A Change in the Use of Second-Person Pronouns

(1)

Formal and informal address

A Change in the Use of Second-Person Pronouns

Explained through Dynamic Systems Theory

(2)

Formal and informal address

A Change in the Use of Second-Person Pronouns Explained through Dynamic Systems Theory

Zwolle, Eric 1823310

Master Thesis Communicatie- en Informatiewetenschappen Rijksuniversiteit Groningen

Marjolijn Verspoor 27-04-2011

(3)

Foreword

This master thesis was written

as

the final assignment of the master study Communicatie- en Informatiewetenschappen at the Rijksuniversiteit Groningen. The intent of this is to investigate the changing custom of second-person pronouns in the Dutch language using Dynamic Systems Theory.

I would like to take the opportunity to express my gratitude to a number of people who helped me during the development of this thesis.

I would like to thank:

My supervisor, Marjolein Verspoor for her inspiration and guidance during the development of this thesis,

my friends of RUG library who helped me relax during numerous coffee- or lunch breaks,

the Zwiers family for allowing me to stay at their home while I would conduct my research in Dalen,

and my girlfriend Saskia and my sister Laura for providing feedback on my thesis.

Most importantly, I would like to thank my parents for supporting me and providing me with the means in order to study.

Eric Zwolle Eric

Groningen, 22-04-2011

(4)

Master thesis Eric Zwolle 3 Abstract

Language change has been extensively studied in the field of sociolinguistics. Many studies have investigated phonological, semantic or syntactic changes using Dynamic Systems Theory. So far, no studies have applied this model to study the changes in discourse-pragmatic phenomena. The use of second-person pronouns in Dutch has been studied and concluded that the linguistic variable is changing over time. This study also investigates the change of the second-person pronouns in Dutch. However, this study distinguishes itself by applying Dynamic Systems Theory in order to study the linguistic variable. Thereby the aim of this study is to evaluate the suitability of this model to investigate discourse-pragmatic phenomena.

Dynamic Systems Theory is a recent approach in the field of sociolinguistics to study language change. Essentially, it regards languages as dynamic systems that continuously change over time. The development of such systems can be understood as nested processes that evolve across any timescale. A change is occurring when a system moves from one attractor state to another. Dynamic systems are characteristically sensitive to initial conditions, are completely interconnected, evolve through internal reorganization and interaction with the external environment and their change is characterized by chaotic variation between attractor states.

Sociolinguists have different ways of studying language change. A language change is either studied in real time, or in apparent time. Apparent time studies investigate cross-sectional data from different age groups. This study adopts the apparent time construct, as a real time construct is not feasible.

In order to evaluate the suitability of Dynamic Systems Theory to study discourse- pragmatic phenomenon, the following research question has been formulated: can Dynamic Systems Theory be used to discern a change in the custom of using U and je in Dutch across different age groups?

A case study has been devised in order to answer the research question. The case study is based on a questionnaire. The questionnaire consists of a collection of photos showing various people of different ages, gender and situations. The aim was to collect data on participants of different age groups addressing the models with U or je.

(5)

The research population consists of the population of Dalen, a town in the northeast of the Netherlands. A sample totaling to 148 participants took part in this research.

The participants were men and women between the ages 20 and 80 years old. The participants were divided into three age groups: between 20 to 40, between 41 to 60 and between 61 to 80.

The data on these age groups has been analyzed with Dynamic Systems Theory, where the focus is placed on variation of responses between the age groups.

Namely, Dynamic Systems Theory predicts that a change in the linguistic variable is signified by variation in the second age group ranging between 41 to 60 years old.

Furthermore, the data is analyzed according to the age grading hypothesis. This is a model which predicts that the use of the linguistic variable changes as people become older. Also the age grading hypothesis predicts a specific pattern in the data.

Based on the previous concepts have the following hypotheses been formulated in order to answer the research question:

! Null-hypothesis (H0): The data shows no clear variation patterns between the age groups.

! Hypothesis 1 (H1): The data shows a clear pattern of variation occurring at the second age group.

! Hypothesis 2 (H2): The data shows a pattern consistent with age grading.

The results of this study showed that for certain contexts, the custom of using U and je is changing. These results have been observed using Dynamic Systems Theory.

Thus hypothesis H1 has been adopted. Since Dynamic Systems Theory has successfully indicated a change in the linguistic variable, the model presents a practical method of investigating discourse-pragmatic phenomena.

