‘Correct’ Gronings and How to Measure This Value

(1)

‘Correct’ Gronings and How to Measure

This Value

Melissa Boven S2530899 MA Multilingualism University of Groningen August 2019 Supervisor: dr. N.H. Hilton Second reader: prof. dr. G.T. Jensma

(2)

Abstract

This thesis seeks to answer what is correct Gronings and whether there is a method to measure this value efficiently. A language test is constructed to determine speakers’ proficiency. This test consists of a Picture Naming Task, a word reading task and a counting task. It is thought that the Picture Naming Task in itself will be a way to validate self-reported proficiency scores, but the results vary. Since the word reading task and counting task cannot be judged objectively due to there not being a standard language, the help from language professionals is enlisted. These professionals are asked to rate the recordings made of participants in order to establish whether there is a shared ideology present among professionals. The language professionals judge highly subjectively and the scores they give the participants are varied. Without the word reading task and counting task the language professionals do not have enough information to form an opinion. When all test items are combined the language professionals are able to, according to them, judge persons’ proficiency in Gronings. They judge mostly on fluency, which they described as a combination between speed and accuracy, with pronunciation being an important factor as well. Even though part of this thesis’ aim is to construct a language test, it seems that it is clear that a shared language ideology was present among the language professionals – they knew what was correct Gronings (according to them) and what was a ‘lesser’ language. In conclusion it seems like it is possible to construct a language proficiency test for a non-standardised minority language.

(3)

List of Tables

Table 1 Images used in the Picture Naming Task p. 20

Table 2 Words that needed to be read out loud p. 21

Table 3 Data on all people interviewed at the Dag van de Grunneger Toal p. 25

Table 4 Nine people chosen for judgement by language professionals p. 27

Table 5 Picture Naming Task scores related to the self-reported proficiency

scores p. 31

Table 6 Answers given by the nine persons selected for the language

professionals p. 32

Table 7 The scores given by W. Bastiaanse p. 32

Table 8 The scores given by dr. C. Bergmann p. 34

Table 9 The scores given by W.J. Friedrich p. 36

Table 10 The scores given by J. Glas p. 38

Table 11 The scores given by drs. J.J. Groenbroek p. 39

Table 12 The scores given by M. Hillenga p. 40

Table 13 The scores given by prof. dr. G.T. Jensma p. 41

Table 14 The scores given by drs. I. Nienhuis p. 43

Table 15 The scores given by ds. K.G. Pieterman p. 44

Table 16 The scores given by H. Schoonhoven p. 45

Table 17 The scores given by F. Schreiber p. 46

Table 18 The scores given by T. Ufkes p. 47

Table 19 The scores given by H.R. Wierth p. 49

Table 20 All participants’ overall scores as judged by language professionals p. 50 Table 21 All participants’ Part I (PNT) scores as judged by language

professionals p. 51

Table 22 All participants’ Part II (counting and reading) scores as judged

by language professionals p. 52

Table 23 All participants’ overall scores as judged by language

(5)

1 – Introduction

The idea for this thesis came from prof. dr. G.T. Jensma who claimed that by watching short interviews conducted at a 2017 Google X event he could determine a speaker’s proficiency in Gronings. At the 2017 event visitors were approached with a short test on the Groningen language. They had to read aloud some words and count to ten. In the more than hundred videos recorded on that day, many people claimed to speak the language – yet they were not able to fluently count to ten or read out loud some words. Seeing as prof. dr. G.T. Jensma is a language professional, I thought that it would be interesting to compare multiple language professionals’ ideologies to that of a form of standardised language (a dictionary) and speakers’ own opinions about their language abilities. The Google X incentive could thus be seen as a pilot for the current study. This thesis shows whether there is a shared feeling of correctness among the language professionals and the minority language’s community as a whole and whether this ideology differs from the standard as prescribed by a dictionary. In short, this thesis examines what constitutes correct Gronings according to language professionals and whether there is a method that can measure this value. Simultaneously, the relationship between self-reported proficiency, actual proficiency as measured with a Picture Naming Task and proficiency as judged by language experts will be discussed.

(6)

2 – Theoretical background

2.1 – Political profile of Gronings

All speakers of Gronings are bilingual, seeing as Dutch is the majority language that greatly influences all speakers of Gronings. Gronings itself is not even acknowledged within The Netherlands as being a language in its own right. Gronings is one of the languages that together with, among others, Twents and Drents are known as Low Saxon. Low Saxon is classified in The Netherlands as a regional language, together with Limburgian. Low Saxon has had this status since 1996. This acknowledgement means that provinces and municipalities are allowed to make their own policies and act on the language however they see fit (Rijksoverheid, z.d.).

Gronings has many variations throughout the province and around the province border. Since the publication of Ter Laan (1923, cited by Hoppenbrouwers & Hoppenbrouwers, 2001) Gronings is distinguished in seven areas: Hunsingo, Fivelingo, Oldambt, Veenkoloniën (all Central Gronings), Westerkwartier, Gorecht and Westerwolde. However, Heeroma (1963, cited by Hoppenbrouwers & Hoppenbrouwers, 2001) writes that the Veenkoloniaals variation is a mixed dialect of among others the Oldambt language, instead of a pure Groningen dialect. He also states that Westerwolds is actually not Gronings, but Drents. Reker (1995) divides Gronings in just North, East and West. The city of Groningen itself is not taken into account here. In his publication, however, he does distinguish between areas within each wind direction he mentioned. In a language guide from 2005 he differentiates between Groningen-Stad (Groningen-City), Oldambt-Plus, Veenkoloniën, Westerkwartier and Hogeland. However, in all publications the regions are distinguished based on the pronunciation of certain words. This means that it is clear that there are different types of Gronings that are distinguishable from one another and that there may be a possibility that certain types are preferred over other types.

2.1.1 – Speaker numbers

A recent study on traditions and customs in the Groningen culture by the Sociaal Planbureau Groningen (2015) includes an estimate on speaker numbers for the Groningen language. The study was done in the form of a survey sent out to a panel of 1426 persons. The survey called Trots op Groningen: beeldvorming en identiteit (Proud of Groningen: imaging and identity) covered questions about the province, the language, regional binding and more. According to the panel results, four out of ten people still speak the language on a daily basis. Furthermore, 73% of people that partook in the survey were able to understand Gronings either ‘well’ or

(7)

‘very well’. The panel members were also asked how important they deemed it to be able to speak and understand the language. 36% of respondents found it important to be able to speak Gronings, opposed to a slight majority of 39% who did not find it important at all. The remaining 25% answered neutral. Understanding the language yielded higher results, 70% of respondents thought it was important to be able to understand the regional language. Only 13% answered that it was not important to be able to understand the language. From the younger generation, in this survey consisting of 18 to 34-year-olds, 59% deemed it important to understand Gronings. The older generations all found it more important. Out of 35 to 49-year-olds 64% found it important, whereas from 50 to 64-49-year-olds and 65+ years old respectively 75% and 72% answered positively.

