Implementation and acceptance of national language policy: the case of Dutch (1750–1850)

The paper discusses implementation and acceptance as crucial elements of a historical-sociolinguistic reappraisal of Haugen’s well-known theory of standardization. The case study that we focus on is the Dutch language in the second half of the eighteenth and the first half of the nineteenth century. In this period, Dutch became an object of political control. Significant aspects of the nationalization of language were the establishment of an officialized orthography (1804) and grammar (1805), which were to be used in the national school system. Education was the societal domain in which the national government tried to secure the transmission of the national language norms. We study the implementation and acceptance of official language norms from two perspectives, viz. by focusing on teaching materials developed for the new national school system, and by analyzing a recently compiled corpus of original language data from this period. We argue that implementation and acceptance, though relatively understudied topics in standardization studies, can usefully be operationalized, and turned into empirical questions that historical-sociolinguistic analysis can answer.


Introduction
The success of standardization depends much less on selection and codification than on implementation, acceptance and functional elaboration. Prescriptive norms are only truly effective if they develop into community norms. Nonetheless, in analyses of historical standardization phenomena more attention is usually paid to selection and codification than to the societal dimensions embodied by implementation, 1 3 acceptance and functional elaboration. In this paper, we focus on implementation and acceptance as crucial elements of a historical-sociolinguistic reappraisal of Haugen's (1966Haugen's ( , 1972Haugen's ( , 1987 well-known theory of standardization. The case study that we focus on is the Dutch language in the second half of the eighteenth and the first half of the nineteenth century. In this period, Dutch was nationalized, that is it developed into a symbol of the newly founded Dutch nationstate, and as such it became an object of political control (Rutten 2016). Significant aspects of the nationalization of language were the establishment of an officialized orthography and grammar (Siegenbeek 1804;Weiland 1805), which were moreover to be used in the national school system. Education was the societal domain in which the national government tried to secure the transmission of the national language norms.
In this paper, we study the implementation and acceptance of official language norms from two perspectives, viz. by focusing on teaching materials developed for the new national school system ("Implementation in education" section), and by analyzing a recently compiled corpus of original language data from this period ("Acceptance in language use" section). We argue that implementation and acceptance, though relatively understudied topics, can usefully be operationalized, and turned into empirical questions that historical-sociolinguistic analysis can answer.
In the "Implementation and acceptance" section, we discuss the concepts of implementation and acceptance, and how we operationalize them in this paper. The "Standard language ideology in The Netherlands" section provides historicalsociolinguistic background to the case study by introducing the rise of standard language ideology in the Netherlands in the period around 1800. "The variable and its historical-sociolinguistic context" section introduces the specific linguistic phenomenon that we focus on, viz. the representation of long vowels in open syllables. "Implementation in education" section discusses the implementation of new spelling principles in teaching materials for schools. "Acceptance in language use" section shifts attention to patterns of variation and change in language use in an effort to determine the acceptance of these new spelling principles. The paper end with the "Discussion" section. Haugen's (1966: 933) classic analysis of the four key concepts of standardization, that is of 'the step from "dialect" to "language", from vernacular to standard' comprises selection of norm, codification of form, elaboration of function and acceptance by the community. The former two concepts relate mainly to the form of language, while the latter two refer to the functions of language. Haugen (1966: 933) conceptualizes selection and acceptance as two 'aspects of language development' that concern society, while codification and elaboration primarily relate to the language itself, even though Kloss (1967: 30), whose Ausbau language is the model for Haugen's elaboration, states that '[t]he concept of ausbau language is primarily a sociological one'. For Haugen, elaboration mainly meant lexical adequacy to fulfill all necessary functions, while Kloss thought of ausbau in terms of genres and domains. Acceptance by the community concerns the function of the standard in society. Haugen (1966: 933) comments that 'a standard language, if it is not to be dismissed as dead, must have a body of users'.

Implementation and acceptance
However, the first three terms indicate human activity: a group of speakers select and codify a norm in the interest of 'minimal variation in form' (Haugen 1966: 931), while seeking to adjust the language to as many functional domains and contexts as possible in the interest of 'maximal variation in function' (Haugen 1966: 931). Acceptance, on the other hand, appears to be relatively passive, leaving a conceptual and empirical gap between the activities employed by a particular social group, and the possible end result of the wider community using the forms selected, codified and elaborated by this group. Haugen's (1966) well-known matrix seems to fail to account for the spread of the selected, codified and elaborated form of language through society. What is more, Haugen's work is framed in the activity-oriented language planning literature. Thus, Haugen (1972) replaces acceptance with propagation in order to have an active rather than a passive 'procedure', in line with the other three terms: the norm should also be propagated in order to ensure its dissemination across the language community. Haugen (1987: 59-64) reevaluates the model and introduces implementation as the fourth term. Implementation is defined as 'the activity of a writer, an institution, or a government in adopting and attempting to spread the language form that has been selected and codified', to which it is added that 'the spread of schooling to entire populations in modern times has made the implementation of norms a major educational issue' (Haugen 1987: 61). Haugen thus conceptualizes standardization as an active language planning activity, which he connects with the conscious spread of the standardized form from one social group to another. This social spread is usually top-down, that is, from privileged groups to less powerful groups. Acceptance is essential, even it is only by 'a small but influential group' (Haugen 1966: 933). Furthermore, Haugen (1966: 933) is quite explicit about the wider socio-political context in which standardization usually takes place: 'in our industrialized and democratic age there are obvious reasons for the rapid spread of standard languages and for their importance in the school system of every nation'. In sum, standardization appears to be tied to the political unit of the nation-state, in which influential groups try to disseminate the norm from above to other groups 'below'. In this act of implementation, education is vital.
