CONTRAST]-s'incorpora al model normatiu el segon. No es tracta pas d'una creació ex novo del llenguatge literari català, sinó de la potenciació d'un dels matisos en joc que, en última instància, corresponia a una cadena "natural" de valors des d'antic: des del valor de [CÀLCUL] fins al valor de [CONTRAST] (Martínez 2019, p. 126) Therefore, by the end of the 19th-20th centuries, one of the two actual values of tanmateix [CONFIRMATION] [CONTRAST] is incorporated to the norm model, namely, the second one. It is not an ex novo creation of Catalan literary language, but the promotion of one of the nuances at play, which, all in all, corresponded to an old 'natural' chain of meanings: from the meaning of [CALCULATION] to the meaning of [CONTRAST]. The previous contributions offer a detailed analysis of the diachrony of the four contrastive markers dealt with in this paper (nogensmenys, emperò, no obstant (això), and tanmateix) by focusing on their origins in Old Catalan and their later development. The research in this paper focuses on their use and evolution in the 20th century (since 1904) and the beginning of the 21st century (until 2013), which means an extension of the analysis into more recent times. It also adopts a socio-linguistic approach, complementary but different from the diachronic perspective in the papers summarized in this section, by confronting the information coming from descriptive and normative documents (i.e., dictionaries and grammars) and corpus examples.

Materials and Methodology
The analysis of the markers described in this paper considers three sources of information, namely: dictionaries, reference grammars, and corpus examples.
The dictionaries analyzed are the three most relevant lexicographic works from a descriptive and normative point of view (see the complete data in the References section): the Diccionari de la llengua catalana (DIEC, 1995(DIEC, /2007, which is the normative dictionary issued by the Institut d'Estudis Catalans, 9 the Diccionari català-valencià-balear (DCVB), a comprehensive descriptive dictionary written by Alcover and Moll and published in 10 volumes from 1930 to 1962, and the Diccionari descriptiu de la llengua catalana (DDLC, 1998(DDLC, -2016, a modern descriptive online dictionary that extends DIEC's information derived from IEC's reference corpus (CTILC).
The grammars analyzed are a selection of the most relevant contributions to the description and prescription of Catalan (see the complete data in the References section). 10 The selection includes: Fabra's grammars (1912Fabra's grammars ( , 1918Fabra's grammars ( /1933Fabra's grammars ( , 1956, Badia i Margarit's Gramática catalana (1962), and Gramàtica de la llengua catalana (1994), and Sanchis Guarner's Gramàtica valenciana (1950). Also, some later updated reference works are considered: The two volumes published under the auspices of the Institut Interuniversitari de Filologia Valenciana aimed at establishing Catalan norms in the Valencian territories (Lacreu's Manual de l'estàndard oral, 1990, andGuia d'usos lingüístics, (GUL, 2002), and the descriptive Catalan grammar directed by Solà, Gramàtica del català contemporani (GCC, 2002). Finally, the new IEC's grammars are the most recent and updated normative grammars in Catalan: Gramàtica de la llengua catalana (GIEC, 2016), Gramàtica essencial de la llengua catalana (GEIEC, 2018), an adapted on-line version of GIEC, and Gramàtica bàsica de la llengua catalana (GBU, 2019), a simplified printed version of GEIEC. These grammars show the evolution of the prescription about the connectives analyzed here, information that can be compared with their evolution in the corpus.
The third source of information is the Catalan language reference corpus, Corpus Textual Informatitzat de la Llengua Catalana (CTILC, https://ctilc2.iec.cat/scripts/, accessed on 25 June 2021) sponsored by the Institut d'Estudis Catalans. The corpus includes written texts from the 19th century (since 1833), but it is mainly constituted by texts of the 20th century and has been expanded to include texts produced as recently as 2018. The texts are literary (fiction, poetry, plays, and so on) and non-literary (essays, press, textbooks, and so on). The analysis focuses on the period 1904-2013 (ca. 70 million tagged words in this period, excluding proper nouns and non-analyzable units). The data in the corpus can be retrieved by year/period, author, work, linguistic variety among Catalan, type of text (either literary or non-literary), and origin (original or translated), which helps to identify relevant variables underlying some frequencies.
The data from the corpus was collected and analyzed as follows. Every connective (and  (11,193 examples). The quantitative data were analyzed considering the 21 periods in which the corpus is divided between 1904 and 2013: an initial ten-year period, 1904-1913, followed by 20 5-year periods thereafter. Not all the periods include the same number of words and, therefore, the occurrences have been normalized per million words in order to properly compare the data. The raw frequency (i.e., number of cases identified in the corpus) and the relative frequency of the four connectives (i.e., number of the cases if the number of words per period was the same) is shown in Figure 1. 11 literary or non-literary), and origin (original or translated), which helps to identify relevant variables underlying some frequencies.
