SAND Volume II – Commentary

Syntactic Atlas of Dutch Dialects – Commentary – Volume II

The Syntactic Atlas of the Dutch Dialects (sand) illustrates the syntactic variation in the Dutch language area at the beginning of the 21st century. The Dutch language area consists of The Netherlands, the northern half of Belgium and a small area in the northwest of France, on the border with Belgium. sand consists of two volumes. Volume I appeared in 2005, at the same time as volume I of the Morphological Atlas of the Dutch Dialects (mand) and the final volume of the Phonological Atlas of the Dutch dialects (fand). sand Volume II and mand Volume II appeared in 2008. Together, these three atlases provide a unique, detailed and extensive description of phonological, morphological and syntactic microvariation in Dutch.

The description and mapping of syntactic variation is important for a number of reasons. Local dialects are disappearing faster than ever before through the increase of mobility and communication and through the general decrease in the ability of a community to remain isolated. These dialects appear to be increasingly approximating regiolects as well as the standard language. Most experts think that a substantial part of this variation will be lost fairly soon. The wealth of variation that is described in these atlases is of important cultural and historical value; it represents an essential part of the cultural heritage of the Dutch-speaking area. The amount of syntactic variation that is described in this atlas is surprising, since most people – laymen and professionals – have the impression that variation is mainly concentrated in the domains of pronunciation and lexicon. However, these atlases show that the amount of grammatical variation in a relatively small language area is astonishing.
This atlas provides unique material for linguistic research. Linguists may use these maps in their research into the nature of the linguistic system. What exactly are the properties of microvariation? How is microvariation related to variation between languages? What are the boundaries of microvariation? How is microvariation related to the general linguistic system? What typological generalisations can be deduced from the atlas? Do typological generalisations between languages also hold for language-internal generalisations? Are there similarities between geographic variation and diachronic variation? Which aspects of geographical variation can be explained as the result of language change? Do the properties of geographic language variation provide us with insights regarding the direction of language internal change? The atlases also provide information concerning language-external properties, such as the history of Dutch, the ethnolinguistic situation, the genesis of the standard language, the influence of related Germanic languages such as German and English, the linguistic position of Frisian as a minority language, the properties of dialect attrition, etc. The collections of fand, mand and sand provide us with a monument for the Dutch language; for the first time in history, linguists may assess hypotheses on language variation with the aid of a complete and precise description of the dialects of a language.

Syntactic phenomena

The aim of the sand is to provide a detailed description of the present-day situation with respect to syntactic variation in Dutch dialects. The selection of topics investigated was inspired by a survey of existing literature. We have chosen for an approach in which a selection of topics is investigated in depth, rather than attempting to provide a descriptive overview of all phenomena that show variation in the syntactic domain. This approach allows us to evaluate generalisations and hypotheses that relate to a particular phenomenon in the literature. The fact that the selected topics are investigated in detail makes it possible to formulate new descriptive generalisations and theoretical hypotheses. Of course, we have found much more variation than shown in the sand-volumes. The complete data collection is available online (see section Dynasand below).

In volume II, we have concentrated on the following issues:

two and three-verb clusters,
the verbal cluster: interruption and morphosyntax
auxiliaries and verb positions
negation and quantification

In Volume I (2005), the focus was on complementisers and complementiser agreement, subject pronouns, subject doubling, subject clitizisation in yes/no-sentences, reflexive and reciprocal pronouns and preposed constituents.

Maps

sandII has two components: a book with maps and a book with commentary. The map book commences with two general maps. The first map provides the location of provinces and regions, in addition to the locations in which the sand interviews were held. This map can be folded out in order to position it alongside the specific maps in the atlas. Next to the fold-out map, there is an alphabetical list of place names, their coordinates and their usual dialectological code, the so-called Kloeke-number. The second general map provides a subjective classification of the dialects, based on a questionnaire that was sent to 1500 informants of the Meertens Instituut in 1939. These informants were asked to judge the degree of similarity between their dialect and the dialects in the neighbourhood (cf. Daan & Blok (1977)). This subjective map is included for comparison with the maps in the sand. The sand makes use of maps with symbols. The keys for each map contain a reference to the relevant section in the commentary.