This view of standardization corresponds to standard language ideology (SLI) as developed in the historical period of nationalism, from the eighteenth century onwards well into the twentieth century (Milroy 2012;Lippi-Green 2012: 55-62). In many European language areas, this period is characterized by governmental efforts to spread the standard variety among speakers, particularly through the school system, which 'was the institution where the ideology of one people, one territory and one language could be translated into reality' (Wright 2012: 71). It is not relevant here whether Haugen subscribed to this view of standardization and SLI himself, or whether he merely aimed to describe the prevalent ideology, although it is clear that his analysis predates the emergence of language ideology as a main topic of sociolinguistic analysis. Importantly, this view of standardization is identical to the view embraced by historical actors in standardization processes, such as proponents of linguistic nationalism in the Netherlands in the late eighteenth and early nineteenth century (Rutten 2016: 40-51, and see below, "Standard language ideology in The Netherlands").
In historical descriptions of the standardization of Dutch, selection and codification play a major role (e.g. van den Toorn et al. 1997;van der Sijs 2004;van der Wal and van Bree 2008). For the volume edited by Deumert and Vandenbussche (2003a), authors were asked to discuss the history of a particular Germanic language on the basis of Haugen's four key concepts. It seems significant that in many chapters, the formal side of selection and codification is extensively discussed, whereas less attention is usually devoted to the two key concepts relating to the functional dimension, viz. implementation/acceptance and elaboration of function. Moreover, discussions of implementation and acceptance often focus on metalinguistic discourse in which the spread of the standard is argued for, or they are restricted to the general observation that the standard is currently widely accepted. Implementation and acceptance are less often operationalized as empirical questions that can be answered on the basis of archival sources.
In their introduction to the volume, Deumert and Vandenbussche (2003b: 7) comment that '[t]he sociopolitical realization of the decisions made at the stages of selection and codification is referred to as implementation, that is, the gradual diffusion and acceptance of the newly created norm across speakers as well as across functions. The implementation stage is the "Achilles heel" of the standardization process: acceptance by the speech community ultimately decides on the success or failure of a given set of linguistic decisions made at the stages of selection and codification.' Implementation and acceptance can be viewed 'as the result of rational decision making', that is as a top-down process, or as the bottom-up result of 'social influence exercised in social networks'. They add that 'the novel forms of elementary national education which emerged from the late eighteenth and early nineteenth century in most European countries, and which provided prescriptive language education to large numbers of speakers, were a central force in the diffusion of standard languages and the formation of a standard/dialect diglossia' (Deumert and Vandenbussche 2003b: 7).
We contend that implementation and acceptance are not only the 'Achilles heel' for historical actors involved in standardization. They are also the 'Achilles heel' of much research on standardization in the past. In this paper, we argue that implementation and acceptance can usefully be operationalized, and thus turned into empirical questions that historical-sociolinguistic analysis can answer. We claim that there are solid historical reasons to consider implementation as an example of top-down 'rational decision making', often focused on the educational domain as the main locus of norm transmission. Furthermore, we consider acceptance to be an empirical matter of language use 'from below': variational patterns that move in the direction of the prescribed forms may indicate increased acceptance of these forms. Thus, we split the overarching concept of implementation into implementation per se, that is official language planning activities geared towards the spread of the standard forms along the lines of Haugen (1987) on the one hand, and acceptance as the actual spread of standardized language among 'a body of users' on the other hand (cf. Haugen 1966: 933).

Implementation and acceptance of national language policy
In Late Modern European nationalism, the implementation of standardized varieties was largely an educational issue. Acknowledging the importance of schools for norm transmission, Vandenbussche (2007: 29) adds that here, 'we also meet the "black box" of historical pedagogy […] We know little, almost nothing, about the methods and practices used in language-teaching in earlier times'. In the eighteenth and nineteenth centuries, however, education came increasingly under national control, and school inspection systems were set up to enable governments to exert influence on teaching practices. This opens up many ways to investigate the top-down promotion of the standard in educational settings. Recent research has focused on school inspection reports as valuable testimonies to teaching practices that otherwise remain hidden for historical analysis (Langer 2011). Schoemaker and Rutten (2017) show that Dutch school inspectors at the beginning of the nineteenth century actively participated in the dissemination of the newly devised standard (see also below, "Standard language ideology in The Netherlands").