The data from the corpus was collected and analyzed as follows. Every connective (and its variants) was searched for in the corpus and all the examples were retrieved and included in a database. The total number of examples is 18,340: nogensmenys (557 examples), emperò (1226 examples), no obstant això/no obstant/això no obstant (5364 examples), and tanmateix (11,193 examples). The quantitative data were analyzed considering the 21 periods in which the corpus is divided between 1904 and 2013: an initial ten-year period, 1904-1913, followed by 20 5-year periods thereafter. Not all the periods include the same number of words and, therefore, the occurrences have been normalized per million words in order to properly compare the data. The raw frequency (i.e., number of cases identified in the corpus) and the relative frequency of the four connectives (i.e., number of the cases if the number of words per period was the same) is shown in Figure 1 The variables period/year and author have been systematically analyzed in order to interpret the data. All examples for each period have been counted and compared in order to establish the evolutive tendencies. A yearly analysis has also been conducted to more precisely individuate turning points for frequency increase or decrease. In addition, the authors, using each marker, were checked, and special attention was paid to grammarians. In fact, the comparison of the prescription coming from grammars and the current use of a marker by a grammarian proves highly relevant and points to some interesting inconsistencies.
Finally, the information in dictionaries and grammars was confronted with the corpus data in order to provide a complete picture of the evolution of the markers and the interaction between explicit norms and formal use.

The Rise and Fall of Nogensmenys
Nogensmenys is a parenthetical connective used in sentence initial (1) and medial (2) position, both introducing independent sentences (1-2) and clauses inside a sentence (3). It can follow a conjunction, mainly i 'and' (3a) and però 'but' (3b). The variables period/year and author have been systematically analyzed in order to interpret the data. All examples for each period have been counted and compared in order to establish the evolutive tendencies. A yearly analysis has also been conducted to more precisely individuate turning points for frequency increase or decrease. In addition, the authors, using each marker, were checked, and special attention was paid to grammarians. In fact, the comparison of the prescription coming from grammars and the current use of a marker by a grammarian proves highly relevant and points to some interesting inconsistencies.
Finally, the information in dictionaries and grammars was confronted with the corpus data in order to provide a complete picture of the evolution of the markers and the interaction between explicit norms and formal use.

The Rise and Fall of Nogensmenys
Nogensmenys is a parenthetical connective used in sentence initial (1) and medial (2) position, both introducing independent sentences (1-2) and clauses inside a sentence (3). It can follow a conjunction, mainly i 'and' (3a) and però 'but' (3b).
The IEC's descriptive dictionary (DDLC) differentiates four uses of nogensmenys (sometimes written as no gens menys). 12 (4)  All its meanings are described by synonymy and illustrated with corpus examples. Meanings 1 and 4 correspond to contrast-concession (tanmateix 'however', no-res-menys 'nonetheless'), and do not seem to be different. Meaning 2 corresponds to additive marking (a més 'moreover') and meaning 3 corresponds to an emphatic adverb (ni més ni menys 'nothing less').
Even if nogensmenys has been mistakenly considered as an archaism (Badia 1962, II: p. 97 ;Lacreu 1990, p. 327;Serra andPrunyonosa 2002, p. 2020), it is a 'neologism' introduced at the beginning of the 20th century by Pompeu Fabra to substitute noresmenys, an old form also reintroduced in that period (see Section 3). Fabra considered noresmenys as a calque of Spanish nada menos 'nothing less' (meaning 3 in DDLC), a use that can still be found in the corpus (5), although residually.
Most normative dictionaries only include the contrastive use of nogensmenys: The definition in (6) shows this reduction of its polysemy, since nogensmenys is defined in DIEC by a concessive paraphrase "Without what has been just said or what is implicit being an obstacle." Normative grammars show differences in the description and norm about nogensmenys. As already indicated, Fabra proposes nogensmenys as a neologism created to become an alternative to noresmenys in contrastive contexts (Martínez 2019, pp. 87-88). 13 This proposal was followed by authors and grammarians, and noresmenys disappeared from formal use.
Badia (1962, II, pp. 100, 233) includes nogensmenys as a contrastive-concessive marker along with tanmateix, amb tot, així i tot, no obstant això, and the parenthetical connective però (used in non-initial position). However, previously in the same grammar (Badia 1962, II, p. 97), Badia indicates that nogensmenys is archaic. In his 1994 grammar, Badia (1994, p. 314) considers tanmateix and nogensmenys as equivalent, at least in some senses, and qualifies both connectives as formal. 14 The modern IEC grammars include nogensmenys as an opposition parenthetical connective while indicating that it is very formal. Other grammatical works do not include this connective (e.g., GUL, 2002;

, based on contemporary corpus examples).
What is the evolution and current use of nogensmenys? The first documented use as a concessive marker dates back to a 1915 translation of Shakespeare's Corolianus (Martínez 2019, pp. 51, 53). Fabra's initiative to promote nogensmenys had an instant reflex in use at the beginning of the 20th century but had a fluctuant evolution thereafter and finally failed, as shown in Figure 2: This proposal was followed by authors and grammarians, and noresmenys disappeared from formal use.