Included in the survey were also questions on whether people not only understood or spoke the language, but were able to write and read in the language. The numbers for reading were already not high, since only 33% of respondents said they were able to read the language ‘very well’ (11%) or ‘well’ (22%). The percentage of people that were able to write the language was even lower, only 9% of participants said that they were able to write Gronings either ‘very well’ or ‘well’. 27% said they were able to write Gronings with great difficulty, whereas the majority of respondents (52%) answered that they were not able to write in the language at all.

It is questionable, however, how reliable the above described data from the Sociaal Planbureau Groningen is. There is no readily available information on the composition within the panel, and how the panel members were recruited. There is one other study, however, that also claims that speaker numbers are still high. However, this is a study from 1995. Reker (2002) writes that said 1995 study by RTV Noord (a regional broadcaster) showed that 79% of inhabitants of the province of Groningen older than 65 said they were speakers of Gronings. In the category 18 to 25 year old this was 44%. A more recent estimate from 2005 by Streektaal.net said that almost 1.8 million people still spoke a Lower Saxon dialect, which amounts to 60% of the total population of the Lower Saxon language area. Unlike the study from the Sociaal Planbureau Groningen, however, these studies do not include how well participants think they speak or understand Gronings.

Driessen’s (2006, 2012) studies have shown that the amount of dialect speakers in the Netherlands is decreasing to a very slim amount, painting a more grim picture than all studies cited above. In 1995 (the same year as the study from RTV Noord), 2001 and 2011 a total of 400 elementary schools were visited, which amounts to data from 7000 pupils per poll. In these years, Driessen gauged how many parents still spoke in dialect among each other,

(8)

whether parents (fathers and mothers measured separately) spoke dialect with their children and whether children amongst each other spoke dialects (with either their siblings or in the school yard, two different measures). One of the dialects that Driessen studied was Lower Saxon, the dialect continuum that Gronings is part of. In his 2012 report he writes that in 1995 34% of parents still spoke dialect among each other. In 2001 this was decreased to 24% and in 2011 this was only 15%. The amount of children that spoke dialect among each other in for example the school yard was lower: in 1995 6% of children spoke in a Lower Saxon dialect, in 2001 this went down to 5% and in 2011 only 1% of children was said to speak in dialect with other children. The other numbers do not differentiate much from these: 7% of children spoke dialect with their mothers in 1995 and in 2011 only 1%. With their fathers 8% spoke dialect in 1995, whereas in 2011 this was decreased to 2%. In 1995 7% of children spoke Lower Saxon with their siblings and in 2011 this also decreased to 1%.

It is clear that there are no true speaker numbers available for either Gronings or Lower Saxon as a whole. This can be attributed to the fact that none of the studies mentioned above have tried to define what proficiency is needed to call someone a speaker of a language. Thus, it is necessary to find out what classifies as a speaker and how you can measure that, especially when the language in question is not standardised. The only thing that is found in all studies, is that over generations the use of dialect is decrease. Heeringa and Hinskens (2012, 2015) have confirmed this decrease in dialect usage as well. They say that the decreasing amount of speakers is due to not only dialect occurring less in daily life, but also because of dialects changing into so called regiolects. According to the authors, a regiolect is a language that is spoken in a larger area than a dialect. They found that dialects not only converge to Dutch, but that they are also becoming more similar to each other. This is due to the influence of Dutch and the fact that there are less differences remaining. In their 2009 study Heeringa, Neerbonne and Kleiweg talk about the same phenomenon. They measured the distance between sounds from different dialects and found that there is a shift taking place from dialect to regiolect. Their examples include the Gronings words scheuvel(e)n (ice skating) that is becoming schoatsen (‘schaatsen’ in Standard Dutch) and the very typical word kopstubber (cobweb remover) that is becoming roagebol (‘ragebol’ in Standard Dutch). These examples show that the dialect word is making place for the Dutch variant, but keeping with a dialectal way of pronunciation.

It has become clear that current research is unable to tell us how many speakers of Gronings are left. There is also no reliable data on how well the speakers of the language still speak it and how often they use the language. However, classifying what is ‘right’ and what is

(9)

‘wrong’ language is another problem altogether. For dialect revitalisation and the creating of learning tools a standard variety that is accepted as the ‘right’ language needs to be established. Gronings is not standardised, although standardisation attempts have been made in the past. K. ter Laan (1871-1963) published his Nieuw Groninger Woordenboek (New Groningen Dictionary) in 1929. He was the first one to create a spelling method for Gronings. However, since then Siemon Reker (1950) has published a new dictionary, together with having developed a new spelling. This spelling is generally used by authors, with some influences from K. ter Laan. Reker’s spelling is seen as the standard. However, it is not officially documented as being the standard of the written Groningen language (let alone having a standard). Since Gronings is not taught in schools, most people do not know how to write in the language (Sociaal Planbureau Groningen, 2015). However, it is possible that speakers do have a (shared) feeling of what is correct language and what is not and this opinion could differ from the standard as prescribed by Reker. In order to fully understand language standards we need to consider language standardisation.

2.2 – Language standardisation

Language is one of the most important symbols of (national) individuality and independence. It is not only a means of social cooperation, but also a means of individual expression. The first leads to uniformity and the latter leads to diversity. Ideally, a balanced form of the two would be used by the majority to ease communication. This is where a standardised language comes in, preferably one where there is room for diversity without imposing a single standard on users with diverse dialects (Haugen, 1972). When referring to a standard language, it is often the written form that is meant. The structures of spoken language are neglected (Bex & Watts, 1999). In spoken language we tolerate more variety than in written language, where something is often considered wrong because it does not comply with the rules we have been taught in schools. According to Milroy and Milroy (1999), standard language is an ideology and not a reality. It is a set of abstract norms to which the user may choose to conform, either to a greater or lesser extent. Standardisation is motivated in the first place by various social, political and commercial needs and this mostly includes the writing system because it is often easily standardised compared to the oral tradition. Haugen (1972) states that with sufficient motivation, written and maybe even spoken language can be deliberately guided and changed. Ferguson (1962, as cited in Haugen, 1972) states that most languages are not standardised at all and are not used for writing. This could be considered the ‘normal’ state of language,

(10)

seeing as this means that languages that are standardised and have written forms are actually less frequent. Ferguson found that writing and standardisation is often imposed by governments, schools, churches, et cetera, but that almost no-one speaks according to the rules. Similar findings are described by Bex & Watts (1999).