At least for the case of Dutch, many other avenues exist to investigate the implementation of the standardized variety. From the late eighteenth century onwards, teachers' societies and teacher training colleges were founded. Teachers' societies organized meetings at which educational issues were discussed, including language issues. For the province of Groningen in the northeast of the Netherlands, a remarkable questionnaire was sent to local schoolteachers in 1828, comprising questions about all kinds of local practices, including language practices (Boekholt and van der Kooi 1996). Equally remarkable is the Icelandic collection of corrected student essays, testifying to the corrective attitudes of nineteenth-century grammar school teachers in Iceland (van der Feest Viðarsson 2017).
In the present paper, we will approach the issue of implementation through the lens of teaching materials ("Implementation in education" section). Reading education in the Early and Late Modern Netherlands, from the sixteenth to the nineteenth century, heavily relied on easy language guides, so-called ABC-booklets that mainly explained the alphabet, on religious prose texts and historical texts such as the Historie van David 'History of David' and the Spiegel der Jeugd 'Mirror of the Youth', and on newspapers and letters for the more advanced pupils (de Booy 1977(de Booy : 48-52, 1980. Grammar was not usually part of reading education, which changed after the school reforms of the early nineteenth century (see "Standard language ideology in The Netherlands"). By focusing on teaching materials that were used in reading education, we get a close look at the norms that were promoted.
Acceptance, to the extent that it is an attitudinal phenomenon, can obviously not be investigated with current methods in attitude research such as questionnaires. Still, contemporary metalanguage can reveal beliefs and ideologies prevalent in the historical language community. In this paper, we take a different approach, maintaining that patterns of variation and change may reveal the influence of norms and prescriptions on language users. Historical analyses of the influence of normative discourse on the wider language community have only become possible after the compilation of historical-sociolinguistic corpora with representative samples of language data from various groups of society. In recent years, the intricate relationships between prescription and historical patterns of language use have attracted quite some attention Poplack et al. 2015;Anderwald 2016). Here, we report on a case study based on a corpus specifically designed to measure the effects of prescriptivism in the history of Dutch ("Acceptance in language use" section).

Standard language ideology in The Netherlands
The Dutch case offers a unique opportunity to empirically assess the effectiveness of standard language policy in a historical context. After approximately 200 years of metalinguistic discourse, the middle of the eighteenth century marked a clear change of focus (Rutten 2016). Previous metalinguistic texts, from the sixteenth century onwards, targeted specific groups of society, for example poets, lawyers and ministers, and tended to conceptualize the proposed normalized variety of Dutch as an additional layer to the existing sociolinguistic spectrum. From c. 1750 onwards, the normalized variety promoted in metalinguistic discourse was increasingly perceived as the only 'real' Dutch language. Simultaneously, non-standard varieties such as regional dialects were increasingly considered to be flawed versions of the one truly Dutch language, and some commentators even argued for the conscious eradication of non-standard language. In language ideological terms, we may say that the supralocal variety used in certain 'high' registers of the written code and codified in eighteenth-century normative grammar underwent iconization (Irvine and Gal (2000: 37), signaling the rise of standard language ideology (Rutten 2016).
In other words, the second half of the eighteenth century witnessed the rise of linguistic nationalism, which was part of a broader development, viz. the rise of cultural nationalism in general. For the Dutch situation, it is usually assumed that cultural nationalism preceded nationalism as a political ideology. From c. 1750 onwards, a cultural process of nation building developed, which was accompanied by the formation of a national state in the 1790s. Thus, the foundation of the modern Dutch nation-state can be dated to c. 1800 (van Sas 2004: 44).
Importantly, the nation-state formation around 1800 also led to fundamental changes in the educational system as well as to a whole new language planning phenomenon, viz. an official language policy. The newly appointed minister of education-also a novelty-sought to nationalize the educational system by bringing it under national control through the establishment of a school inspection system in 1801 (Dodde 1968). The third and final school act of this period, issued in 1806, explicitly mentioned knowledge of the language as one of the core elements of the national curriculum, along with reading, writing and arithmetic (Boekholt and de Booy 1987: 99). In addition, a list of prescribed schoolbooks for these subjects was called for. The new educational policy also led to the establishment of a journal specifically oriented to school teachers, school inspectors and others working in the educational domain: the Bijdragen betrekkelijk den staat en de verbetering van het schoolwezen 'Contributions concerning the state and the improvement of the field of education', published from 1801 onwards.