Badia (1962, II, pp. 100, 233) includes nogensmenys as a contrastive-concessive marker along with tanmateix, amb tot, així i tot, no obstant això, and the parenthetical connective però (used in non-initial position). However, previously in the same grammar (Badia 1962, II, p. 97), Badia indicates that nogensmenys is archaic. In his 1994 grammar, Badia (1994, p. 314) considers tanmateix and nogensmenys as equivalent, at least in some senses, and qualifies both connectives as formal. 14 The modern IEC grammars include nogensmenys as an opposition parenthetical connective while indicating that it is very formal. Other grammatical works do not include this connective (e.g., GUL, 2002;
Nogensmenys is used 557 times in the period (171.7 cases per million words) by 183 authors. Since its introduction by Fabra in the first third of 20th century, its use increased steadily from 1905 to 1933 and in the 1959-1983 period, especially between 1964 and 1983. From 1984 on, the connective started a steady decrease and since 1999, the corpus includes less than 16 occurrences per 5-year period. 15 In brief, the new connective purposely added to the list of concessive parenthetical connectives as a neologism becomes a marked form tied to formal or even archaic registers. As Martínez points out, this is an interesting case of an artificial introduction tied to the language codification process leaded by Pompeu Fabra and followed by many authors and grammarians alike.
Nogensmenys, tot i romandre en un àmbit d'ús altament formal, és una mostra d'una proposta d'incorporació d'un element nou en el moment d'elaboració de la norma del català contemporani normatiu i d'exclusió d'un altre element tradicional, el vell noresmenys. (Martínez 2019, p. 96) Despite remaining as highly formal in use, nogensmenys illustrates a proposal to incorporate a new element at the time of elaboration of contemporary normative Catalan while excluding another traditional element, the old-fashioned noresmenys.
Some dictionaries describe nogensmenys as a polysemous item (DDLC), transferring the various traditional uses of noresmenys to the new form, but most dictionaries and grammars only account for its contrastive meaning. More modern grammatical works reflect its decrease in use and restrict its conditions of use or even exclude it from the list of contemporary contrastive connectives. In the end, Fabra's strategy somehow failed. Noresmenys disappeared from current use, but, after a period of uneven increase, nogensmenys followed a similar path.

The Mysterious Case of Emperò
Emperò is a reinforced variant of però, both as a conjunction (sentence or clause initial, (7)) and as a parenthetical connective (medial (8a) or final position (8b)).  All dictionaries define emperò as a secondary form of the basic contrastive conjunction però 'but', both as a conjunction and as a parenthetical connective. DIEC defines it as equivalent to the conjunction però ('but') in its three connective uses (però 1, però 2, però 3; a fourth use refers to its lexicalization as a noun, peròs 'objections').
In Old Catalan, there is, finally, the concurrent form emperò, that appears very often in [Cancelleria's] documents substituting intermediate però and less often substituting initial però, which does not allow us to establish a difference in meaning between the two forms. Since però and emperò are equivalent, the preservation of both forms would be of no use, and so modern writers have recognized it, rightly deciding in favor of the exclusive use of però.
Along the lines of Fabra, Badia (1962, II, p. 97) explicitly says that emperò should be substituted in favor of però: "tampoco incorrecta, pero igualmente poco recomendable, aunque tiene un uso muy generalizado en algunas comarcas, como en tierras de Gerona, con el valor del simple però" ('which is not incorrect, but not much advisable, even it has a very general use in some areas, such as Girona, with the value of the simple però'). Surprisingly, Badia himself makes a relatively extensive use of emperò, even in the last third of the 20th century (30 cases in 4 different books published in 1976, 1979, and 1981). In the same vein, Sanchis (1950, p. 285) also says that però is the preferred variant ("emperò sinònim de però, encara que menys recomanable" 'emperò is a synonym of però, but less advisable'), but he uses emperò extensively (55 times in his 1950 grammar), only as a conjunction and mainly introducing an independent sentence, after a full stop.
The evolution of emperò in the period analyzed is shown in Figure 3. Emperò had a significant use at the beginning of the 20th century; 1226 occurrences (equivalent to 396 occurrences per million words), of which 370 cases (140.6 pmw), belong to the 1904-1923 period. In the corpus, emperò is used by 321 authors. As a consequence of the new explicit norm by Fabra, its use fell to less than 100 cases since 1926, with occasional periods of relative increase that are never higher than 41 occurrences per million words and are often lower than 20 or even 10 cases. It must be pointed out that, in the latter periods, the occurrences tend to concentrate in one or a few works. 16 The final period, 2009-2013, only includes 11 cases (2.1 pmw). Nowadays it is only heard spontaneously in some areas of Eastern Catalan (Badia 1962, II, p. 97).