Haugen (1972) describes language standardisation as the result of a process called language planning: the management or manipulation of language in which normative and/or prescriptive linguistics are combined. Language planning was first defined as an activity of preparing a normative orthography, grammar and dictionary for the guidance of users in a non-homogenous speech community. However, Haugen later saw this as the outcome of language planning and not as the actual language planning process itself. The process of language planning, according to him, is the exercise of judgment in the form of choices among available linguistic forms. Planning implies an attempt to guide the development of a language in a direction desired by the planners (and thus perhaps not the speech community). Language planners want to influence users’ choices and in doing so, they assume that there must be a standard against which these different linguistic innovations can be evaluated. Language planning is not committed to one task in particular, the ideology behind it is very variable. Among other things, it could be used to either promote or prevent change, advocate either uniformity or diversity or it could be used for language maintenance. Language planners see writing as the primary language form and therefore spoken language is secondary. The language changes proposed by a language planner are often shaped by the speech habits of a social elite. This has a result that the newly created standard language has two mutually supporting aspects, namely on the one hand a generally accepted orthography (seeing as the elite are mostly also the governing and thus prescribing party) and on the other hand a prestige dialect that the socially ambitious would want to imitate. Language planning can be organized and deliberate, but also spontaneous. It can be undertaken from anywhere on the scale from private individuals to dictators. There is a wide range of organizations which have undertaken language planning: academies (as in France and Italy), churches, government commissions, schools of literature and science, et cetera.

Haugen divides language standardisation in four steps: selection, codification, elaboration and acceptance. This means that the first step is to select the language that you wish to standardise. Haugen states that there is a certain ‘ecology of language’ that consists of more than just the grammar, rules, etc. of a language (that have been prescribed by a certain party). We should also consider the social status and function of the language. Language ecology is the study of interactions between any given language and its environment. It is

(11)

primarily determined by the people who learn the language, use it and transmit it to others. Voegelin and Voegelin (1964, as cited in Haugen, 1972) advise that one should not pay selective attention to just a few languages in an area, but that one needs to consider all the languages in that area before picking the one (aspect of) language you wish to standardise. After the selection, codification takes place. In this phase a language planner develops a model for the language (usage). The elaboration phase includes enlarging the vocabulary. The acceptance stage is the last stage, where the plan is presented to the public and (hopefully) accepted. The government can help with this acceptance by implementing the newly standardised language.

When deciding what the standard should be, the ‘right’ way of speaking a language is often kept in mind by the prescriptive party (Milroy & Milroy, 1999). This standard language ideology may be present for Gronings as well and there are people who could be considered for a prescriptivist role, these people are interviewed for this thesis as well. Noreen (1982) states that wrong language is that what is misunderstood, not understood at all or understood with great difficulty. Bad language is also hard to pronounce, hard to remember, over-complicated or longer than necessary. Wrong language is also that what is novel without really adding anything of value to the language. All these things can be summarized as aspects of efficiency. Correct language is that what is easily understood without unnecessary words. However, as previously stated, it is often just the written tradition for which standardisation can be easily enforced and where correct language can be more easily distinguished from incorrect language.

‘Correct’ language is, as already briefly discussed above, often the language of the elite. This language is often the most prestigious variant and the one that plays the largest role when standardising a language (Bex & Watts, 1999). For the current study this means that there is some sort of elite with an opinion that they want to prescribe to all other speakers of a language. Regional languages or dialects are not as prestigious and therefore are often neglected when deciding on a new standard. Reker (2002) writes that the attitude towards Gronings from speakers and non-speakers is very different, especially young people that do not speak the dialect are very negative towards it. Age seems to play an important factor in whether or not someone finds the language to be of any importance. Reker shows that the older a person is, the more value they attach to the language. The statement “Gronings is important” was judged to be “very important” by only 14% of 18 to 25-year-olds and 28% said “not important”. However, 45% of people that were 65 years of age or older claimed that they found the language very important. Overall more than 50% of respondents said Gronings

(12)

was either important or very important to them. Reker also found that education level and Gronings were negatively correlated. The lower someone was educated, the more often they spoke Gronings. Surprisingly, however, he also found that the higher someone was educated, the more value this person attached to the survival of the language. Reker thinks that there may be a possibility that non-speakers attach more value to Gronings than speakers, seeing as he found that in Groningen the majority of people thought that Gronings was the most important handicap in their steps on the social ladder. According to them, in order to have opportunities and to be able to move further out into the world than just the nearby region, they need to speak Standard Ducth instead of the traditional dialect spoken in their direct environment.

Minority languages can also be subject to standardisation. An example of this is Basque. According to Oñederra (2017) the standardisation of written Basque has so far followed Haugen’s four definitial features (selection, codification, elaboration and acceptance). According to her, it would be good for spoken Basque to also follow these steps. A standard pronunciation has already been proposed, but it has not been widely communicated to the public. This is mainly because of the different cultural, political and historical situation of today being different from the dictatorship under which standardised written Basque was introduced. Standardisation implementations almost never have an effect on spoken language. However, according to Oñederra an attempt at standardising spoken Basque should be made. She sees it as a way to preserve the language. There are a lot of new speakers of Basque who are not bound to a certain dialect. They could benefit greatly from a standard that prescribes careful pronunciation of the language. She does advise language planners to take variation into account and she argues that stylistic variation should stay accessible for those speaking the more dialectal forms of Basque.

2.2.1 – Problems in language standardisation

According to Haugen (1972) problems with language standardisation arise when dealing with an illiterate population without a written tradition (thus, nothing to standardise). Problems could also arise when there are competing orthographies or inadequate orthographies present. As described above, there is a consensus on which ortography to use when writing Gronings, but this writing system is not known to most of the speakers. Online on websites as Facebook, where you find most of the speakers using the written form of the language, you see that speakers are using a form which they deem correct. This is usually visible when using verbs as ‘kopen’ (to buy), which in correct Gronings would be ‘kopen’ as well, but a layman would

(13)

write is as ‘koopm’ (or similar, but with an apostrophe before the ‘m’), seeing as that is more close to how it is pronounced (these are my own observations).