(6)

Master thesis Eric Zwolle 5 Table of contents

1. Introduction ...6

2. Theoretical framework ...7

2.1 Using U or je ...7

2.2 Language change...10

2.3 The wave model...12

2.4 Transmission...14

2.5 Incrementation ...15

3. Applied models ...22

3.1 Dynamic Systems Theory ...22

3.2 The apparent-time construct ...26

4. The case study ...29

4.1 Research question...29

4.2 Hypotheses...30

4.3 Research method ...30

4.4 Participants ...30

4.5 Procedure...31

4.6 Materials...31

4.7 Location...33

4.8 Variables ...33

4.9 Prediction ...34

5. Results...36

5.1 Distribution of age...36

5.2 Distribution of gender...37

5.3 Aggregate data ...37

5.4 Data per gender group...39

5.5 Data per situation...42

5.6 Partitioned data per age group ...44

6. Discussion ...61

7. Conclusion ...67

References ...68

Appendix...73

(7)

1. Introduction

Languages change. Linguistic change is a subject that has been extensively studied in the field of sociolinguistics. A great deal of such studies has been conducted to investigate changes in phonologic, semantic or syntactic changes. As a result, several models of linguistic change have been produced. A relatively recent approach to the study of linguistic change comes from Dynamic Systems Theory. Dynamic Systems Theory will henceforth be referred to as DST. This theory characterizes languages as dynamic systems that continually evolve and change over time. Whereas this model has been applied to study phonologic, semantic or syntactic elements of a language, no previous study has used it to investigate discourse-pragmatic phenomena.

Discourse-pragmatic phenomena describe literal and non-literal aspects of communicated linguistic meaning that are determined by principles that refer to the physical or social context. However, discourse-pragmatic phenomena are equally prone to change as phonologic, semantic or syntactic changes.

Therefore, DST should be applicable to study such linguistic changes.

This paper is essentially a case study that investigates the custom of second- person pronouns in Dutch over time. However, this study distinguishes itself by applying Dynamic Systems Theory in order to study the linguistic variable.

In doing so, this study investigates the propensity for DST to describe the change of the linguistic variable. Furthermore, researching this custom is largely a discourse-pragmatic endeavor. Therefore this study will serve as an evaluation for the application of DST to study discourse-pragmatic phenomena.

Previous studies observed that the use of second-person pronouns is

changing in the Netherlands. Vermaas (2002) observed a change in the use

of second-person pronouns in Dutch from the thirteenth to twentieth century in

the Netherlands. Therefore, the change of this linguistic variable is a given

fact. Thereby the research by Vermaas provides adequate comparison

material for this study. As the discourse-pragmatic change is already known

this study can compare these observations in order to test DST.

(8)

Master thesis Eric Zwolle 7

2. Theoretical framework

The following chapter describes various important theories and models from the field of sociolinguistics. They serve as important background information for the conceptualization of this paper as well as providing inspiration for its methodology.

2.1 Using U or je

Preparatory to the case study, it is important to understand the subject. As a discourse-pragmatic variable, it is a more complicated phenomenon to grasp than, for instance, prescriptive aspects of language such as semantics or phonology. Therefore the following section will explain various aspects to using U and je.

A tu and Vous (T/V) distinction is used in sociolinguistics to describe the use

of second-person pronouns in a language. In Dutch ʻUʼ and ʻjeʼ are second-

person pronouns used for formal or informal address where the V form is U

and the T form is je. Using U and je is characterized through various

dimensions such as social distance, courtesy, familiarity, solidarity, formality

or insult between speakers. Therefore its application depends more on

contexts and frames than on grammar or lexicon. The T/V distinction also

relies on the face-keeping principles by Goffman (1955). They characterize

linguistic interactions as projections of wants and needs that establishes

public image. By these principles speakers present themselves as competent

members of society and ensures security of oneʼs own and oneʼs other face

(Houtkoop; Koole, 2000). In such cases the participants usually scramble to

correct a mistake or indicate competence and so maintain the social balance

between the interlocutors. Brown and Levinson (1987) expand on this theory

by identifying speech acts as potential Face Threatening Acts (FTA). With this

theory, Brown and Levinson offer two facets to the concept of face, namely

positive and negative face. Positive face reflects the desire for oneʼs self

image being appreciated and respected. Negative face on the other hand is

characterized as the desire for privacy, autonomy and the basic claim for

freedom to act. During communication, being verbal or non-verbal, these

(9)

ʻfacesʼ can either be threatened or reinforced. During a conversation a speaker may unintentionally threat oneʼs face. According to Brown and Levinson, the speaker can adopt various strategies to limit the threat of both the positive and negative face.

Furthermore, face-work conducts relationship negotiation between the communicators. When engaging in conversation, “Speakers are necessarily doing a certain amount of ʻidentity workʼ, through their use of conversational style as well as their use of a particular accent, dialect or language”

(Warhaugh, 1998). They often display unequal relationships as one or more speakers enjoy a more powerful position over others. Rajend et al. (2009) refers to this as symmetrical or asymmetrical talk. While a symmetrical relationship indicates equality in power between participants of a conversation an asymmetrical relationship indicates a difference. Such differences may be explained in status-related terms and will affect the overall organization of an interaction (Rajend, et al, 2009). Speakers can indicate their relative status through the use of second-person pronouns. The V form negates a higher status of the addressee than the speaker. It is used to indicate politeness and to convey respect towards the other. Moreover, the speaker would be positively polite towards the listener by diffusing any positive face FTAʼs. On the other hand, T is used to indicate similarity or familiarity between people.

Using T creates camaraderie as it negates social distance from one-another.

The V form however creates social distance. Thus using T/V during a speech act can be an arduous matter for those untrained in its use.