Another important task for the new minister of education was to 'take all possible measures to purify and cultivate the Dutch language, and to regulate its spelling', as it was said in the Instructie 'Instruction' (1799: 6; our translation) handed to him on his appointment. This task led to the so-called schrijftaalregeling 'written language regulation', that is the publication of a Dutch orthography (Siegenbeek 1804) and a Dutch grammar (Weiland 1805), ordered by and printed on behalf of the national government, and meant for use in the administrative and educational domains.

The variable and its historical-sociolinguistic context
The issue that we focus on in this paper is the spelling of the long vowels a, e, o and u in open syllables. Historically, lenghtening of the short vowels <a>, <e>, <i>, <o> and <u> in closed syllables was represented by adding <e> or <i> or by doubling. The most common representation of long /a:/ in closed syllables, for example, used to be <ae>, while <ai> was an infrequent alternative; <aa> developed into the majority variant in the course of the seventeenth and eighteenth centuries. Long /i:/ was commonly written <ie> throughout the history of Dutch and until the present day, and both in closed and in open syllables. In closed syllables, doubling became the most common strategy in the Early Modern period in the case of a, e, o and u, i.e. <aa>, <ee>, <oo> and <uu>.
In open syllables, however, these digraphs were sometimes avoided, and a single grapheme was used, such as <a>. The issue of double or single graphemes in open syllables was an important topic in metalinguistic commentary in the eighteenth century, and therefore also one of the main issues addressed by Siegenbeek (1804) in his officialized orthography. Siegenbeek (1804: 97-136) opted for single graphemes in open syllables, and this has in fact been the main rule since then. Thus, the singular forms taal 'language', speel 'play', zoon 'son' and vuur 'fire' are today written talen 'languages', spelen 'play', zonen 'sons' and vuren 'fires' in the plural. The rule does not apply to long /i:/, so that the plural of bier 'beer' is bieren 'beers'.
For long a and long u, the issue was purely orthographical. In the case of long e and o, the issue was significantly more complicated. In the northwest of the language area, particularly in large parts of Holland, etymologically different long e's and o's have merged into /e:/ and /o:/. This merger is also part of the standard variety. Many dialects, however, maintain the phonological difference between so-called softlong and sharplong vowels, as they are traditionally called in Dutch historical linguistics. Softlong ē and ō are long vowels that developed through lengthening of the originally short vowels /ε/ and /I/, and /o/ and /u/ in open syllables, for example the vowels in the first syllables of spelen 'play' and open 'open'. Sharplong ê and ô are long vowels that developed through monophthongization of the West Germanic diphthongs *ai and *au, for example in steen 'stone' and boom 'tree'.
The coexistence of merging and non-merging varieties led to various writing traditions (Rutten and van der Wal 2014: 35-36). In phonology-based systems, the etymological difference between softlong and sharplong vowels was represented in the orthography. This was traditionally done by doubling the grapheme in the case of sharplong vowels in open syllables, for example steenen 'stones' and boomen 'trees' as opposed to softlong spelen 'play' and open 'open'.
Writers for whom the historical difference had become opaque developed alternative spelling systems. In a morphologically-oriented system, open syllables were spelled <ee> on analogy with closed syllables, and <e> in the absence of possible analogical forms, for example steenen because of the singular steen, and also speelen because of speel, but hemel 'heaven' with <e>. Finally, syllabic systems came into use which had either <e> or <ee> in open syllables, independent of phonological or morphological considerations, for example steenen, speelen and heemel, or stenen, speelen and hemel. Table 1 compares these writing practices. Syllabic I is the present-day standard Dutch system. Siegenbeek (1804) prescribed the phonological system.
In language use, these systems also occurred, though many writers adopted variable writing practices, particularly in regions where the vowels had merged, or were in the process of merging. Rutten and van der Wal (2014: 34-44) show that even in regions where the etymological difference is maintained until the present day, such as the Zeeland dialect area, a remarkable shift in writing practices took place between the seventeenth and the eighteenth centuries. Sharplong and softlong e's were neatly separated in seventeenth-century private letters from that area, along the lines of the phonology-based system. In the eighteenth-century data, however, both types of long e patterned similarly attracting mostly <ee>. Thus, the dominant approach became syllabic system II. This means that the merger, alien to the area of Zeeland, became represented in the spelling. Rutten and van der Wal (2014: 67-72) show that this development mirrored the changes taking place in Holland, in particular in Amsterdam, where however the merger was also part of the spoken language. In sum, the phonology-based system gave way to a syllabic system, and this happened both in Amsterdam and Zeeland, despite the fact that Zeeland maintained the etymological difference in the spoken language.