Milroy & Milroy (1999) state that non-standard speakers are those who suffer from language standardisation. According to them, standardisation is a limitation of diversity. There are larger minority languages that have been standardised or have made an attempt at standardisation (such as Basque, see example above). Lane (2011) states that promoting language standardisation is a way to validate groups, but at the same time it also limits group-internal diversity. Furthermore Lane states that developing a standard for a minority is never a neutral process and that this has consequences for the status of the language and how its users relate to the new standard. One of the inherent problems with standardisation is whether the language users will identify with the standard and accept is as their own. When a language is being standardised, there is always a risk that the newly established standard does not resonate with its users and thus, a standardisation attempt that was supposed to empower a group of minority language speakers may create a new form of stigma for those that do not feel that they can live up to this new standard. This can cause speakers to reject the standard or to abandon their (minority) language altogether. In conclusion, developing a standard for a minority language (or any language for that matter) may not always have a positive outcome. This is also cited in Costa, De Korne and Lane (2017).

It has become clear that standardisation is a process that can entail many different facets of language. Determining what should be the standard is not always easy, seeing as deciding what is ‘correct’ and ‘incorrect’ language is often subjective (Noreen, 1982) or based on prestige (Bex & Watts, 1999; Milroy & Milroy, 1999). It is therefore not surprising that speakers of minority languages or non-standard speakers in general find themselves becoming victims of standardisation (Milroy & Milroy, 1999), seeing as they might not agree with a standardised version as imposed by language planners (Lane, 2011). Language planners can be anyone from private individuals to dictators and as shown by Haugen (1972) they are mostly males from a higher social class. When considering the current study one may wonder whether Gronings is in urgent need of a standard. Standardisation attempts are currently being made in the form of a project called Woordwaark. This project is one giant database containing all possible forms of Gronings as used by speakers themselves, or in other words: it is a large dictionary. This dictionary is a form of codification, but no-one is sure whether there is a general sense of acceptance among the language community. Are the speakers of Gronings ready for their language to be codified and standardised? Or is there no public acceptance and is the project just an example of prescriptivists at work? For this thesis

(14)

language professionals were asked to decide on what is right and what is wrong language. Their opinions could be seen as a codified form of Gronings as well. These professionals were mainly older males, something that will be discussed in the methodology section as well. The language professionals’ opinions were needed to examine whether there is some sort of unofficial standard that they all adhere to and that they deem to be the correct way of speaking a language. Only when there is a shared sense of correctness you are able to construct a language test and to test people’s proficiency on basis of this codification as well.

2.3 – Language testing

Language testing can in itself be seen as a form of standardisation, but standardisation does not need language testing per se. When creating a test, one must decide on what are the correct answers. When one specific answer is deemed to be better than another, you could say that the test maker has a preference for a certain form (or standard). This does not apply when turning it around. When you have standardised a language, you do not necessarily need a test – until you want to test proficiency, achievement or something similar. Tests for non-standardised languages are not common and this may be obvious, seeing as there is nothing to refer to when creating a test. What are the correct or desired answers when there are no guidelines to adhere to? In these cases a language test could be made on the basis of available codification. For Gronings a method is still being developed, as will be shown in this thesis.

Language testing is often an important part of language teaching and learning (Douglas, 2014). The results of language tests can impact a language learner’s life. Proficiency tests, for example, are useful to make predictions about someone’s ability to use the language in communicative situations. Douglas (2014) states that a language test is a measuring device that measures language ability. Language tests are often complex, because of language itself being complex with many features that can be measured. However, language testing does not just exist for determining someone’s proficiency.

Bachman (1990) writes that there are two major uses of language tests. The first is as a source of information for making decisions within the context of educational programs and the second one is as indicators of abilities or attributes that are of interest in research on language, language acquisition and language teaching. Alderson, Clapham and Wall (1995) explain that language testing exists for multiple purposes. According them there are five broad categories that tests tend to fall into: placement tests, progress tests, achievement tests, proficiency tests and diagnostic tests. Placement tests are designed to assess students’ level of language ability, so that they may be placed in the appropriate class. Progress tests are given

(15)

at various stages throughout a language course to see what a student has learnt. Achievement tests are similar to progress tests, but they are usually given at the end of a course. A proficiency test is not based on a particular language programme, but they are designed to test the ability of students with different language training backgrounds. Diagnostic tests seek to identify those areas in which someone needs further help. Diagnostic tests also includes test that check for dyslexia, dementia or aphasia.

The construction of a language test follows different phases. First of all you need to have an idea about what you want to test. If you know what you wish to test, you can write the so-called test specifications. In the test specifications you specify the purpose of the test, who the test is meant for, how many tasks the test has and what its overall duration will be, what the target language situation for the test is, what type of text types or exercises need to be chosen (fitted to the type of audience), what language skills will be tested and what elements should be tested, what sort of tasks are required, how many items are required for each section, what methods will be used, what rubrics are to be used as instructions for candidates and finally which criteria will be assessed by the markers. After writing the test specifications, you can start writing items and moderating the test. Moderating means that you enlist the help of a third party, so they can judge whether your test items measure what they ought to measure. It can be the case that you find one answer to be obvious, but that the third party (and therefore perhaps also a tested person) may think of something else entirely when seeing the question. When you have written your test, you need to pre-test and analyse it, to make sure it works in practise. If you know it works and you are not going to be testing all participants yourself, you need to train the examiners. You also need to cater for their reliability. After testing scores need to be reported on and marks need to be set (they can be fixed beforehand as well). It is important that the test is validated and after the testing period is over you could write post-test reports (which in this case, is this thesis). All tests could or should be reviewed and evaluated afterwards, either by the person who constructed it or a third party. (Alderson, Clapham & Wall, 1995).

As previously described, there are no clear speaker numbers or speaker proficiencies known for Gronings. This could be because of a lack of research, but also because some of the studies that do exist were problematic and not clear in what they defined as a speaker (and to what extent that person actually spoke the language, either according to professionals or according to themselves). Seeing as Gronings has not been standardised and there are no proficiency tests for it, another method for deciding on right or wrong answers was chosen, besides just adhering to Picture Naming Task scores. It needed to be verified by experts that

(16)

the Picture Naming Task was effective. There have been few to no (socio)linguists in the past that have studied proficiency in Gronings, but there are various people active in the work field. They use Gronings in their professional and/or personal lives and often they do know how to write in the language. They tend to adhere to Reker’s orthography and all of them seemed to have a certain standard that they themselves adhere to or that they think should be (or is) the standard of the language that other speakers should want to achieve. These professionals are regarded by their peers and by the public as having a certain status. Thus, these persons could be seen as the ‘elite’ as described by Haugen (1972). These persons were approached to participate in this study, seeing as their expertise on the language could be useful in deciding what is right and what is not when testing or judging a test or tested person. More information on this will be provided in the upcoming chapter.