The theoretical framework by Brown and Levinson provides an understanding of the constituents of the T/V custom. However, the framework does not provide a specific description of Dutch customs. Rather, it is a theoretical framework describing a universal theory of politeness applicable to many languages including Dutch.

To describe the custom of T/V in Dutch is challenging at best. As a discourse-

pragmatic phenomenon it relies on frames and politeness theory. Therefore

there are countless possibilities and combinations of situations that influence

(10)

the use of the T/V forms. Also, there are no linguistic references that outline the general rules on this subject. Vermaas (2004) also determined that these rules are omitted in linguistic publications in the last three decades.

Nevertheless Wagenaar (2004) provides a description of how T/V should be used. According to the author: U should be used when addressing someone who is unfamiliar with the person in question, or whether one is engaging in a business environment or buying something at a shop. In cases where the person is familiar, younger or seems to enjoy a similar or lower status, the word ʻjeʼ should be used. Wagenaar describes the general rule as: always say U, unless the usage of je has been agreed upon or has been established in a natural manner (2004). Furthermore Vermaas (2004) provides an illustration of the custom based on the politeness and familiarity dimension. Table 1 makes an attempt to illustrate the general rule for using the T or V form based on publications by Vermaas and Wagenaar.

Table 1: rules of T/V:

Say ʻUʼ when addressing: Say ʻjeʼ when:

Someone older Someone is younger

In formal settings In informal settings Someone has a higher socio-

economic status

Someone has a lower socio- economic status

Strangers Someone familiar

Family Friends

Vermaas (2004) has investigated various situations where T/V are used. Her

research especially describes the use of address by dimensions of status and

solidarity. According to the author, U is a feature of status whereas je is a

feature of solidarity. Vermaas observes that the dimension of solidarity is

starting to prevail over the status dimension. The author attributes this to a

shift in the societal developments of authority relations and emancipation

during the 1960ʼs. Vermaas concludes that the use of je is increasingly

becoming prevalent over U.

(11)

2.2 Language change

As previously mentioned, languages are prone to change. Discourse- pragmatic elements can change as well as semantic or phonologic ones. The following section explains several sociolinguistic views on the subject.

Sociolinguists developed different approaches towards studying languages.

On the one hand, synchronic linguistics is concerned with the constituents that establish a language as a system. Diachronic linguistics on the other hand concerns the replacement of linguistic items within a system. In other words, diachronic linguistics focuses on changes within a language over time.

According to diachronic linguistics, languages do not exist as static constituents (Appel et al, 2002). Instead, languages are dynamic and can be proven to change dramatically over a given period of time by specimens of ʻthe same languageʼ such as English (Rajend, et al, 2009). De Saussure (1983) describes languages as adaptable states-of-being. Such states do not exist as a point of time but rather as: “a period of time of varying length, during which the sum of all changes occurring is minimal”. (De Saussure, 1983: 142) This conception is vital to our understanding of languages as an object of study. The language-state studies describe languages as dynamic entities and predict that any language changes gradually across time as a never- ending sequence of states. Per definition, this model predicts that languages change over time.

An important question often reviewed in the field of sociolinguistics is whether change is observable while it is occurring. Divergent sociolinguistic paradigms provide contrasting answers on the matter. For instance, some of the founders of the modern discipline, e.g. De Saussure and Bloomfield, have maintained that change in itself cannot be observed and that only the results and consequences of a given change may reveal itself (Wardhaugh, 1998).

De Saussure, Bloomflield as well as Hockett, being structuralist linguists

adopting the comparative method, traditionally observe linguistic change at

two or more distinct points in time. By doing so, they consider the observation

of changes in progress theoretically impossible (Chambers, et al, 2002).

(12)

Variation further complicates studying linguistic change. Not only dialects, but also jargon and slang account for great linguistic variation. The kinds of variation this produces raise significant problems about the very nature of language itself. It does not only complicate the specification of parameters that characterizes a language, but also make it difficult to attribute variation to linguistic change. Equally is it difficult to determine if variation signifies linguistic change, or if it simply signifies naturally occurring variation in a language.

Despite the issues involved, many sociolinguistics study linguistic changes through variation. The wave model developed by Bailey and Bickerton is a dynamic model that suggests that all variation in language results from changes in progress, which means that variation is therefore the mark of progress (Wardhaugh, 1998). Also, the models of transmission and incrementation by Labov (2001) predict how variation leads to linguistic change. Moreover, the study conducted by Chambers and Trudgill (1998) describes how the uvular ʻrʼ spreads from city to city in several European countries and later on spreading across the countryside. They have reported on what they believe is change in progress. Additionally, studies by Chambers et al (2009) led them to conclude that linguistic changes are characterized by periods of relative stability followed by periods of considerable flux. Appel et al, (2002) reinforce this by describing variation being the ʻmotorʼ behind linguistic change.

Thus there is a clear division in the field of sociolinguists regarding studying

language change through variation. While some sociolinguist claim that

linguistic change cannot be studied through variation, others provide evidence

proving the contrary. This study adopts the latter by using the Dynamic

Systems Theory model. The following section will clarify the ʻwave modelʼ in

the field of sociolinguistics and provides an explanation of how language

changes over time.