The evolution of emperò in the period analyzed is shown in Figure 3. Emperò had a significant use at the beginning of the 20th century; 1226 occurrences (equivalent to 396 occurrences per million words), of which 370 cases (140.6 pmw), belong to the 1904-1923 period. In the corpus, emperò is used by 321 authors. As a consequence of the new explicit norm by Fabra, its use fell to less than 100 cases since 1926, with occasional periods of relative increase that are never higher than 41 occurrences per million words and are often lower than 20 or even 10 cases. It must be pointed out that, in the latter periods, the occurrences tend to concentrate in one or a few works. 16 The final period, 2009-2013, only includes 11 cases (2.1 pmw). Nowadays it is only heard spontaneously in some areas of Eastern Catalan (Badia 1962, II, p. 97).

The Deictic Anchoring of No obstant (això)
This contrastive marker has three possible forms: no obstant (13), això no obstant (14), and no obstant això (15), which exhibit different evolution and changing normative consideration. The three variants can occur in sentence or clause initial position, sometimes preceded by a conjunction (mainly i 'and', però 'but' and que 'that') and also in medial position.   We can see that the normative dictionary (16) locates first the variants with the demonstrative no obstant això/això no obstant and the simple variant no obstant follows, although they are treated as equivalents and defined synonymically. The descriptive dictionary (17), which is more modern, introduces the connective through its simple variant (no obstant) and adds the variants with the demonstrative as secondary forms.
Thus, DIEC establishes a somehow circular definition by synonymy: no obstant (això) > tanmateix > nogensmenys. As for the variants, DDLC seems to consider no obstant as the basic unit and the forms with the demonstrative (no obstant això/això no obstant) as variants. In contrast, the first edition of DIEC, following Fabra's Dictionary, did not accept the variant without the demonstrative. Nowadays, the normative dictionary includes no obstant as a second possibility.
DCVB considers the prepositional uses of no obstant and the conjunction no obstant que as a second meaning of obstant, the present participle of the verb obstar ('impede', 'be an obstacle'). The first meaning ('Although, without being an obstacle') corresponds to the conjunctive and prepositional uses, whereas the second one (b) is that of the connective ('despite this', 'nonetheless'). This dictionary includes the forms with or without the demonstrative as alternatives, but the variant without això (no obstant) is in the first position and, hence, it cannot be considered as a secondary variant.
Grammars describe no obstant això as a contrastive-concessive marker related to other connectives, mainly tanmateix, nogensmenys, amb tot, and així i tot (e.g., Espinal 2002Espinal , p. 2788. From a prescriptive perspective, Fabra (1912) lists no obstant as a coordinator. However, in his posthumous grammar, only the variant with the demonstrative, no obstant això, is included (Fabra 1956, p. 84). Interestingly, Fabra uses the connective only once in this work and the form is no obstant, as in many of the letters he wrote.
Incorrecta ha de considerarse la conjunción no obstant [n. a.], que presenta hoy a veces un cuadro de usos parecido a los de las locuciones castellanas "no obstante" y "sin embargo". (Badia 1962, II, p. 97, Fn 12) The conjunction no obstant [non-acceptable] must be considered incorrect; nowadays it often exhibits a series of uses similar to the Spanish expressions "no obstante" and "sin embargo".
(2) Needless to say that the clearly foreing expression *sin embargo, too frequent in many people's speech, is even worse.
Badia's description suggests the real reason to ban no obstant, that is, its verbatim correspondence with the Spanish connective no obstante. Informal Catalan tended to use the loan sin embargo, which is rightly considered incorrect, but the prejudice against no obstant is not based on historical data and cannot be justified on genuinity grounds.
Later grammars do not exclude the form without the demonstrative, which, as already said, was already possible in Old Catalan, but label the variant with the demonstrative as preferent in formal registers. For instance, GUL points out that the omission of the demonstrative is not advisable (2002: 153). Similarly, the grammar portal Optimot (www. optimot.cat access on 3 July 2021) indicates that the variant with això is more frequent in formal registers and points out that until IEC's 2016 grammar, no obstant was not accepted as normative.
Other works remind the reader that the variant without the demonstrative, even if not accepted, was gaining social acceptance.
Lacreu provides examples of remarkable authors and newspapers to make his point. In fact, as Garachana (2019a, in print: 38.5.2.4) shows, the use of no obstant was nothing but new and it was the predominant variant for decades. Previous grammarians accepted the form without the demonstrative, based on extensive use by well-known authors (e.g., Moll or Coromines). In fact, no obstant is the only variant included by Sanchis (1950, p. 285) in a list of corrective conjunctions, along with nogensmenys, tanmateix, amb tot, and altrament.
What does the corpus say? The marker is used 5364 times in the period (1584.8 cases per million words) by 1130 authors, including many cases in newspapers, with no specific indication of author. Figure 4 shows the evolution of the use of the marker, regardless of the variant. La reducció de la locució no obstant això a no obstant, encara que no està acceptada normativament, té cada vegada més un major grau d'acceptació social. (Lacreu 1990, p. 327, Fn 3) The reduction of the expression no obstant això into no obstant, even if it is not normatively accepted yet, is gaining social acceptance.