Symbol maps

Symbol maps are used to reflect the geographic distribution of distinct variants of a particular syntactic variable, and to visualize potential correlations between variables. The various syntactic variants of a specific variable are depicted by means of coloured squares. Within a set of thematically coherent maps, the colour of a specific variant is kept constant as much as possible. Within a single symbol map, the use of related colours (e.g., light green vs dark green) indicates related phenomena. If a location allows more than one variant of a particular variable, this location has more than one coloured square (with a maximum of six squares, three horizontally and two vertically). The order of colours in a combination of squares is kept constant and the squares are located in a predictable manner with respect to each other. The keys of each map present the full set of coloured squares, which is given in the fixed order that is used on the map. A location with more than one variant can thus simply be distinguished from neighbouring measuring points. If none of the syntactic variants on the map were found in a particular location, the sampling point is printed as a grey point. A grey point may also indicate that there are no relevant data for the particular location. The symbol maps were drawn with MapInfo software.

Commentary

The detachable commentary comes in two versions: a Dutch and an English version. The four chapters in the commentary book correlate with the four chapters of the map book. Each chapter in the commentary book is introduced by a general introduction in which the nature and the linguistic relevance of the specific phenomenon is discussed. We have tried to avoid theory-specific terminology and theory-internal analyses in order to enhance the accessibility of the atlas and to make the atlas less dependent on the theoretical state of affairs in the early 21st century. The introduction also contains a paragraph in which the syntactic phenomena are discussed in their historical perspective. The second section of each chapter contains a short discussion of the most relevant literature for the topic under discussion. This overview of the literature is not intended to be exhaustive. In the third section, the different maps are discussed individually. For each map, we have provided the following information: the type of syntactic phenomenon, the test sentences used in the interviews, the way the data were elicited and the geographical distribution of conspicuous properties. In general, the discussion of individual maps is kept rather short since the maps speak for themselves in most instances. Often, there is a discussion about whether a particular map confirms or runs contrary to an existing generalisation. Each chapter concludes with a list of the relevant literature. This list comprises more literature than the discussion in the second section, but it is not exhaustive either. For more literature, we refer to the sand bibliographythat can be consulted at www.meertens.knaw.nl.

Dynamic Syntactic Atlas

Inspired by this printed atlas many readers will want to consult the underlying data, draw their own maps and combine various maps with each other. This is possible with Dynasand, an on-line database with search engine and cartographic software which is publicly available at www.meertens.knaw.nl. This database includes all the data that were collected in the sand-project: the sound recordings and partially tagged transcriptions of the oral interviews, the results of the written questionnaires and the results of the telephone interviews. The user-friendly search engine allows the user to look for information with respect to sentences, locations, location codes (Kloeke numbers), strings of words, parts of words, word categories, and lemmas. With the cartographic component, the user is able to display the results of the data (s)he has found on a map. Sets of data may be combined into one map, thereby allowing the user to investigate possible correlations between phenomena that have been entered into the database. The user is able to determine the properties of the map with respect to the area that is depicted, the symbols and the colours. An extensive description of the dynamic-sand is given in Barbiers, Cornips & Kunst (2007).

Sampling points

The dialects in the Netherlands (and Friesland), the Dutch-speaking half of Belgium and a small part of northern France constitute a continuum. It is impossible to determine exact boundaries between different dialects. Consequently, it is impossible to determine the number of dialects in the Dutch language area. In order to determine the number of locations necessary for a reliable picture of the occurring syntactic variation, various criteria have been taken into account.