(13)

2.3 The wave model

Prior to the introduction of the ʻwave modelʼ, diachronic linguistics was traditionally characterized by the ʻcomparative methodʼ. The comparative method essentially reconstructs languages according to various stages of all language families across the globe, i.e. it describes a language in succeeding stages and exploits differences and similarities between related languages.

This method relies heavily on the ʻtree modelʼ for visualizing the output of its results. The tree model is an analogy of a family tree, where the branches represent different language families. August Schleicher developed the tree model, in his words the ʻstammbaumʼ model, in 1853 (Cited in Fox, 1995).

However, the tree model has its shortcomings. Fox discusses these shortcomings as: “not a very satisfactory model for the representation of the historical relationships between languages” (Fox, 1995: 128). His main argument is that the model fails to accommodate the external influences that change a language, and that the model implicitly assumes that changes occur independently in each branch.

The ʻwave modelʼ or ʻwave theoryʼ is a theoretical construct devised by Johannes Smidt (Cited in Fox, 1995). An early version of the wave model depicts linguistic developments radiating out from a point, like a stone thrown in the water. This model primarily describes the interaction of changes as the waves traverse languages and dialects. It was meant to deal with the relationships between languages and how new languages were formed from pre-existing languages. Furthermore, unlike the ʻstammbaumʼ, describes this model a language as fully interconnected.

Figure 1: Smidtʼs wave theory (from Fox, 1995: 128)

(14)

The wave theory was adopted and further developed as a general model by Bailey (1973) and Bickerton (1971) to describe linguistic changes over time.

Essentially, a language change is not an instantaneous event. As predicted in the language-state conception of De Saussure, a language changes in a gradual manner. Many studies concerning language change depict the phenomenon of change essentially as a wave. Recurrent findings in language change studies observed that initial alterations start off slowly, incrementally rapidly changing at a mid-section and flatten at a completion state, where the frequencies are cumulative (Altmann et al, 1983, Bailey 1973, Kroch 1989, Labov, 1994; cited in Tagliamonte et al, 2009).

Figure 2: The S-curve (from Tagliamonte et al, 2009: 58)

An example of the application of the wave model is the theory of lexical diffusion. Especially, Wang (1977) has worked out the theory to describe how a language can change word by word across a progressive gradient. The theory primarily describes phonological changes in a language and depicts phonological changes as an s-curve. The s-curve of figure 1 therefore acts as a visual example to describe wave theory. In the words of Wardhaugh: “The theory of lexical diffusion has resemblances to the wave of language change:

a wave is also a diffusion process” (Wardhaugh, 1998: 211).

(15)

As the s-curve is applied as a general linguistics change model, caution is needed when describing discourse-pragmatic changes through the s-curve.

After all, the s-curve of lexical diffusion explains phonological changes and therefore is not necessarily compatible for discourse-pragmatic phenomena.

However, the transmission and incrementation models by Labov (2001) formulate an adequate model for discourse-pragmatic changes. Whereas these models were developed for depicting phonological phenomena Tagliamonte (2009) has observed the suitability of Labovʼs models in describing discourse-pragmatic changes. The next section describes Labovʼs transmission and incrementation models of language change.

2.4 Transmission

Transmission is a term that describes a linguistic change process that involves the language acquisition of children. It describes a continuous sequence of childrenʼs acquisition of their native language. Labov provides a definition describing the social process of linguistic acquisition: “A language (or dialect) Y at a given time is said to be descended from language (or dialect) X of an earlier time if and only if X developed into Y by an unbroken sequence of instances of native-language acquisition by children.” (Labov, 2007: 3)

In its simplest form, transmission is a flawless reproduction of an invariant

linguistic pattern of the parents performed by the child. The child will obtain

the language with the acquisition devices it possesses given that parents

continually speak the language as they learned it. The term transmission

therefore presumes that children reproduce the form of language used by

older generations perfectly. However, there is a problem with the transmission

model. Namely, children do not replicate a language perfectly. This lies at the

core of the transmission problem. The transmission problem states that

children learn to talk differently from their previous generations and that these

differences move in the same direction in each succeeding generation (Labov,

2001). Thus linguistic acquisition of children results in imperfect duplication of

(16)

linguistic usage in regard to previous generations. Each generation in its turn will make differences in the same direction.

2.5 Incrementation

The linguistic changes that result from the transmission problem process is referred to as incrementation. This process can be distinguished as a ʻchange from belowʼ. According to Labov (2007), childrenʼs incrementation of the linguistic change will take the form of increases in frequency, extent, scope or specificity of a variable. He refers to this process as ʻvernacular re- organizationʼ. Additionally, Labov states that age is a crucial variable of vernacular re-organization, as it takes place between the first acquisition and the effective stabilization of language use. Thus vernacular re-organization is deemed to occur in a certain age-period. The beginning hereof has been examined in some detail by the works of Payne (1980), Kerswill (1994, 1995, 1996) and in Kersill and Williams (2000), where mounting evidence concludes that vernacular shifts begin to take place after the age of four years (Cited in:

Tagliamonte, 2009). Stabilization is a term describing speakers maintaining a constant level of language use after acquisition. However, at what age stabilization takes place has not been determined by empirical research.