Lacreu provides examples of remarkable authors and newspapers to make his point. In fact, as Garachana (2019, in print: 38.5.2.4) shows, the use of no obstant was nothing but new and it was the predominant variant for decades. Previous grammarians accepted the form without the demonstrative, based on extensive use by well-known authors (e.g., Moll or Coromines). In fact, no obstant is the only variant included by Sanchis (1950, p. 285) in a list of corrective conjunctions, along with nogensmenys, tanmateix, amb tot, and altrament.
In brief, corpus examples, both old and modern, do not confirm-or even contradictmost grammarians' prescriptions aimed at eliminating the variant without the demonstrative. Yet, the norm had an influence on the reduction of use of no obstant, which accounts for the triumph of no obstant això over the simple variant during the 20th century and the beginning of the 21st century.
The preference for no obstant això is not based on the traditional Catalan use of the connective but is the result of a prescriptive intent to differentiate Catalan from Spanish (maybe considering the Italian parallel connective ciò nonostante). The recent normative acceptance of no obstant will probably imply the increase of a variant that has been long forbidden or hardly tolerated because of its verbatim correspondence to the Spanish no obstante.
(20) Al Llibre dels fets de Jaume el Conqueridor apareix un cavaller Bernat de Rocafort, que ha estat identificat per Ferran Soldevila amb un majordom de la cúria reial del mateix nom que acompanyà el monarca a la campanya de Mallorca del 1228-1229. No hi ha, tanmateix, cap indici, fora de la coincidència del nom, que permeti relacionar aquest cortesà del segle XIII amb el nostre guerrer. (MARCOS HIERRO, ERNEST: Almogàvers. La història, 2005) In Llibre dels fets by James the Conqueror there is a knight named Bernat de Rocafort, who Ferran Soldevila identifies with a butler of the royal curia who accompanied the monarch in the campaign of Majorca in 1228-1229. There is, tanmateix ('however'), no indication, apart from the coincidence of the name, that this 13th-century courtier can be related to our warrior. The dictionaries differentiate several meanings of tanmateix. DCVB includes two adverbial uses (1. Realment, de veres 'really, certaintly'; 2. Serveix per a fer l'afirmació d'un fet expressant ensems una limitació de tal fet o l'existència de certes dificultats que s'oposaven a la seva realització 'It is used to assert a fact while indicating a restriction of that fact or the existence of certain difficulties to its realization') and a connective use (3. De totes maneres 'nevertheless').
Nowadays, the adverbial uses (1 and 2) occur only in either very formal registers or informally in some small areas of Northern Catalan (Martínez 2020). Conversely, the adversative-concessive meaning (meaning 3) expands and generalizes. The normative dictionary (DIEC) differentiates three concessive meanings associated with the connective uses of tanmateix. Two of the meanings are defined paraphrastically: "Expressa que una cosa és lògic que ocorri, esp. malgrat allò que semblava oposar-s'hi" ('It indicates that it is logical for something to take place, esp. notwithstanding everything that seems to oppose it'); "Expressa que una cosa ha d'ésser perquè en resulti explicable una altra" ('It indicates that something must happen for some other thing to be explicable'). The third meaning is defined by a synonym, nogensmenys 'notwithstanding'. The meanings, however, are not clearly differentiated.
The grammatical description of the marker is uneven. Some grammarians consider tanmateix as a concessive connective, while others define it as purely adversative; some treat it as archaic or formal, whereas more modern approaches treat it as a general contrastive marker. Let us revise the different approaches.
The marker tanmateix is included in Fabra's 1918grammar. Badia (19621994, p. 314) acknowledges the difficulty in establishing a unique meaning.
Casi imposible resulta dar una versión univalente de partícula tan matizada como tanmateix: expresa que es natural que suceda una cosa, en especial a pesar de aquello que parecía oponerse; que algo ha de realizarse para que de ello pueda darse razón de otra cosa; que lo que se acaba de decir no es obstáculo para algo. (Badia 1962, II, p. 97, Fn 11) It is almost impossible to give a univalent version of a particle as nuanced as this one: it expresses that it is natural for something to happen, especially in spite of what seemed to oppose it; that something must occur to justify something else; that what has just been said is not an obstacle to something.
Badia considers tanmateix as a formal concessive marker. Lacreu (1990, p. 327) labels it as archaic, just as nogensmenys, and points out that it is used as a direct translation of Sp. sin embargo.
More modern grammars consider tanmateix as a general contrastive marker. GUL (2002, p. 153) describes it as an intermediate between adversativity and concessivity, along with amb tot and no obstant això. Cuenca (2002, pp. 3196-97) classifies it as a general opposition marker, one of the most neutral and prototypical of its group along with en canvi 'however' and the conjunction però 'but'. Along her lines, the recent IEC grammars (GIEC, GEIEC, and GBU) classify tanmateix as an opposition marker.
Regarding its use in the 20th century, the corpus includes 11,193 occurrences of tanmateix, equivalent to 3085 tokens per million words. It is used by 1091 'authors' (including many cases in newspapers, with no specific author indicated). Its evolution in the 20th century can be seen in Figure 6. Badia considers tanmateix as a formal concessive marker. Lacreu (1990, p. 327) labels it as archaic, just as nogensmenys, and points out that it is used as a direct translation of Sp. sin embargo.