In the eighteenth-century normative tradition, all three major types were promoted (phonological, morphological, syllabic). Grammarians such as Moonen (1706) and Sewel (1712) and many of their followers in the subsequent decades opted for the morphology-based system (Rutten 2011: 96-97, 116-118). Already in 1723, ten Kate gave a detailed explanation of the etymological difference, which was the foundation of the phonological system. This position was also taken by Kluit (1763), and through him it became part of Siegenbeek's (1804) prescriptions (Rutten 2011: 98-103, 116-118). This was not the only etymologically driven spelling choice by Siegenbeek, but it is still remarkable that he maintained the phonology-based system, based on an etymological difference between vowels that had merged in his native Holland region centuries before, viz. around 1600. Moreover, Siegenbeek's choice did not only imply a move away from the spoken language of his time, it also went against the changes in written usage as documented by Rutten and van der Wal (2014: 67-72), who argue that syllabic systems were on the rise in the eighteenth century, in the slipstream of the phonological merger. In sum, Siegenbeek was following a particular tradition in normative grammar, whereas other normative grammars as well as patterns in language use preferred entirely different writing systems.

Implementation in education
As explained above ("Implementation and acceptance" section), we will approach the complicated issue of implementation through an analysis of teaching materials.
In the late eighteenth century, two types of sources were fundamental in reading education: language guides and reading materials, though note that language guides often also comprised texts for reading exercises. Reading materials usually comprised short texts with a strong moral purpose. The language guides were often quite different from the more complex publications that made up the normative grammatical tradition. Instead, they focused on the alphabet and on spelling principles, integrating grammatical knowledge only to a limited extent. In the nineteenth century, after the study of the national language had become a mandatory part of the school curriculum, the market for schoolbooks grew tremendously, including also language guides with more extensive grammatical information. Based on the historical literature, contemporary book reviews, remarks found in inspection reports and the number of reprints and new editions, it is possible to compile a reliable list of the most widely used teaching materials for reading education (de Booy 1977;Huiskamp 2000). We focus on the second half of the eighteenth century, from 1750 until the publication of Siegenbeek's official orthography in 1804, and on the first half of the nineteenth century, from Siegenbeek until 1850. For the first period, we identified twenty widely used teaching materials. This list comprises both language guides and reading materials. 1 The selection includes reading materials with a long tradition, such as Trap der Jeugd, Spreuken van Salomon and Historie van David, and reading materials that were first published in this period, such as van Alphen's Proeve van Gedichten voor Kinderen, van Oosterwijk Hulshoff's Geschiedenis van Jozef, and Buis's Natuurkundig Schoolboek. In addition, grammatical works were taken into account that were perhaps not used in the class room, but that were still oriented to schoolteachers, such as van Belle's Korte Schets der Nederduitsche Spraakkonst, van der Palm's Nederduitsche Spraekkunst voor de Jeugdt, and Stijl/van Bolhuis's Beknopte Aanleiding tot de Kennis der Nederduitsche Taal.
In these twenty books, we analyzed the representation of the long vowels a, e, o and u in open syllables. Some language guides formulated a specific rule. If explicit rules were absent, as in the reading materials, we took the first fifty words with a long a, e, o or u in open syllables. These tokens were then first scrutinized for syllabic systems, preferring either one grapheme (v) or a digraph (vv; v = vowel). If there was variation between single graphemes and digraphs, we investigated whether morphological spelling principles (vv (m)), or in the case of e's and o's, phonologybased considerations conditioned the variation (vv (ph)). Table 2 shows that syllabic spelling systems were rarely used. Van Bolhuis (1793) and Nyland (1800) opted for syllabic system I ( Table 1). The Spiegel der Jeugd (1752) has both single and double graphemes in all positions, though not depending on phonological or morphological principles. Other publications with a variable result are van Bolhuis/Stijl (1776) and the Spreuken van Salomon (1784). Van Bolhuis was the editor of Stijl's text. It turns out that Stijl preferred the morphological system, whereas van Bolhuis preferred the phonological system. The Spreuken van Salomon used a syllabic system with one grapheme in open syllables, except in a few instances where words with a so-called sharplong ê are spelled <ee>.
The general trend is that phonological and morphological systems occurred in the 1750s, 1760s and 1770s, after which the morphological system took over. This system was easier, of course, as it applied to all four long vowels in open syllable, whereas the phonological system only applied to long e's and o's. We argued above ("The variable and its historical-sociolinguistic context" section) that Siegenbeek's preference for the phonological system ran counter to the merger that characterized many Hollandic dialects, and that also made it into standard Dutch, as well as to changes in language use, where syllabic systems were on the rise. Here, it turns out that Siegenbeek's choice also ran counter to the evolution in writing practices in eighteenth-century schoolbooks. For the second period from 1805 to 1850, we selected a larger number of schoolbooks and language guides as there were many more publications in use due to the increase in schoolbook production. Thus, we were able to distinguish two types of publication: grammar books meant for use in schools, and reading materials. Twenty grammar books were scrutinized, 2 and eighteen books meant for reading practice. 3 As with the first period, we focused on books that were widely used, and often saw several reprints.