Language testing problems arise when a test does not measure what it is supposed to measure. Bachman (2000) argues that validation research should be done more, to be able to create better language tests. For this thesis the problems with self-reported proficiency are important. Self-reported proficiency scores are reported in the literature as being unreliable, with speakers either over- or underestimating their own use and abilities. Shameen (1998) found that Fiji-Hindi speakers rated their proficiency higher than it actually was, this was validated by an oral language test. Martin-Jones (1991) finds something similar, namely that speakers may over-report their usage. This means that they say that they use the language regularly, but in reality they use the language less frequent than estimated by themselves. On the other hand, Nicholas (2009) found that speakers actually tented to under-report themselves. The over- and under reporting may be because of the social status that a language has (Holmes, 1997). Holmes calls self-reported proficiency inaccurate and unreliable.

2.4 – Current study

The aims of this thesis are to study what constitutes correct Gronings and whether there is a method to test this value. Our main research question examines what the self-reported proficiency scores of Gronings speakers are and how these relate to scores as tested with a language test that was based on the available codifications and how these in turn relate to what language professionals think to be the correct way of speaking Gronings. The general idea is that a short language test containing a Picture Naming Task (based on available codification), word reading task and counting task can show whether or not a person speaks the language correctly. The hypothesis, based on Haugen (1972) and Noreen (1982) is that the language professionals will each have a different, highly subjective opinion on what is correct

(17)

Gronings and what is not, based on their own preferences and their own ideology and possibly the ideology of their peers (the language of the elite they belong to). It is expected that with the help of the language professionals, proficiency can be measured.

Other questions that will be answered in this thesis are what the self-reported proficiency of the general public is, how this self-rated proficiency relates to measured proficiency in a Picture Naming Task and how these outcomes are assessed by language professionals. How much data the language professionals need to form an opinion and what they base their opinion on will also be examined, together with whether they share a sense of correctness or not. Whether there is a shared sense of norms in the speech community as a whole will be discussed as well. Based on Shameen (1998) it is thought that the general public will over estimate their own proficiency skills. This is also supported by the idea from Martin-Jones (1991) that people over-report how much they use their second language. This means that the hypothesis is that the self-reported proficiency of the general public will be high, whilst their actual proficiency as measured with a Picture Naming Task wil be lower. Language professionals will, as mentioned above, positively judge those speaking their own or their preferred type of Gronings, whilst having a clear dispreference for other types of Gronings. Assuming that they have their own opinions, it is thought that they will not need much information to form an opinion on someone’s ability to speak Gronings and whether it is – according to them – correct or good Gronings. This also implies that it is possible that the language professionals disagree with the constructed test and therefore think it is not a reliable tool for determining a speaker’s actual proficiency.

(18)

3 – Methodology

3.1 – Background

The test used in this thesis consists of both a Picture Naming Task and the reading aloud of some words, together with being able to count to ten. These three tasks were chosen mainly due to them being practical and easily understood by the (mostly elder) participants. The tasks are easy (for both the participant and the tester) and quick. A Picture Naming Task is often constructed to analyse cognitive deterioration (as is the case with Alzheimer’s disease). They are also used to screen for developmental reading disorders, because they measure verbal fluency. Picture Naming Tasks do occur in studies concerning bilingualism as well, as is the case in Gollan, Montoya, Fennema-Notestine and Morris (2005). They found that bilinguals named pictures in their dominant language more slowly and with more errors than monolinguals, but that after the fifth presentation they were just as quick as monolinguals. Bilinguals retrieved English picture names more quickly if they knew the name in both Spanish and English. Their findings suggest that bilinguals differ from monolinguals at a postconceptual processing level, that “implicit activation of lexical representations in the nontarget language can facilitate retrieval in the target language and that being bilingual is analogous to having a lexicon full of lower frequency words, relative o monolinguals”.

Kameyama (2018) investigated the production of late bilingual functions while executing controlled code-switching tasks among Japanese native speakers who use English at work. They needed to name pictures in one of the two languages. They answered in English more rapidly than in Japanese, suggesting that they were more prepared to use their second language than their native tongue. This could be summarised as having suppressed their native language more than their L2.

Jones et al. (1991) used fMRI to study brain activation during a bilingual Picture Naming Task and a word reading task. They found that activation is higher in bilinguals compared with monolinguals, because for bilinguals word retrieval is more demanding. They need to rehearse the articulation of each word and they need to carefully monitor their speech output to avoid errors when competition for word selection occurs between, as well as within, language.

The reading aloud of words is often judged by speed and not pronunciation. A word word reading task is successful when you read as fast as possible whilst making as little mistakes as possible. The speed (with points deducted for a wrong answer) determines your word reading fluency (Protopapas, Katapodi, Altani, Georgiou, 2018). In the current study

(19)

speed is less important than accuracy (or pronunciation in this case). However, as will be shown in the results section, pronunciation and speed often go hand in hand, seeing as in order to be deemed fluent you need a certain speed whilst talking correctly. The same thing applies to the counting exercise.

3.2 – Material

Alderson, Clapham and Wall’s instructions on how to make a language test were followed when constructing this test. After deciding what was going to be tested, with which participants (the visitors at the event that will be discussed in 3.3), a method for testing this needed to be constructed. In accordance with the literature in the previous chapter, it was decided to do a Picture Naming Task. Based on Gollan, Montoya, Fennema-Notestine and Morris (2005) and Kameyama (2018) it was thought that the task at hand would not pose any problems for the bilingual speakers, seeing as they should be able to access their lexicon to find the right target word just as well as a monolingual speaker would. Since Kameyama (2018) found that the L2 was actually accessed faster, it was thought that the participants (who were either L1 speakers for both Dutch and Gronings or L2 speakers of Gronings, but their usage of the language in their daily lives differed) would be able to provide the correct target words with ease. The fMRI study by Jones et al. (1991) showed that the activation in bilinguals was higher, meaning that they answered more carefully. Gronings speakers are by definition bilingual, seeing as Dutch is always an influence as well. It was therefore thought that a Picture Naming Task and a reading and counting task were good choices, seeing as they show how fluent a person is when pronouncing a word, whilst simultaneously their accuracy could be assessed. The words used for the reading task will be discussed below.