More modern grammars consider tanmateix as a general contrastive marker. GUL (2002, p. 153) describes it as an intermediate between adversativity and concessivity, along with amb tot and no obstant això. Cuenca (2002, pp. 3196-97) classifies it as a general opposition marker, one of the most neutral and prototypical of its group along with en canvi 'however' and the conjunction però 'but'. Along her lines, the recent IEC grammars (GIEC, GEIEC, and GBU) classify tanmateix as an opposition marker.
Regarding its use in the 20th century, the corpus includes 11,193 occurrences of tanmateix, equivalent to 3085 tokens per million words. It is used by 1091 'authors' (including many cases in newspapers, with no specific author indicated). Its evolution in the 20th century can be seen in Figure 6. Tanmateix increases its use from 1919 on and exhibits sustained high rates since 1969. As Martínez points out, and modern Catalan grammars reflect: …Al llarg del segle xx, l'ensenyament, la lectura i els mitjans de comunicació deuen haver estat les vies a través de les quals el tanmateix [ADVERSATIVOCON-CESSIU] ha arribat a estendre's, fins i tot, en registres orals mitjanament formals. (Martínez 2019, p. 128) …Throughout the 20th century, teaching, printed material and the media must have been the ways through which tanmateix with an [ADVERSATIVE-CONCESSIVE] meaning has come to extend, even in medium formality oral registers.
Tanmateix increases its use from 1919 on and exhibits sustained high rates since 1969. As Martínez points out, and modern Catalan grammars reflect: . . . Al llarg del segle xx, l'ensenyament, la lectura i els mitjans de comunicació deuen haver estat les vies a través de les quals el tanmateix [ADVERSATIVOCONCESSIU] ha arribat a estendre's, fins i tot, en registres orals mitjanament formals. (Martínez 2019, p. 128) . . . Throughout the 20th century, teaching, printed material and the media must have been the ways through which tanmateix with an [ADVERSATIVE-CONCESSIVE] meaning has come to extend, even in medium formality oral registers. To summarize, the polysemous tanmateix has extended its use as a contrastive connective, while the adverbial uses have reduced in frequency. Its meaning has also generalized from concession to general contrast in formal registers, maybe as a way to find an equivalent to sin embargo in Spanish. Sin embargo was also a concessive connective that indicates general contrast nowadays and, as shown in Section 2.1, has no verbatim counterpart in Catalan.
All in all, as Martínez argues, tanmateix illustrates the promotion of a word to avoid a norm problem. The connective has been extended outside formal written contexts of communication. Its extension of use is parallel to a change in meaning derived from the selection of one of its older uses.
Tot i difondre's inicialment en l'àmbit d'ús formal, tanmateix constitueix un altre exemple d'una proposta, en aquest cas molt reeixida, de promoció d'un valor d'un element lingüístic en el moment de construcció de la norma fabriana. (Martínez 2019, p. 139) Although initially disseminated in formal use contexts, tanmateix is another example of a proposal, in this case very successful, to promote the value of a linguistic item at the time of construction of Fabrian norm.

Conclusions
The analysis presented in this paper has followed the recent diachronic history of four contrastive connectives that have experienced relevant changes during the 20th century, namely nogensmenys 'nonetheless', emperò 'but', no obstant (això) 'notwithstanding', and tanmateix 'however'. The driving force behind the changes has been mostly formal institutions (or the people representing them, mainly Pompeu Fabra, and grammarians, such as Badia and Sanchis, who followed Fabra).
The evolution in the use of the four contrastive markers is summarized in Figure 7. The graphic shows different but related diachronic histories, which are the result of the codification process and a bilingual context in which Catalan has experienced interference from Spanish. The selection of tanmateix with an [ADVERSATIVE-CONCESSIVE] meaning probably had to do with Pompeu Fabra's concern to substitute, among others, the Spanish expression sin embargo.
All in all, as Martínez argues, tanmateix illustrates the promotion of a word to avoid a norm problem. The connective has been extended outside formal written contexts of communication. Its extension of use is parallel to a change in meaning derived from the selection of one of its older uses.
Tot i difondre's inicialment en l'àmbit d'ús formal, tanmateix constitueix un altre exemple d'una proposta, en aquest cas molt reeixida, de promoció d'un valor d'un element lingüístic en el moment de construcció de la norma fabriana. (Martínez 2019, p. 139) Although initially disseminated in formal use contexts, tanmateix is another example of a proposal, in this case very successful, to promote the value of a linguistic item at the time of construction of Fabrian norm.

Conclusions
The analysis presented in this paper has followed the recent diachronic history of four contrastive connectives that have experienced relevant changes during the 20th century, namely nogensmenys 'nonetheless', emperò 'but', no obstant (això) 'notwithstanding', and tanmateix 'however'. The driving force behind the changes has been mostly formal institutions (or the people representing them, mainly Pompeu Fabra, and grammarians, such as Badia and Sanchis, who followed Fabra).