Whereas the eighteenth-century results displayed quite some variation, with various norms and writing systems being in competition, the nineteenth-century results are completely homogeneous. All 38 publications follow Siegenbeek (1804) in the case of the spelling of long vowels in open syllable, that is, they adopt a single grapheme in open syllables (talen, spelen, zonen, vuren), except for the representation of the sharplong e's and o's that derive from West Germanic diphthongs (steenen, boomen).
This means that the implementation of the Siegenbeek spelling in language teaching materials in the first half of the nineteenth century was extremely successful, right from the start, despite the fact that Siegenbeek's prescriptions constituted a breach with prevalent writing practices both in schoolbooks and in private letters.
Within one year after the publication of Siegenbeek's official spelling in 1804, books oriented to the field of education adopted the newly prescribed orthographical conventions. Some books made this explicit even on the title page or in the preface. As early as 1805, du Mortier entitled his 'grammar for the youth': Letterkunst voor de jeugd, of handleiding om de kinderen in de scholen te oefenen in de Nederduitsche Spelling van den heer Matthys Siegenbeek, that is 'Grammar for the youth, or guide to teach the children the Dutch spelling of Matthijs Siegenbeek'. Similarly, Wertz (1805: i-ii) indicated that his book was in accordance with the officialized spelling of Siegenbeek (1804). Someone like Wester, a prolific language commentator himself in the late eighteenth century, who had prescribed the morphological spelling of long vowels (see Table 1), switched to the phonology-based system adopted by Siegenbeek (1804). Other authors, too, adopted the Siegenbeek spelling after 1804. W. van Oosterwijk Hulshoff, for example, published his Geschiedenis van Jozef in 1797, while a fifth edition came out in 1806. Whereas the earlier publication was characterized by the morphological system, the 1806 edition followed Siegenbeek (1804). A similar change can be witnessed in the Natuurkundig schoolboek by J. Buis. The implication of this successful implementation after 1804 is that the written language that children came into contact with was uniformly influenced by Siegenbeek (1804).

Acceptance in language use
Another aspect of implementation relates to language use. Did language users implement the new orthographical rules in their writings? For this aspect of implementation, Haugen (1966) suggested the term acceptance. In this section, we report on a corpus study of long vowels in open syllables. In view of the high frequency of long vowels in open syllable, and the complexity of the variable, particularly also in actual usage data, which tend to be even more variable than edited and published works, we decided to restrict the analysis to one long vowel in particular, viz. long e. Another reason was the considerable work already done on long e, particularly from a historical-sociolinguistic perspective, on which we can build in the present analysis (Rutten and van der Wal 2014). In "The Going Dutch Corpus" section, we introduce the corpus, and in "Results" section we present the results.

The Going Dutch Corpus
Our analysis is based on the Going Dutch Corpus (GDC), specifically built to determine the influence of Siegenbeek (1804) and Weiland (1805) on patterns of language use. The GDC has a built-in diachronic dimension with original language data from approximately one generation before the official language policy of 1804/1805 as well as data from c. one generation later. We allowed for twenty-year windows to ensure that enough archival sources were available so that period 1 spans the years 1770-1790, and period 2 the years 1820-1840.
Taking a historical-sociolinguistic perspective, we selected two types of ego-document, viz. private letters as well as diaries and travelogues. It has been shown that these handwritten texts can give access to relatively informal, speech-like language use of the past (e.g. Elspaß 2012; Rutten and van der Wal 2014). In addition, the GDC comprises newspapers. By including newspapers, we are able to compare manuscript sources to print sources, which is particularly relevant in a situation of top-down language planning. A crucial issue is whether editors and printers were inclined to adopt the official norms, and whether this enlarged differences between manuscript and print language. Moreover, newspapers were still locally produced and distributed, making them interesting print sources from a sociolinguistic perspective. Thus, the GDC is a diachronic multi-genre corpus, developed on the assumption that diachronic changes will affect different genres in different ways. All texts were manually transcribed from pictures of the original archival sources.
The GDC comprises seven regions within the northern Low Countries; see Figure 1. Previous historical-sociolinguistic corpora mainly comprised sources from North and South Holland and from Zeeland (Rutten and van der Wal 2014). The GDC thus expands the regional coverage with four regions to the east of Holland and Zeeland: Groningen, Friesland, Utrecht and North Brabant. Finally, the two types of ego-document are distributed as equally as possible over the categories of male and female writers.