The first criterion is that sampling points must be evenly distributed over the entire language area. The area was divided into equally sized squares. In each square, one reckoning with the relative position of squares.
Secondly, much variation was to be expected, following relevant literature and dialectological knowledge. This caused an increased density of transition zones. This is the reason that, for instance, the province of Oost-Vlaanderen has a somewhat higher density of Noord-Holland. This criterion is also responsible for the fact that there are more Netherlands than in the western part. Because of their relative isolation, a (former) Dutch islands.
Groningen, Heerlen, Nijmegen, Rotterdam and Utrecht. More so than in smaller locations, the informants in these cities are representative of their neighbourhood, and not of the whole city. Dialects may differ accordingly among different parts of these cities.
The fact that some locations do not have a long history or undergone large demographic changes was taken into account. This led usto discard the polders in the IJsselmeer.
Final considerations in determining the selection of were available to this project.
The application of these criteria led to a total of 267 158 of which are located in the Netherlands, 102 in Belgium and 7 in France (cf. the fold-out map).

Informants

The informants were selected using the following criteria:

The informant and his/her parents are born and raised in the place of residence.
The informant has lived in the place of residence until the age of seven years.
The informant uses the dialect at home and in at least one public domain.
The informant preferably belongs to the lower or the lower middle-class.
The informant is between 55 and 70 years old.
At first, the informants were selected via family, friends and acquaintances or societies for regional/local geography and history or regional/local dialect societies. Furthermore, existing files of informants such as the files of the Meertens Instituut, the phonological atlas (fand) and the three dialect dictionaries of southern Dutch: the dictionaries of Brabant dialects, Limburg dialects and Flemish dialects. Potential informants were screened to see whether they met the criteria mentioned above, to ensure that they did not have a normative perspective towards their dialect and to see whether they would qualify as an assistant interviewer (see Methodology). For each location, two informants were needed. In many cases, a selected informant in a particular location was able to provide the name of another suitable informant..

Methodology

When a speaker of a standard language interviews a dialect speaker, it often happens that the dialect speaker uses a language variety that is somewhere between the local dialect and the standard language. This phenomenon is known as accommodation. The sand interviews were organised in such a way to minimise the risk of accommodation. This has led to a methodology for the Netherlands that was somewhat different from the methodology that was used in Belgium and France. In the Netherlands, we have selected two informants at each location, one of which assumed the role of assistant interviewer, while the other was the target informant. Each interview required an entire day. In the morning session, the field worker instructed the assistant interviewer to record the questions for the questionnaire on tape (DAT-recorder or minidisc). The actual interview was recorded in the afternoon. The assistant interviewer instructed the informant, clarified individual questions if necessary and sometimes discussed the answers to the questions with the informant. During the interview the local dialect was used. The field worker stayed out of this interview as much as possible. In Belgium, the field worker did the target interview him/herself, using a variety of the regional language. Again, two (or more) speakers of the local dialect were present during the interview, but, in these cases, both had the role of informants. The difference in methodology in Belgium and the Netherlands was motivated by the fact that dialect speakers in Belgium are more stable in their use of local dialects. A more detailed description of the methodology of the sand project and the elicitation techniques that were used, can be found in Cornips & Jongenburger (2001) and Cornips & Poletto (2005).

Elicitation techniques

Four different methods were used to elicit the anwers to the questionsin the questionnaires:

Indirect grammaticality judgements. In this method, sentences in local dialect were presented to the informant who was asked to assess whether the sentences occurred in the local dialect. The advantage of using sentences in the local dialect is that these sentences are not discarded due to lexical or phonological properties. An advantage of the indirect procedure is that normative considerations are prevented as much as possible. Sometimes this method was used to get relative judgements. Variants of a particular sentence were offered to the informant. In addition to the question of whether a particular sentence occurred in the local dialect, the speaker was also asked to provide relative judgements, comparing different sentences.
reflexive and reciprocal pronouns (provided the informant with a picture and asked him/her to complete a sentence in the local dialect.
Translation. In many cases, the informant was asked to translate a sentence presented in the standard language into the local dialect.
Indirect grammaticality judgement + translation. A combinationof methods (a) and (c) was sometimes used. The informant was asked to judge whether a particular Standard Dutch sentence wascommon in the local dialect, and to translate the sentence into the local dialect. The advantage of this method was that the translation consolidated and evaluated the judgement.