According to Tagliamonte (2009), it is known that certain types of changes are easily integrated into the vernaculars of speakers of any age, but that there is a limit to what is learned in terms of phonological, morphological and syntactic components beyond a certain age. Nonetheless, it is unknown at what age this critical thresh-hold occurs.

Labov (2001) describes several hypothetical models in order to illustrate

linguistic change created by incrementation. The first model describes the

effects of incrementation on the language use of an individual.

(17)

Figure 3: A linear model of incrementation for a single female speaker from 1 to 45 years of age (from Labov, 2001)

Figure 2 is a hypothetical model that portrays the value of change for a single female speaker year by year. The first acquisition is set between 1 to 3 years of age. It levels exactly with that of the caretaker, level 0 in this graph. The change occurs between ages 4 and 17, with incrementing change occurring at the same rate for 13 years. The incrementation stops and vernaculars stabilize.

Uniform incrementation

Uniform incrementation describes the effect of incrementation on succeeding generations. Figure 2 supposes the incrementation trajectory of a speech community, composed by women. When a woman becomes a mother herself,

“this vernacular is passed on to the next generation of speakers, who undergo the same process of reorganization, incrementation and stabilization”

(Tagliamonte, 2009: 66). Figure 4 is a visual rendition of the uniform

incrementation process.

(18)

Figure 4: Age profiles of a linguistic change in progress with uniform incrementation (from Labov, 2001: 449)

The pattern of this model resembles a step-wise trajectory. It shows how a change instigated by a given generation compounds in consecutive generations. A generation set at 25 years in this model. Linguistic change is portrayed by one-year incrementations, over a time frame of 100 years.

Again, the hypothetical test-subjects are women. It is important to note that the figure does not track an individual. Rather it portrays a snapshot of all the women in a hypothetical speech community. Also, Labov used arbitrary units in his model. It acts purely as an example and as a visualization of the concept and thus should not be interpreted literally.

The first speech community is tracked at the year 1925. As the women reach age 5, they reach the first level of incrementation at 1 years old.

Subsequently, the trajectory follows an upward slope until the age of 17, where the vernaculars stabilize at level 13. This level is maintained by all speakers aged 17 to 25, born between 1900 and 1908. Speakers born before these years are unaffected by vernacular change and thus remain at base line 0. This trajectory is repeated for the years 1950, 1975 and 2000.

According to Tagliamonte (2009), the uniform pattern of incrementation has

been confirmed by empirical research. However, its generality is limited.

(19)

Moreover, Labov (2001) observes that when uniform incrementation occurs, it appears to be restricted to males. He attributes the advancement of linguistic change to the generational influence of female caretakers. This means that men inherit the increment of their caretakers but do not create advancements as women.

Logistic incrementation

Labov substituted another more plausible model of language change due to the lack of generality of the uniform pattern. In addition, Labov acknowledges coinciding studies that describe language change as an s-curve. Accordingly, he produced a model of logistic incrementation.

A logistic expression known as the Verhulst logistic growth curve lays at the foundation of the logistic incrementation construct (see Tsoularis, 2001).

Labov adopted the following expression to predict a hypothetical logistic growth curve:

Logistic growth expression (from Labov, 2001: 450)

I = K1/(1 + K2 / N⁰ x e-^rt)

In the equation is K

1

the maximum possible change in 1 year, K

2

the limits of change, N

0

is the starting point, r is the rate of change and t the time in years.

A distribution curve similar to the s-curve in figure 1 is predicted by this expression. When N

0

is set at 1, and K

1

and K

2

are set at 100, it predicts a hypothetical 100-year progress of a change with logistic incrementation. The result is the prototypical wave model of linguistic change.

However, predicting the rate of incrementation produces a different curve.

Figure 5 shows the increment of each year over the preceding year.

(20)

Figure 5: Rise and fall of rate of incrementation of a logistics curve (from Labov, 2001: 451)

Finally, Labov combined the model of logistic incrementation with transmission. The resulting values of Labovʼs calculations are reproduced in figure 6.

Figure 6: Age profiles of a linguistic change in progress with logistic incrementation of change (from Labov, 2001: 453)

(21)

Integrating transmission with logistic incrementation has created substantially different values from the step-wise trajectory predicted in the uniform incrementation model. Instead, the results are linear curves that are typical of trajectories produced in sociolinguistic research of female speakers (Labov, 2001). These curves do not maintain an upward curve as age decreases, but instead are interrupted by peaks. Also, each successive level produces higher levels of change across the linguistic community. On that account logistic incrementation predicts that children acquire a larger increment than children of the previous generations. Thus, during vernacular re-organization these children will increase more units of change compared to change induced by earlier speakers. It therefore accounts for maintained advancements of change, and explains why the slopes have an upward curvature from right to left. The drop-off after the peak, towards the younger children, is according to Tagliamonte (2009) attributable to greater length in the participation of these children in the incrementing change process. In figure 6, the vernacular re- organization stabilizes at around the age of 17. That means that at this age, these speakers have participated longer in the change than their younger peers. Specifically, 17 year olds had 12 years of incrementation, 12 year olds had 7 years, 8 year olds had 3 years etc. Therefore the 17 year olds accumulated more years of incrementation than the younger children. While the children younger than 17 have not yet reached the critical age of stabilization, they still continue to participate in the change process.