The evolution in the use of the four contrastive markers is summarized in Figure 7. The graphic shows different but related diachronic histories, which are the result of the codification process and a bilingual context in which Catalan has experienced interference from Spanish. Some connectives or variants have decreased in use. This is the case of emperò, which has almost disappeared from current use, and the variant no obstant, only recently fully accepted by the norm. Some connectives have increased in use and then decreased because of their formal and archaic-like character, as in the case of nogensmenys. As for no obstant això and tanmateix, the connectives have experienced an extension of use.  1904-1913 1914-1918 1919-1923 1924-1928 1929-1933 1934-1938 1939-1943 1944-1948 1949-1953 1954-1958 1959-1963 1964-1968 1969-1973 1974-1978 1979-1983 1984-1988 1989-1993 1994-1998 1999-2003 2004-2008 2009-2013 nogenmenys tanmateix no obstant això emperò Some connectives or variants have decreased in use. This is the case of emperò, which has almost disappeared from current use, and the variant no obstant, only recently fully accepted by the norm. Some connectives have increased in use and then decreased because of their formal and archaic-like character, as in the case of nogensmenys. As for no obstant això and tanmateix, the connectives have experienced an extension of use.
As we have seen, Fabra banned emperò and introduced nogensmenys as a neologism. The evolution of emperò can be compared with that of nogensmenys (Figure 8 The connective emperò exhibited relatively high figures at the beginning of the century, but it soon began to reduce in frequency because of Fabra's normative proposal of abandoning it as a redundant form. Nogensmenys (557 cases, 171.7 pmw) was introduced anew by Fabra and seemed to consolidate during the first third of the century. It gained popularity in the period between 1964-1983 (251 occurrences, 45% of cases of all the period), but it ended almost as restricted in use as emperò. Overall, emperò and nogensmenys followed similar paths, except that emperò had frequent use before the 20th century and nogensmenys is a neologism substituting another form which also disappeared, noresmenys. Different beginning, same end.
The marker based on the present participle obstant exhibits three competing variants: no obstant això, això no obstant, and no obstant. In previous stages of Catalan, the three variants alternated in different proportions through time, no obstant being the most frequent form. However, the codification of Catalan marked no obstant as unacceptable. The banning of no obstant, which happens to be the variant that coincides with Spanish no obstante, implied the promotion of no obstant això. Formal use reflects this explicit norm.
Tanmateix is a completely different case. There is no explicit norm affecting the evolution of the marker, although implicitly traditional grammarians and writers might have been looking for a counterpart to the Spanish connective sin embargo 'however'. Tanmateix increased its frequency by mid-century and consolidated high rates of use while experiencing a change in meaning and expansion in use.
Let us now turn to the use of these connectives by reference grammarians, considering the works included in the corpus. The frequencies are uneven and sometimes surprising when compared with the explicit norm that grammarians propose.
(a) The neologism nogensmenys occurs 3 times in Fabra's works included in the corpus, 35 times in Badia's works, and 8 times in Sanchis' grammar. Only Badia seems to make an extensive-very extensive, in fact-use of this connective.   1904-1913 1914-1918 1919-1923 1924-1928 1929-1933 1934-1938 1939-1943 1944-1948 1949-1953 1954-1958 1959-1963 1964-1968 1969-1973 1974-1978 1979-1983 1984-1988 1989-1993 1994-1998 1999-2003 2004-2008 2009-2013 emperò nogensmenys The connective emperò exhibited relatively high figures at the beginning of the century, but it soon began to reduce in frequency because of Fabra's normative proposal of abandoning it as a redundant form. Nogensmenys (557 cases, 171.7 pmw) was introduced anew by Fabra and seemed to consolidate during the first third of the century. It gained popularity in the period between 1964-1983 (251 occurrences, 45% of cases of all the period), but it ended almost as restricted in use as emperò. Overall, emperò and nogensmenys followed similar paths, except that emperò had frequent use before the 20th century and nogensmenys is a neologism substituting another form which also disappeared, noresmenys. Different beginning, same end.
The marker based on the present participle obstant exhibits three competing variants: no obstant això, això no obstant, and no obstant. In previous stages of Catalan, the three variants alternated in different proportions through time, no obstant being the most frequent form. However, the codification of Catalan marked no obstant as unacceptable. The banning of no obstant, which happens to be the variant that coincides with Spanish no obstante, implied the promotion of no obstant això. Formal use reflects this explicit norm.
Tanmateix is a completely different case. There is no explicit norm affecting the evolution of the marker, although implicitly traditional grammarians and writers might have been looking for a counterpart to the Spanish connective sin embargo 'however'. Tanmateix increased its frequency by mid-century and consolidated high rates of use while experiencing a change in meaning and expansion in use.
Let us now turn to the use of these connectives by reference grammarians, considering the works included in the corpus. The frequencies are uneven and sometimes surprising when compared with the explicit norm that grammarians propose.