Data collection

The empirical research for the sand project (2000-2003) was phased in the following way.

Development of an inventory of existing knowledge on syntactic variation. Among other things, this included a study of the literature (2000). The resulting bibliography can be consulted through the website of the Meertens Instituut ).
Written questionnaire (2000). A pilot study of 424 test sentences was carried out among 368 informants obtained from the database of the Meertens Instituut. The aim of this pilot was to get a first impression of the distribution of specific phenomena. This was necessary to make the final questionnaire as efficient as possible. The results of this study were used to assess the data that were collected in the oral interviews (phase iii).
Oral interviews (2001-2002). The interviews at the 267 measuring points (cf. Sampling points) consisted of a set of approximately 160 test sentences and a variable set of sentences that were selected on the basis of the results of the phases (i) and (ii). The interviews were recorded with DAT or minidisc recorders.
Telephonic interviews (2003). After an overview of the results of the oral interviews, we added a telephonic interview phase to complete the data set with data that were missing or unreliable. The telephonic interview consisted of approximately 100 questions for each measuring point. Telephonic interviews were held with informants from 252 locations. These interviews were recorded as well.

Digitalisation and transcription

The interviews were recorded with DAT recorders (in the Netherlands) and minidisk recorders (in Friesland and Belgium). These recordings were transferred directly to computers without conversion, using the Sadie DAW system. The sample frequency was 44.1 kHz, 16 bits. The result is that the quality of the recordings is high enough to allow phonetic research. Transcription of the recordings was achieved by means of the PRAAT programme (Boersma & Weenink; http://www/fon/hum.uva.nl/praat/). Phonetic transcription was impossible due to limited human resources and finances available and it is not really necessary for syntactic research. We made use of a normalised orthographic transcription with different rules for lexical and functional morphemes. Lexical morphemes were transcribed according to the rules of Standard Dutch, abstracting from phonetic and phonological differences. For instance, the verb /kinne/ on tape was transcribed as kennen ‘to know’. Functional morphemes such as inflection, pronouns and determiners were not normalised, since these morphemes are crucially involved in morpho-syntactic variation. These morphemes were literally transcribed, in such a way that a one-to-one correspondence between sound and orthography was approached. For instance, the sentence Wa denk je wien ik gezien heb (lit. what think you who I seen have) was transcribed without a /t/ on wa, whereas the normalised spelling is ‘wat’, and with an additional /n/ on wien, although the normalised orthography is ‘wie’. Clusters of morphemes were transcribed as a whole, since the boundaries between functional morphemes are often difficult to determine and require additional analysis. Consequently, the embedded clause dat ik het hem geef (‘that I it him give’) might be transcribed as ‘daketem geef’. A preliminary morphemic analysis was assigned in those cases (da-’k-’t-’m ...). A more detailed descriptionof the transcription guidelines is given in Barbiers & Vanden Wyngaerd (2001).

Organisation

Participating institutions

Meertens Instituut (KNAW, Amsterdam), University of Gent, University of Antwerp, University of Leiden, University of Amsterdam, and Frisian Academy (KNAW, Leeuwarden).

Financing

Flemish-Dutch Committee for Dutch Language and Culture (VNC), Dutch Organisation for Scientific Research (NWO), Foundation for Scientic Research Belgium (FWO), Meertens Instituut, Royal Netherlands Academy of Arts and Sciences (KNAW), Research Foundation of the University of Antwerpen, and the Royal Academy of Dutch Language and Literature (KANTL, Gent).

Initiative and supervision

Hans Bennis (Meertens Instituut), Hans den Besten (University of Amsterdam), Magda Devos (University of Gent), Johan Rooryck (University of Leiden), and Johan van der Auwera (University of Antwerp).

Project management

SjefBarbiers (Meertens Instituut), Hans Bennis (Meertens Instituut), Magda Devos (University of Gent), Guido Vanden Wyngaerd (University of Antwerp/Catholic University of Brussels; until December 2001), and Johan van der Auwera (University of Antwerp).