Eventually it intensifies the sum of their increments the linguistic development of their native language.

It is important to remember that the model is a rendition of arbitrary values produced by mathematical calculations. However, Labov (2001) reviewed several studies, such as studies in Panama City by Cedergren and the Philadelphia Neighborhood Study, in order to test the incrementation model.

He concludes that the model has shown its legitimacy. Also studies conducted

by Tagliamonte (2009) investigating various linguistic variables in Toronto,

has tested and validated the logistic incrementation model. Tagliamonteʼs

study has concluded that the logistic incrementation model is not only

(22)

applicable to describe morphosyntatic- and semantic changes, but also to

describe discourse-pragmatic changes. The logistic incrementation model of

Labov therefore provides a crucial insight in understanding the evolution of

language. Following Tagliamonteʼs conclusion it now does not only describe

language changes in morphologic, semantic or syntactic terms, but also to

depict discourse-pragmatic changes. Moreover, Tagliamonteʼs studies verify

that discourse-pragmatic variables can be studied through sociolinguistic

methods. The ramifications hereof are vital for the formation of this research.

(23)

3. Applied models

The aforementioned theoretical concepts clarify the background of this case study. The transmission and incrementation models of Labov provide insight in the forces behind linguistic change. The next section on the other hand will describe the models used in this case study to investigate the linguistic variable.

3.1 Dynamic Systems Theory

A relatively recent approach to studying linguistic change comes from DST. A dynamic system can be defined as: “a set of variables that mutually affect

each otherʼs changes over time”. (van Geert, 1994: cited in Köpke et al, 2007:

58) DST was originally developed as a mathematical theory describing the effects of various variables upon a system, where sets of variables interact with one another over a given period of time. DST is basically a sociolinguistic adaptation of general Systems Theory. DST is characterized by two themes:

“Development can only be understood as the multiple, mutual, and continuous

interaction of all the levels of the developing system, from the molecular to the cultural; Development can only be understood as nested processes that unfold over many timescales from milliseconds to years.” (Thelen, 1996: 258).

DST can be applied to many different disciplines. A paper published by de Bot et al (2007) links DST to second language acquisition. The authors argue that languages can be seen as dynamic systems to which DST can be applied as a framework and instrumentation for understanding languages. Moreover, a number of other publications have emerged describing language as dynamic systems (Larsen-freeman, 1997; Herdina and Jessner, 2002).

DST ascribes various characteristics to a dynamic system. First of all, it is

sensitive to initial conditions. This means that in some systems small

differences in initial condition can have a profound impact in later stages of its

development. On the other hand, big differences in initial conditions can

sometimes only have a small effect in later stages of its development. This

also indicates a non-linear relationship between the initial conditions and the

long-term effects.

(24)

Secondly, DST regards systems as completely interconnected. This means that large systems consist of many subsystems. As all systems are connected, a change in one system can lead to the change of others. The system is also not only dependent on internal factors; external forces also influence the state of a system.

Therefore, a third important characteristic is the change and development through internal organization and interaction of the environment. Internal reorganization occurs when the system progresses from one stable state to another ʻattractor stateʼ. Attractor states are preferred states wherein the system tries to settle. Such attractor states are synonymous to De Saussureʼs linguistic states. A rendition of how the attractor-state analogy works comes from the work of C. H. Waddinton (cited in Thelen, 1996). The following illustration depicts the attractor state as an egg on top of a slope.

Figure 7: Waddingtonʼs slope analogy (from in Thelen, 1996: 261)

Waddingtonʼs analogy shows the path of an egg on an epigenetic landscape.

The egg can roll in different directions off the slope and possibly end up at

four different locations. The rolling of the egg across the plane represents a

dynamic system moving across time, where four different outcomes are

possible. Yet, the outcome of the development is unknown beforehand. The

top of the slope represents the initial conditions and the bottom stands for the

(25)

attractor states. To reach the ensuing attractor state the egg must traverse a space between attractor states.

To what attractor states the system will develop itself is difficult to predict as often many attractor states are possible. Additionally, sometimes certain input can lead to the system simply adapting itself into the new element without internal reorganization. It is also possible for internal organization to occur without external influence. Consequentially it is difficult to predict for dynamically developing systems.

Waddingtonʼs analogy serves as an appropriate illustration of DST.

Nevertheless, it is not a factual description of reality. This description attributes the path the egg follows to a random occurrence. Nonetheless, linguistic variables are not conditional to random chance. The evolution of a language is rather determined by innumerable human factors. Moreover, linguistic variables do not travel across a predetermined course, but rather seem to create its ʻpathʼ as it goes along. In essence, Chaos Theory best describes such an analogical path where continuing changes alter the state of the system. A good example of this is the weather system. Its state is constantly changing due to small and big changes in diverse variables.