In sum, the GDC comprises two periods, three genres, seven regions, and in the case of the ego-documents also two genders. When building the corpus, we assumed that in terms of orthographical and grammatical uniformity and standardization, the newspapers would outperform the diaries and travelogues, which would on their turn surpass the private letters (cf. Schneider 2013). This led to different requirements for the various subgenres with respect to the minimal number of words. In the case of newspapers, we aimed at 5000 words per region and period. For diaries and travelogues, we raised the desired number of words to 10,000, for private letters to 15,000.
The entire GDC has 421,878 words distributed across the external variables as shown in Table 3. Period 1 includes 200 private letters written by 154 individuals, and 26 diaries and travelogues written by 25 individuals. Period 2 has 200 letters by 144 individuals, and 25 diaries and travelogues by 25 individuals.
As can be seen in Table 3, male writers are overrepresented. However, with 30-40% of the sources being written by female authors the GDC considerably differs from the traditional near-absence of female writers in language histories.
The corpus search generated 6086 tokens of softlong ē, and 3487 tokens of sharplong ê. Table 4 gives the diachronic overview of the orthographical representation of the long e's in the GDC.
The results for the sharplong variant confirm the earlier results obtained by Rutten and van der Wal (2014): the digraph <ee> is preferred. This is already the case in period 1 and continues in period 2. A major change, however, occurs in the case of the softlong variant ē. Whereas the eighteenth-century results show a modest preference of 59.5% for <e>, which is the preferred variant in the phonology-based system, the share of <e> increases to 92.5% in period 2. The result is Implementation and acceptance of national language policy a neat distribution of <e> and <ee> in period 2, in conformity with the system prescribed by Siegenbeek (1804).
Since the sharplong ê was already spelled <ee> in more than 90% of the instances in the pre-Siegenbeek period, a pattern which remained stable in the post-Siegenbeek period, we will restrict our analyses of genre, region and gender to the vowel that saw a major change, i.e. the representation of the softlong ē. The first external variable that we focus on is genre. In Table 5, we present the proportion of the 6086 tokens with softlong ē spelled <e>, i.e. the variant officialized by Siegenbeek (1804).   Table 5 shows that all three subgenres in the GDC show a marked change in the direction of the prescribed variant. In newspapers from period 2, only 1 token with <ee> was found. Diaries and private letters also have extremely high proportions of <e>, especially against the background of the much more variable picture for period 1. Private letters from period 1 display an even distribution of <e> and <ee>, possibly indicating the merger of softlong and sharplong e's, and the concomitant randomness of the distribution of related graphemes. Newspapers, however, already display a preference for <e> in period 1. Interestingly, this is also the case for diaries and travelogues, which confirms our idea that these are markedly different from private letters, also linguistically, with diaries and travelogues veering towards published materials.
Turning to region, Table 6 shows the proportion of the 6086 tokens for softlong ē spelled with the officialized variant <e>.
In four regions, the pre-Siegenbeek distribution is at chance level (Friesland, North Brabant, North Holland and South Holland). In two regions that maintain the difference in many dialects, <e> is already preferred in period 1 (Zeeland, Groningen). This also applies to Utrecht, for which we do not have an explanation at this point. An equal distribution at chance level clearly suggests a merger in the spoken language, whereas the preference for <e> in Zeeland and Groningen confirms the maintenance of the original difference in the spoken language. Our main research question, however, concentrates on changes between period 1 and 2, and here all regions show a remarkable increase of the prescribed variant <e> to more than 85%. This means that independent of the absence or presence of the phonological merger, the etymological difference between softlong and sharplong e is represented throughout the language area in the nineteenth century, as advocated by Siegenbeek (1804). In Table 7, we cross-tabulate the results for genre with gender, incorporating the diachronic dimension as well. As before, we focus on the proportion of <e>-spellings for softlong ē. Table 7 shows that the generic difference between private letters and diaries/ travelogues holds across gender. Both men and women distribute <e> and <ee> at chance level in their private letters from period 1. There is a significant difference with diaries and travelogues. In period 2, both men and women shift to <e> as the dominant variant, in accordance with Siegenbeek (1804), though the change appears to be even more pronounced among men than among women. The gender difference is in line with the large-scale sociolinguistic analyses of seventeenth and eighteenthcentury Dutch by Rutten and van der Wal (2014: 395-396), who argue that gender differences in the degree of participation in the written culture may be important.

Discussion
In this paper, we have tackled the issues of implementation and acceptance through an analysis of normative and linguistic practices in the eighteenth-and nineteenthcentury Netherlands. In 1804 and 1805, an official spelling and grammar of Dutch were published as part of a wider policy aimed at the nationalization of language and education. The new Dutch spelling and grammar (Siegenbeek 1804;Weiland 1805) were meant for use in the educational and administrative domains. The first official Dutch language policy and language-in-education policy are landmarks in the iconic intertwining of language and nation in the early days of European nationalism.