Methodology

Leonie Cornips (Meertens Instituut) and Willy Jongenburger (Meertens Instituut; until September 2001).

Coordination

Tamar Israël (April 2000 until September 2002), Jeroen Van Craenenbroeck (December 2001 until April 2003), Margreet van der Ham (April 2001 until January 2004), and Susanne van der Kleij (January 2000 until February 2002).

Field work

Composition of the questionnaires:
SjefBarbiers (Meertens Instituut) and Guido Vanden Wyngaerd (University of Antwerp/Catholic University of Brussels).
Field work:
Belgium: Gunther De Vogelaer (University of Gent), Annemie Neuckermans (University of Gent/Antwerp), Hugo Ryckeboer (University of Gent), and Vicky Van den Heede (University of Antwerp/Gent).
JeroenVan Craenenbroeck (University of Leiden), Margreet van der Ham (Meertens Institute), Susanne van der Kleij (Meertens Instituut), Marjo van Koppen (University of Leiden/Meertens Instituut), Vivien Waszink (Meertens Instituut), and Hedde Zeijlstra (University of Amsterdam).

Transcription

Transcription protocol: Sjef Barbiers (Meertens Instituut) and Guido Vanden Wyngaerd (University of Antwerp/Catholic University of Brussels).
Transcription:
Belgium: Gunther De Vogelaer (University of Gent), Annemie Neuckermans (University of Antwerp/Gent), Vicky Van den Heede (University of Gent/Antwerp).
Friesland: Arjan Hut (Frisian Academy), Bouke Slofstra (Frisian Academy), and Henk Wolf (Frisian Academy).
SjefBarbiers (Meertens Instituut) and Guido Vanden Wyngaerd (University of Antwerp/Catholic University of Brussels).
Tagging:
Irene Haslinger, Alies MacLean, Margreet van der Ham, Marjo van Koppen, Vivien Waszink (Meertens Instituut), and Bouke Slofstra (Frisian Academy).

Software-development

Jan Pieter Kunst (Meertens Instituut): database, search engine, tagging application, cartografic software; Ilse van Gemert (Meertens Instituut): cartographic software; Kees Grijpink (Meertens Instituut): digitisation.

Cartography

Selection and check of the data:

Sjef Barbiers (Meertens Instituut), Gunther De Vogelaer (University of Gent), Irene Haslinger (Meertens Instituut), Jeroen Van Craenenbroeck (University of Leiden), Vicky Van den Heede (University of Gent/Antwerp), Margreet van der Ham (Meertens Instituut), Marjo van Koppen (University of Leiden/Meertens Instituut).

Production of symbol maps:
Margreetvan der Ham with help from Jan Pieter Kunst and Marco René Spruit (Meertens Instituut).

Commentary

SjefBarbiers, Johan van der Auwera, Hans Bennis, Eefje Boef, Gunther De Vogelaer and Margreet van der Ham. Barbiers and Bennis had the main responsibility for chapters 1, 2 and 3; Van der Auwera and De Vogelaer for chapter 4.

Acknowledgements

Thanks are due to:

all of the informants;
the members of the board of advice: Marinel Gerritsen, Ton Goeman, Liliane Haegeman, Aafke Hulk, Willy van Langendonck, Jef van Loon, Pieter Muysken, Anneke Neijt, Johan Taeldeman, Cor van Bree, Roeland van Hout, Piet van Reenen, Henk van Riemsdijk, and Jan-Wouter Zwart;
the readers of the preliminary chapters of the commentary: Frits Beukema, Hans Broekhuis, Gert De Sutter, Magda Devos, Marcel den Dikken, Walter Haeseryn, Joop van der Horst, Olaf Koeneman, Gertjan Postma, Henk van Riemsdijk and Willy Vandeweghe.