Nevertheless this figure assists in understanding linguistic attractor states by visualizing the concept.

The last characteristic of a dynamic system is also the most important one.

Namely, DST predicts that systems seemingly show chaotic variation over time. Due to the potential complex interaction between variables, a system can portray similar attractor states. As this occurs, extensive variation in the system comes across as chaos. Thereby linguistic DST predicts that chaotic variation in any dynamic system is an antecedent to change.

A study conducted by Partha Niyogi and Robert Berwick (1995) illustrate the

change in the use of S(subject) V(erb) and O(bject) with no verb-second

(SVO-V2) and V(erb) O(bject) S(ubject) + verb second constructions in

French by applying the Dynamic Systems Theory. The model has predicted a

(26)

sudden change over a course of three-four generations, resulting in the following graph.

Figure 8: Change according to DST (from Niyogi and Berwick, 1995: 5)

The model depicts the shift in syntax use as an X-shaped graph. The linguistic change of SVO-V2 moves from a high frequency of use towards a lower one.

The VOS+V2 curve moves from a low frequency to a high one. They intersect at a certain point where the frequencies of the linguistic variable are equal.

According to DST, this intersection of frequencies signifies chaotic variation in a system, which is an indication of a system in a transition of moving from one attractor-state to another.

The graphic representation of the results observed by Niyogi and Berwick serves as an important reference for this research. The model illustrates how results of language change can be observed through DST. Therefore, this model will be applied as a reference to compare results observed in this study.

In retrospect, the models by Labov are essentially a derivation of Dynamic

Systems Theory as it shares certain characteristics. The most important

similarity between the two models is that they both describe the process of

(27)

change as the result of a dynamically changing system. However, the model of incrementation describes the linguistic process of change for individuals and explains the forces of linguistic development for women and their daughters. The result is an incrementally changing language. On the other hand, DST considers all the forces of change that influence linguistic development across a community of speakers. Using U and je is dependent on many factors and its change possibly as well. Thus DST will provide a wider scope to investigate the linguistic variable. Furthermore, DST explains circumstances through which language change can be observed over time.

Namely, DST describes that chaotic variation can be a symptom of linguistic change. Labovʼs models on the other hand do not specify identifiers through which language change can be observed. DST is therefore a more suitable model for this study, as it will assist in identifying a linguistic change.

Whereas DST explains a method of analysis, explains the apparent-time construct provides a method of data-collection. This methodological model will be explained in the following section.

3.2 The apparent-time construct

We learned earlier on that language change is traditionally studied by

researching variation at two distinct points in time. Such studies require a

great deal of time and resources to conclude. A longitudinal study of language

change usually has a span across several decades. The cost of such a study

is immense. A more viable method is to investigate language changes though

the apparent-time construct. Apparent-time surveys permit database

collection in identical settings and circumstances, thereby eliminating the

comparability problems. Research conducted by Labov on sound change in

New York was based on apparent-time methods (Labov, 1966; cited in Labov,

1994). Apparent-time studies overcome the problems of longitudinal studies

by intensively studying a community for a short period of time to examine if

any deviations occur. The population is thereby divided into age groups. In the

words of Rajend: “Where older age groups show low use of a variant while

younger groups show increasingly greater use, we can assume that there is a

(28)

change going on in ʻreal time” (Rajend, 2009: 120). Chambers et al (2002) reinforce this by determining age as the primary social correlate once a language change has been observed. By any given alterations in language the age variable creates a prototypical pattern across generations. For instance, some minor variation in speech of the oldest generation occurs more frequently in younger generations. Increasing frequency down the age scale will mark such a language change. The pattern described by the apparent-time construct is similar to the logistic incrementation model of Labov. Therefore, age acts as a substitute for linguistic changes observed in real time. Additionally, this construct assumes that individual vernaculars remain stable after adulthood similarly to the logistic incrementation model.

However, Chambers et al (2002) asserts that this assumption has not been tested in full, and that therefore one must take caution with this method. One of the arguments disputing apparent-time is the age-grading hypothesis.

While the apparent-time construct assumes that age is the metonym of real- time, the age-grading hypothesis states that the variation in language use across generations in fact be attributed to a different application of language as one gets older. The age-grading hypothesis therefore challenges any apparent-time constructs as it rebuts the validity of such studies. Namely, the measured linguistic variation is an indicator of linguistic change or evidence of the age-grading hypothesis. Constructing research on language change can therefore be arduous since within the field there seems to be controversy about what is being studied.

Despite the age-grading hypothesis, numerous studies have validated the apparent-time construct. Research conducted by William Labov (1963, 1966, cited in Guy Bailey; and Labov 2001) produced new methodological innovations in studying linguistic changes in progress despite such difficulties.