The target words for the Picture Naming Task were all nouns, seeing as images of nouns are easier to identify than verbs or other word categories. Since many words in Gronings are cognate to their Dutch counterparts, the goal was to find 10 words that were non-cognate. Haugen (1972) says that one of the end results of language planning, or standardisation, is having a dictionary. Seeing as there is no clear standard for the language, the words for the Picture Naming Task were taken from a dictionary (Reker, 1998). As you can see in Table 1 below, not all images had only one possible response. This can be attributed to the fact that there are different variations within the language and because there is a strong influence from the majority language Dutch. Not all responses that were acceptable for the target word according to the dictionary were non-cognate, the consequences and limitations of this will be discussed in the discussion section of this thesis.

(20)

Table 1: Images used in the Picture Naming Task

English Dutch Gronings

skate schaats scheuvel

trousers broek boksem; broek

moped bromfiets plof(fiets(e))

bread brood stuut; stoet

boot laars stevel; leers; leerze

girl meisje fam(ke); maaid; meid; moagdje; or I; orke; wicht I

daffodil narcis poaskebloum

ladle pollepel slaif; slief

pig varken swien

(plastic) bag (plastic) zakje / tas bule; buul; poil; pong; pongel; pude; puil; puut; tuutje / madde; mat; taas; taze

After selecting the words that were deemed suitable for the test, more than 10 black and white images that showed the target nouns were selected via Google Images. These images were subsequently tested for recognisability and some images were found to be unclear (either in what it projected or the quality of the picture itself). About 10 to 15 people from different ages were approached with the images and asked to name them in whatever language they felt most comfortable in (this included Dutch, Frisian and English). Eventually 10 images that were easily identifiable remained and were printed for use during the Dag van de Grunneger Toal 2018.

The words used for this test, see Table 2, were chosen because some of the words were both Dutch and Gronings, meaning that they could be read out loud in either language. As described by Jones et al. (1991) this meant that participants needed to be careful when selecting a word in their mental lexicons, seeing as errors could easily occur. A pronunciation that sounded Gronings, would indicate that a person had the correct target word image in their head when reading and processing the words. Other words in the test were typical Gronings and contained markers that, when pronounced correctly, are easily perceivable as Gronings and are estimated to be widely identifiable as Gronings by listeners. The first word, tegenvallen, could be read out loud in either a Gronings or Dutch way. This also goes for the words potje, dansen and Martinitoren. The word potje deserves a special mention, since it is an ambiguous word. When read as a Gronings word, the t-sound in the word shifts towards a d-sound. If read that way, it is clear that the reader understands the word as the typical Gronings word for a baby or an infant. If read in the regular Dutch way with a clear t-sound it

(21)

does not mean infant, but instead it means ‘jar-DIM’. The words loat, vergreld and nuver do not exist in Dutch and should be read out loud in Gronings.

Table 2: Words that needed to be read out loud

Gronings Dutch English

tegenvallen tegenvallen (to) disappoint

loat laat late

vergreld boos angry

nuver aardig, mooi nice, pretty

potje baby, dreumes baby, infant

dansen dansen (to) dance

Martinitoren Martinitoren Martini Tower

3.3 – Procedure

Throughout the province of Groningen there are educational institutes, theatres, musicians, comedians, authors and poets, and media all using Gronings as their medium of expression. The Dag van de Grunneger Toal (Day of the Groningen Language) is their annual showcase. During the Dag van de Grunneger Toal 2018 videos were made that were later used for this thesis. The day is an annual event about the language, during which visitors (a couple of thousand) can get acquainted with the variety and increasing popularity of the language (Huis van de Groninger Cultuur, n.d.). They can present their work and/or organisation to the public at an information market, or as one of the performing artists or speakers (these include university professors, authors, singer-songwriters, et cetera) that are present each year. The day is organised by the Centrum Groninger Taal en Cultuur (Centre Groningen Language and Culture), which is a recent merger of the Bureau Groninger Taal en Cultuur (Bureau Groningen Language and Culture) and the Huis van de Groninger Cultuur (House of the Groningen Culture).

During the Dag van de Grunneger Toal 2018 visitors were approached by second year bachelor students Minorities & Multilingualism from the University of Groningen with the question whether they would like to be both interviewed and quizzed on the Groningen language. For the students it was a way to get acquainted with field work that they would possibly have to carry out later in their lives. If they answered positively, they needed to answer general questions about themselves (birth place, age), rate their own proficiency on a scale from 1 (not proficient at all) to 10 (fluent) and answer some general questions about the language. These questions included whether they spoke Gronings on a regular basis and what they thought of the language. Since most of these questions are not relevant to this thesis, they

(22)

will not be discussed here. Only background information that was useful when selecting the participants for the second part of the study will be used. The full interview, translated into English, can be found as Attachment 1 for reference. At the end of the interview people were asked to participate in a Picture Naming Task, so that their actual proficiency could be established afterwards. The details of the Picture Naming Task will be discussed separately below. After the Picture Naming Task the interviewees had to count to ten and read aloud some words. The interviews were videotaped on tablets provided by the Multilingualism department from the University of Groningen. The images from the Picture Naming Task were given to the interviewees on a laminated sheet of A4 paper, with 6 images on the front and 4 images on the back, totalling to 10 images. The images can be found in Attachment 2. The words that needed to be read were taped to the back on the tablets in a large font, so that the interviewees would look into the camera when reading. The interviews were approximately four minutes long. The interview was constructed so that it would be fun for the participants to partake in a scientific experiment, without having the idea that they would have to do so. The majority of visitors were older, so beforehand it needed to be discussed with the students (in this study they were the examiners as described by Alderson, Clapham and Wall, 1995) how to approach the visitors and how to conduct the test. They were given the instructions on the day of the event. They were taught how to use the tablets and how to make sure the Picture Naming Task, word reading task and counting task were carried out in full. They were also reminded to ensure participants about their privacy (all participants signed a document detailing their participation).