References

Barbiers, S., L. Cornips & J.P. Kunst (2007). ‘The Syntactic Atlas of the Dutch Dialects. A corpus of

elicited speech as an on-line Dynamic Atlas.’ In J. Beal, K.P. Corrigan & H. Moisl (eds.) Models and Methods in the Handling of Unconventional Digital Corpora. Volume 1: Synchronic Corpora. Palgrave-Macmillan.
Barbiers, S. & G. Vanden Wyngaerd (2001). Transcriptieprotocol voor de Syntactische Atlas van de
NederlandseDialecten. Manuscript, Meertens Instituut.
Cornips, L. & W. Jongenburger (2001). ‘Het design en de methodologie van het sand project.’
Nederlandse Taalkunde 16:3. 215-232.
Cornips, L. & C. Poletto (2005). ‘On standardising syntactic elicitation techniques I.’ To appear in
Lingua.
Daan, J. & D.P. Blok (1977). Van Randstad tot Landrand. Toelichting bij de kaart Dialecten en
Naamkunde. Amsterdam: Noord-Hollandsche Uitgevers Maatschappij.
Wattel, E. & P. van Reenen (1994). ‘Visualisation of extrapolated socialgeographic data.’
[Rapportnr. WS-429, Faculteit Wiskunde en Informatica, Free University Amsterdam].
sand syntax: A guide to the use of this atlas
Commentary
The commentary is kept as neutral as possible; we did not try to include potential explanations of

the relevant data from a theoretical or historical perspective. Such explanations are dependent on theoretical persuasion and the state of the art of linguistic theory. We opted for a presentation that is as free from theory as possible. We leave it to the linguistic reader to provide explanations for these phenomena.

Each chapter of the commentary has an introduction in which the phenomenon is described in

general terminology.

Each chapter of the commentary contains a short historic overview of the phenomenon in question.
The most relevant literature is discussed in the text of the commentary. At the end of each chapter, a more comprehensive list of references is provided.
For each map, a separate discussion in the commentary is provided. This discussion includes a motivation for the inclusion of the particular phenomenon, the manner in which the phenomenon was tested and striking patterns in its distribution.
In the commentary, the number of the map refers to the page number in the atlas.
The commentary ends with an index in which important terms and forms are included.
In the commentary, the term ‘Vlaanderen’ has been reserved for the combination of the Belgian provinces of West-Vlaanderen and Oost-Vlaanderen, together with the Dutch region Zeeuws-Vlaanderen and Frans-Vlaanderen.
The term ‘Brabant’ refers to the provinces of Noord-Brabant in the Netherlands and Antwerpen and Vlaams-Brabant in Belgium; the two Belgian provinces together are referred to as ‘Belgian Brabant’.
The term ‘Limburg’ is used to refer to the combination of Dutch Limburg and Belgian Limburg.
The term ‘Holland’ refers to the combination of the provinces of Noord-Holland and Zuid-Holland.
In the ordering of references, we have opted for the Belgian variant: prefixes such as van or de are taken as being relevant for determining the position in the references.
Atlas
The atlas begins with a map with the names of the locations that were used as measuring points and with the names of the provinces. This map can be placed alongside the other maps in the atlas by folding it out.
Beside the fold-out map, there is an alphabetical list of measuring points with Kloeke-numbers.
Following the fold-out map, a map is included that illustrates the language situation as it was in 1939; this map allows a comparison with the situation in 2000.
Symbol maps make use of coloured squares.
If locations have squares with the same colour, it follows that they have the same syntactic variant; if they have related colours (e.g., light green vs dark green), it refers to related constructions.
If locations have more than one variant, the location is marked with more than one coloured square; a maximum of six squares is allowed for one location (three horizontally and two vertically) and these are ordered in a fixed manner. For clarification, the fixed order of coloured squares is represented in the map keys.
Coloursfor specific phenomena are kept constant within a chapter as much as possible.
A grey dot indicates that none of the variants that are described in the map key have been attested in that particular location, or that no relevant data were available.
Some maps have big and small grey dots. Small dots indicate that the variant is not tested, while a big dot indicates where the variant is tested, but not found.
The number of the paragraph of the commentary in which the map is discussed, is given in the map key.