In his orthography, Siegenbeek (1804) opted for single graphemes for the representation of long vowels in open syllables, such as <a> and <u>. There were two exceptions to this general rule. Long e and long o derived from West Germanic diphthongs through monophthongization should be spelled with digraphs. This led to pairs such as spelen 'play' versus steenen 'stones', and open 'open' versus boomen 'trees', in which the vowels in the first syllables were homophonous for Siegenbeek's choice was remarkable in many respects. The dominant phonological system of Holland that constituted the input for the spoken standard had the merger of the different long e's and of the long o's. As a result of this merger, writing practices in private letters were increasingly favoring syllabic writing systems, with either a single grapheme or a digraph in any open syllable. Writing practices in schoolbooks were increasingly morphological. Private letters and schoolbooks thus opted for different strategies, but in both cases graphemization was taking place, that is 'the reduction of phonological considerations and the increase in choices directly linked to the written code' (Rutten and van der Wal 2014: 41). In the normative grammatical tradition, the morphological system was also important, which was however in competition with the phonology-based system adopted by Siegenbeek. Despite going against the grain of so many patterns of use, Siegenbeek's preference for the phonological system was suprisingly successful. In schoolbooks from the first half of the nineteenth century, the phonology-based system is used across the board. The new spelling was immediately and with no or hardly any exceptions implemented in teaching materials for reading and grammar education. Of course, this total shift should be seen against the background of the existing normative tradition: normalizing language was not a new idea as such by 1800. Still, officially promoted interference with orthographical and grammatical choices was a novelty. Likewise, the political level at which it was instigated, viz. the level of the nationstate, was new. While there was opposition from various angles to the efforts at nationalization of the educational domain (e.g. Boekholt and de Booy 1987: 101), the nationalization of the language seems to have been much less contested, strongly suggesting that authors of schoolbooks generally subscribed to the standard language ideology and the concurrent desire to implement a homogeneous language through schooling, as did teachers and school inspectors.
The new journal for people working in the educational domain, the Bijdragen betrekkelijk den staat en de verbetering van het schoolwezen 'Contributions concerning the state and the improvement of the field of education' was supported by the government: the first full article that it published was a mission statement by the minister of education (Bijdragen 1801: 25-48). Already in 1805, short reports of the goings-on in particular schools noted that Siegenbeek's orthography was given as a prize to diligent pupils (Bijdragen 1805: 71). Furthermore, a large part of the journal was devoted to book reviews, right from the first issue onwards. In these reviews, the degree to which new publications such as schoolbooks conformed to the norms of Siegenbeek and Weiland was a recurrent topic (see e.g. Bijdragen 1806: 11). Thus, bad reviews were a potential threat to the financial success of schoolbook publishers.
To a large extent, those who were involved in the implementation of the new language policy had similar backgrounds and were active in similar networks. The call for a reorganization of the entire educational domain, including the establishment of a language-in-education policy, originated from the many learned societies and amateur associations characteristic of the eighteenth century. People such as Siegenbeek and Weiland as well as the first minister of education, Johan van der Palm, were active in these societies and subsequently took up official and semi-official positions related to the field of education. Siegenbeek and Weiland wrote the officialized normative publications, van der Palm became minister of education, Siegenbeek was also a school inspector. Interestingly, the implementation of new educational regulations, which also covered the language-in-education policy, was to a large extent the responsibility of the individual school inspectors (see also Schoemaker and Rutten 2017).
The shift in language use towards a distribution of <e> and <ee> in line with Siegenbeek's prescription is perhaps even more remarkable. Newspapers and diaries/travelogues already tended towards the Siegenbeek distribution in period 1, but private letters did not. In period 2, after the establishment of the language policy, there appears to have been a general acceptance of Siegenbeek's prescription-in all seven regions, in three genres, by both men and women. Recall that the two etymologically distinct e's had merged in many regions, resulting in regional differences in spelling in period 1, which however vanished after 1804. Krogull (2018) shows a similar effect for another spelling variable. Krogull et al. (2017), however, argue that such a strong top-down effect is largely lacking in the morphosyntactic case of relativizers. This could indicate a significant difference between orthography and morphology, although we also know that morphological issues are among the most heavily debated present-day usage problems.
The shift in language use is all the more surprising given earlier tendencies towards graphemization. However, we do know from the present study that schoolbooks and newspapers, which were also used as reading materials in school, changed in the direction of Siegenbeek (1804). This implies that a lot of reading materials that children were exposed to in nineteenth-century schools had adopted the official spelling.
The discussion so far shows that we do not have definite answers yet to all questions raised by this particular case. However, our prime concern in this paper was to demonstrate the feasibility of an empirical approach to the concepts of implementation and acceptance. These two crucial aspects of a theory of standardization can be operationalized in a meaningful way, and turned into empirical questions that historical-sociolinguistic analysis can answer.