The Picture Naming Task score was calculated by adding up how many responses were similar to the target words in Table 1. If eight responses were similar to those in the dictionary by Reker (1998), a person scored an 8. This means that the Picture Naming Task was an absolute test of the codification available for Gronings. The disadvantage of this is that language professionals did take accent into account as well, something we did not do when judging the Picture Naming Task according to the codification. The codification is just lexical. If you manage to produce the target word, with or without the correct pronunciation, you score a point. However, the method used by the language professionals could also have disadvantages: it could well be possible that they prefer one variant of Gronings over another. A slight deviation from the target word was allowed. For example, in eastern Groningen an –e insertion at the end of the word is common. Therefore, if someone said stoete instead of stoet for bread it was deemed to be correct. A complete change in vowels was not allowed. People that said brouk instead of broek or boksem for trousers did not get a point for that word. Since

(23)

it is not listed in the dictionary, the Dutch form tas for bag was unacceptable, whereas an eastern pronunciation with a vowel change (taze) was justifiable. Since the dictionary features no pronunciation rules, taze could either have a long a-sound or a short one, thus both were counted as correct responses. If a person said kunstscheuvel (figure skate) instead of just scheuvel (skate) it was still deemed correct, since the image was indeed a figure skate, so a type of scheuvel. The other way around was not acceptable. Some people referred to the daffodils as being bloumen (flowers), this was not specific enough. Most people used a diminutive for the little girl (wichtje), this was seen as a correct response, seeing as the target word was part of their answer. Whether someone used a plural or singular form for an image was not important at all, as long as they got the target word correct – the usage of a plural form when there was only a singular item on the image occurred almost solely for the word stevel or leers (boot), where people used the plural stevels or leerzen. Some participants hesitated or used another word before naming the target word, or said the target word and then doubted and said that was not the correct term. In both cases, the responses were counted as correct, seeing as they did say the word and clearly had it in their mental lexicon.

The people interviewed at the Dag van de Grunneger Toal may have responded different than they would have under normal circumstances, due to them being videotaped and watched by peers, students and others whilst speaking. For the language professionals the videos did play an important role, since they were also able to read someone’s facial expressions and some language professionals actually took these expressions into account (someone’s visual happiness, insecurity and hesitations were all mentioned and affected scores). Furthermore, the persons at the Dag van de Grunneger Toal were not approached in Gronings, but in Dutch. Even though Dutch may have been some people’s daily language, it may have caused an interference due to them being interviewed on Gronings at the Dag van de Grunneger Toal. This corresponds to the findings of Jones et al. (1991) who states that between two languages there’s a high level of interference and thus high possibility that errors will be made when not enough attention is paid to what is said. All interviewees being visitors to the day is in itself problematic as well. The 43 persons interviewed were incredibly biased. It may be assumed that everybody visiting the Dag van de Grunneger Toal has affinity with the language in one way or another, meaning that when participating in an interview the correct mindset is already present. Not only is the population biased, which means that possible different results could be yielded when conducting the same experiment elsewhere, the test itself is also limited as was already discussed above. The Picture Naming Task only accounts for one small part of our linguistic ability.

(24)

The semi-structured interviews consisted of two parts. During the first part of the interview, the professionals needed to answer some general questions about their ideology relating to Gronings and the work that they had carried out relating to the language. During the second part of the interview, the professionals were shown the nine videos from the Dag van de Grunneger Toal that were selected for judgement. They were asked to first judge the Picture Naming Task, rate the person and give an explanation on why that score was given. They then proceeded to the second part, where the person counts and reads, to give another score and explanation on why they scored a person that particular . They then had to give an overall score, which did not have to be the average of both previous scores, since their opinion could have changed based on the second part.

All language professionals were questioned on what they based their judgement upon as well. Their individual judgements will be discussed in the results section. The judgement criteria were very different. Some professionals took a mathematical approach for the first part of the video, stating that if in their eyes the person had eight words correct, they deserved an 8. Other professionals judged based on gut feeling, meaning that they based their scores based on whether or not they thought that the person in case spoke Gronings on a regular basis.

3.4 – Participants and language professionals

Eventually a total of 47 people were interviewed at the Dag van de Grunneger Toal. Afterwards their scores needed to be set, just as Alderson, Clapham and Wall (1995) stated. Out of the 47 participants, 3 did not speak Gronings at all and a fourth person was disqualified since she was helped by her husband with almost all of the words, making it impossible to determine her score. Some of the other participants were helped with one word as well, but in these cases the response was simply counted as being wrong (seeing as they were unable to produce the word themselves). The sex, age, self-reported proficiency scores and actual proficiency scores as measured with a Picture Naming Task of the remaining 43 interviewed visitors are displayed below in Table 3. Some of the self-reported proficiency scores are not a single number. Person 8 needs a special mention, since this person gave himself a 6 or a 7 for speaking, whilst he gave himself an 8 or 9 for his passive skills in the language. People’s self-reported scores are put in the table just how they said it. If they said “6 to 7” or “6 or 7” it is written in that way, or as Person 37 put it “at least a 9”.

(25)

Table 3: Data on all people interviewed at the Dag van de Grunneger Toal

Person Sex Age Self-reported Proficiency Actual Proficiency (PNT Score)

1 M 22 6 5 2 M 25 7 to 8 6 3 F 29 6 7 4 F 29 6 or 6.5 7 5 M 30 4 6 6 M 34 4 8 7 F 42 10 8 8 M 43 6 to 9 5 9 M 50 8.5 8 10 M 54 8.5 or 9 10 11 F 55 10 10 12 M 56 9 4 13 F 56 6 or 7 10 14 F 57 8 8 15 F 58 7 6 16 F 59 7 8 17 M 60 7 8 18 F 63 7 8 19 M 63 8 9 20 F 64 8 8 21 F 64 10 9 22 M 65 8 9 23 M 67 10, 9.5 9 24 F 68 5 8 25 M 68 6.5 9 26 F 69 between 8 and 9 9 27 F 70 10 9 28 M 70 7 or 7.5 10 29 F 71 6 6 30 M 72 7 8 31 M 73 8 8 32 M 73 9 9 33 M 75 5 7 34 M 75 6 or 7 8 35 F 76 8 9 36 M 76 7 or 8 4 37 M 76 at least a 9 9 38 M 77 7 5 39 M 77 8 7 40 M 77 6.5 5 41 M 79 9 8 42 F 82 8 7 43 M 91 9 9

Table 3 shows all interviewees’ data, ordered from youngest to eldest. The average age of the participants was 61,39 with a range of 69 (people from 22 to 91 years old were interviewed). The Picture Naming Task score has a mean of 7,67. The most prevalent score was an 8, which occurred 13 times. Two (2) people scored a 4, a 5 or a 6 were both scored four (4) times,

(26)

whilst a 7 score occurred five (5) times. Eleven (11) people scored a 9 and four (4) people scored a perfect score (10).