(a) The neologism nogensmenys occurs 3 times in Fabra's works included in the corpus, 35 times in Badia's works, and 8 times in Sanchis' grammar. Only Badia seems to make an extensive-very extensive, in fact-use of this connective. (b) Even though emperò was considered unadvisable since 1918, Badia uses emperò in 34 cases in the CTILC, 30 of which are in 4 different books (published in 1976, 1979, and 1981), always as a parenthetical connective. Sanchis (1950/1993, p. 285) also adheres to Fabra's preference for però, but uses emperò 57 times in his grammar, thus contradicting the explicit norm therein. (c) The questioned variant no obstant is used by Fabra in grammars and also in letters he wrote. Badia and Sanchis only use no obstant això (32 and 2 cases, respectively), which is consistent with their explicit norm. (d) Tanmateix is not problematic. Its frequency is variable (Fabra: 7 examples,Badia: 34 examples, Sanchis: 51 examples) not linked to any explicit norm.
The contradictions between the (explicit) norm and the use by writers and even by the leading grammarians-implicit norm-put forward the instability of the formal use of a language that was being codified at that moment. The contradictions also illustrate how direct prescription, especially when coming from Pompeu Fabra, the 'father' of modern formal Catalan, translated into use and highlighted the importance of paradigmatic relationships, especially the influence of Spanish on Catalan in a diglossic context of subordination of the latter.
In all the cases analyzed in this paper, changes in the use of a connective are directly linked to institutional actions expressed as explicit norms. The internal driving force for change was the paradigm as considered on its own, but mainly in relation with Spanish (Section 2.1), and alleged 'redundacy' in the case of emperò (vs. però) and an explicit purpose to 'depurate' Catalan from the influence of Spanish. Brumme (2020, p. 497) makes this idea very clear: The pressure of Spanish on the different areas of use has caused a certain purist attitude among some of the codification agents and institutions, although language users have protested against this on various occasions throughout history.
This purist attitude is associated with the linguistic ideology of Pompeu Fabra and, along his lines, of most grammarians and many writers. As Martí i Castell (2020) explains, the relation between language codification and the Catalan nation is a key principle of Fabra's actions. Fabra's main aim was to reconstruct a national language following four basic tenets: (i) resorting to Old Catalan, as a purer expression of the language, assuming that Catalan has suffered from the pressure of Spanish since the 16th century, (ii) resorting to informal Catalan when it was genuine (i.e., not interfered by Spanish), (iii) taking into account geographical variation, and (iv) comparing the alternatives with other Romance languages (mostly to find solutions different from those in Spanish).
Fabra and later grammarians after him were searching for 'authenticity' as an iconic representation of the 'spirit' of a speech community that strives to survive in a complex context because it lacks a state (Argenter 2020, p. 119). The voice of 'authenticity', according to Woolard (2016), is the reverse of the voice of 'anonymity', a universal voice not linked to any particular place or speech community that needs no justification and does not fight against a dominant language.
De manera que Fabra construïa un artefacte que havia de vehicular l'anonimat en l'àmbit territorial i social en què s'havia de difondre, en competència amb l'anonimat de què ja gaudia l'espanyol a les terres de llengua catalana hispàniques. Aquest era un objectiu polític a assolir. Així ho demanava la creació de la consciència lingüística nacional dels catalans i la plena funcionalitat del català com a llengua pariona de l'espanyol, del francès o de l'italià. (Argenter 2020, p. 120) So Fabra was building an artifact intended to implement anonymity in the territorial and social space in which it was to be spread, competing with the anonymity that Spanish already enjoyed in the Catalan-speaking territories in Spain. This was a political goal to be achieved. And this was needed in order to create the national linguistic consciousness of Catalan people and the full functionality of Catalan as a language which is equal to Spanish, French or Italian.
In conclusion, the analysis of four contrastive connectives has illustrated the dynamics of norm and formal use in the specific context of 20th-21st century Catalan, a key period in its codification and standardization. Paradigmatic relations combined with the pressure of Spanish have proven crucial to interpret norm changes and how they have an influence on an unstable formal use based on a linguistic ideology of authenticity linked to a context of diglossia.
The case of a group of contrastive connectives in Catalan highlights several methodological principles to be considered when analyzing the evolution of some forms and constructions and the interaction between their use and the norms applied to them.
(a) Changes in use cannot always be accounted for by resorting (only) to internal factors.
External factors can trigger or have an influence on linguistic change. (b) The analysis of an item must take into consideration its paradigmatic relationships. (c) A complete analysis of the origin and evolution of prescriptive rules and their translation in use can only be based on comparing the information from different sources (at least, dictionaries, grammars, and corpus examples). The (explicit) norm and the current use should be confronted in order to uncover potential contradictions and inconsistencies. (d) The codification of formal use is a complex process in which ideology plays an important role. (e) In a language contact situation, (socio)linguistic analysis cannot be limited to one of the languages, especially when dealing with a minoritized language.
Funding: This research received no external funding.

Conflicts of Interest:
The author declares no conflict of interest.