Out of Table 3 nine people were chosen to be judged by language professionals. The choice for these videos were based on multiple criteria. Due to the event where the videos were made being very noisy and students having accidentally covered the microphone with their fingers, many videos were close to inaudible for the average listener. Audibility was thus an important factor in selecting the videos. Another factor was that three videos needed to be of people that had the exact same self-rating score as Picture Naming Task score, three videos needed to be of people that actually scored higher than their self-reported proficiency score and another three videos were of persons that scored lower on the Picture Naming Task than their self-reported proficiency would predict. 5 women were used, opposed to 4 men. The ages were variable. All language professionals were shown the videos in the order as seen below in Table 4 (person 1 to 9).

Person 1 was a high scoring contestant, who rapidly named the images and confidently spoke the language. Since on paper she made no mistakes, she was shown first to all language professionals. Person 2 was of a similar age, but he named many of the images in Dutch, explaining his low score. His counting and reading, however, was fast. The idea was that this difference in both parts of the test would throw the language professionals off guard and show them early on that both halves of the test can be very different depending on the person. Person 3 was again in her fifties, but she did not claim to speak Gronings as well as many of the other visitors of the day. She did admit to being raised in it, but she said that it had since faded from her memory. Her Picture Naming Task score was similar to her self-reported proficiency, however it was slightly lower. Person 4 was again female, and her self-reported proficiency and actual proficiency as established with a Picture Naming Task were equal. She did not claim to score a 10, which she did not because she named two words in Dutch. She had to really think about some of the words, eventually getting them right. Person 5 is a young woman who has lived her entire life in the city of Groningen, yet she never really learnt it. She is interested in the language though, because she is drawn to music sung in the language. She scored slightly higher on the Picture Naming Task than she rated herself, however she was not able to count to ten or read out loud the words in a fluent manner. Person 6 was a woman that scored very high on the Picture Naming Task (a 9), but still below what she rated herself (a 10). Out of the 9 people used, she responded the fastest: her Picture Naming Task is very rapid. Person 7 was a man who scored himself only a 6.5, but on the Picture Naming Task he scored a 9 – a big increase. His counting and reading sound very fluent too,

(27)

presumably higher than a 6.5. Person 8 is a new speaker of Gronings, he is originally from Rotterdam and took a language course in Gronings out of interest. He gave himself a 4, since he could make himself understood, but admitted that it was not where he wanted it to be at. He did do very well on the Picture Naming Task, scoring an 8. His counting is not fluent and his reading is quite close to a native speaker’s pronunciation, which he is trying to imitate. Person 9 was by far the oldest person to have participated, however he showed that his mental capabilities were still excellent, in his interview he reviewed on the language of his youth that he had barely used since, because he was sent to a military school were speaking Gronings was prohibited. He scored himself a 9, which was also the result of his Picture Naming Task.

Table 4: Nine people chosen for judgement by language professionals

Person Sex Age Self-reported proficiency Actual proficiency (PNT score)

1 F 55 10 10 2 M 56 9 4 3 F 58 7 6 4 F 57 8 8 5 F 29 6 or 6.5 7 6 F 64 10 9 7 M 68 6.5 9 8 M 34 4 8 9 M 91 9 9

After selecting the nine videos that needed to be reviewed, language professionals were approached to judge these videos. The idea behind this is that standardisation is usually done by a certain elite (Haugen, 1972) and therefore an elite that could judge the language skills of others needed to be gathered. Certain persons were approached to participate and from here onwards they will either be called experts or, preferably, language professionals. In this case language professionals are people who (have) semi- or professionally work(ed) with or in the language of Gronings, as described in chapter two of this thesis. A third party (a consultant at the Bureau Groninger Taal en Cultuur) created a list of persons that he deemed to be suitable for partaking in this study as being an expert. According to him, they were his peers and they were experts on the language (either overall, or because they performed well in their respective fields). This meant that they were all qualified to judge about other people’s Gronings, seeing as they themselves were either fluent in it or worked with it. A lot of the persons were authors or at least wrote in the language often. It was very noticeable that there were no persons born after 1986 on this list. There were also almost no females on the list and

(28)

sadly due to some of the females not responding to an e-mail or them being away on holiday at the time of testing, only one female was interviewed. The lack of female professionals is not easy to explain, seeing as the possible interviewees were selected by a third party. It can only be speculated that there are either not many women working in or with Gronings, or that these women are not deemed to be experts or outstanding in their fields by their peers. That the professionals were almost all males coincides with Haugen’s 1972 statement that language planners are males from a certain elite. This is supported by the fact that most, if not all, of these language professionals were well-educated people with distinctive careers. The potential professionals were approached by e-mail whether they wanted to participate, if they responded positively a meeting was scheduled at either the university, their home or another place where it was guaranteed to be silent. One person was sent the instructions and videos by e-mail and the interview setting was imitated using Skype (dr. C. Bergmann). The youngest person that was interviewed as a language professional was born in 1986, whereas the oldest person was born in 1936. This makes the average age of the professionals at the time of interviewing 60 (ranging from 32 to 82).

The language professionals are discussed in the remaining part of this chapter. They are listed alphabetically and any titles are noted in Dutch. Part of the information below was based on what they themselves told during their respective interviews, whilst other information was taken from online sources.

W. (Wim) Bastiaanse (1946) regularly teaches courses on the language. He also gives speed courses on the literature, language and culture of Groningen. He regularly gives lectures and presentations on the language as well (Webloug, n.d.).

Dr. C. (Christopher) Bergmann (1986) is a linguist who recently obtained his PhD at the University of Groningen. During his stay in Groningen, he became interested in Gronings and decided to take on a language course, coincidentally under the guidance of one of the other language professionals (H. Schoonhoven). The language course gave him insights in the language, his interest in the characteristics of the language was piqued and he decided to study it more in depth. Together with the teacher from his language course he decided to write a book, which resulted in Hogelandster waarkwoorden (Webloug, n.d.). Since Dr. C. Bergmann is originally from Germany and learned to speak Gronings only recently, he could be classified as a new speaker.

W.J. (Willem) Friedrich (1941) is an author and photographer, with a specific interest for the region Oldambt in the eastern part of the province of Groningen. For 30 years he worked at the cultural centre De Klinker in Winschoten, where he came in contact with a lot

‘Correct’ Gronings and How to Measure This Value