Legitimacy of apparent-time has been observed in studies stratifying the

famous ʻMarthaʼs Vineyardʼ investigation. Onsets of using (ay) or (aw) on

Mathaʼs Vineyard in apparent-time have been mirrored by increase of these

features in real-time. (Labov, 1963; cited in Guy Bailey, ) Furthermore, Labov

(29)

revisited his own study on sound change in New York in 1962 to confirm his

apparent-time findings in real-time. Also Joy Fowler replicated the New York

experiment in 1986. With a similar methodology she discovered a similar

pattern as Labov did in 1962. For both in 1962 and 1986 respondents in

different department stores used differing forms of (r). The 1986 figures show

somewhat higher usage of one form of (r-1) over the other than the 1962

figures. Over 23 years, the rate of r-pronunciation had increased an average

of 7 percent (Labov, 2006) Thus Labov concluded that change has taken

place in real time and that the apparent-time construct has been ratified.

(30)

4. The case study

This chapter will describe the research question and the methodology of this case study. The research method for this investigation finds its inspiration from the previously described models and concepts. Linguistic change models by Labov and Dynamic systems Theory describe methods to investigate phonological or semantic variables. The models of Labov illustrates why and how linguistic change occurs. DST on the other hand describes the characteristics of linguistic variation when a language shifts from one attractor-state to another. Research by Tagliamonte observes the suitability of Labovʼs model to describe changes in discourse-pragmatic phenomena. The question therefore arises: can a discourse-pragmatic change also be observed through DST? Thus the primary concern of this research is to investigate the change of the U and je custom through DST. A combination of DST and the apparent-time construct is employed in order to achieve this.

4.1 Research question

The research question can be formulated as follows:

Can Dynamic Systems Theory be used to discern a change in the custom of using U and je in Dutch across age groups?

The research question is based on the properties depicted by DST. The

model will be applied as an analytical tool for the data of this research. The

data will initially be compared to the predicted results. Subsequently, the data

will be analyzed through DST. The analysis will be modeled after the

observations of Niyogi and Berwick but will also be scrutinized through

various statistical procedures. Essentially this study is looking for a

prototypical DST pattern as described in previous studies. Therefore, this

study looks for the level of variation between the three age groups. More

specifically, if a change is occurring it is expected to happen at the second

age group.

(31)

4.2 Hypotheses

The following hypotheses are based on the research question:

Null-hypothesis (H0):

The data shows no clear variation patterns between the age groups.

Hypothesis 1 (H1):

The data shows a clear pattern of variation occurring at the second age group.

Hypothesis 2 (H2):

The data shows a pattern consistent with age grading.

The nature of these patterns will be explained in the ʻpredicted resultsʼ section.

4.3 Research method

The case study is based on a sample of the local population of Dalen, in the northeast of the Netherlands. This sample is selected as a representation of the Dutch research population. Data collection is performed by means of a questionnaire. The questionnaire consists of a collection of eight photos showing various people of different ages and in different situations. The aim of this was to collect data on how different age groups address the people represented in the photos of the questionnaire. The respondents were asked to determine whoever they would address this person with U or je.

While the research has the essential quality of a trend study, is the research method cross-sectional. Conducting a trend study is unrealistic due to limited finances and resources. For this reason this research is based on the apparent-time construct.

4.4 Participants

A number of 148 participants have been selected to take part in this study.

The group was composed of men and women between 20 and 87 years old.

They were a homogeneous group of middle-class people from the town of

Dalen.

(32)

Master thesis Eric Zwolle 31 4.5 Procedure

Target group selection was performed in a semi-randomized fashion. Initial data was collected through acquaintances. The remaining data was collected in public spaces such as shopping centers. The respondents were selected on a middle-class socioeconomic status. Doing so created a more homogeneous group of respondents. They were classified in three categories per 20 years, which is the average categorization between generations.

After selecting participants, they were asked to respond with U and je to the pictures of the questionnaire while the researcher marked down the answers on a scorecard. Afterwards, the researcher would also mark down the gender and age of the participant on the scorecard.

The questionnaire created a specific context for the respondents. Namely, the respondents were addressing strangers. Also, the ages of the models vary between 20 and 60. Furthermore, the respondents are asking for directions on side streets or were situated in a professional setting. The researcher regulated the context of each photo in order to minimize the effect of interfering variables.

The selection procedure continued until a total of 148 respondents had answered the questionnaire. Afterwards, the respondents were categorized according to three age groups. Age group one consisted of people between 20 to 40 years old, age group two consisted of people between 41 to 60 years old and age group three consisted of people of 60+ years old.

Subsequently, the data was analyzed through various statistic operations.

These assisted in analyzing the data according to determine if a change is occurring according to DST. The operations and analysis are thoroughly explained in the results section.

4.6 Materials

The materials of this study consist mainly of the questionnaire and a

scorecard to record the responses, age and gender of the participants. The

following table describes the attributes of each photo in the questionnaire.

(33)

Table 2: description questionnaire

Picture

1 2 3 4

Context* A B A B

Gender Female Female Male Male

Age group

20-40 40-60 40-60 40-60

Picture

5 6 7 8

Context* A B A B

Gender Female Female Male Male

Age group

40-60 20-40 20-40 20-40

* Context A: Asking for directions; context B: Professional setting

(34)

Master thesis Eric Zwolle 33 4.7 Location

The case study was conducted in the town of Dalen in order to optimally control the variables.

The town is situated in the North East of the Netherlands. It is a small town in the province of Drenthe. It has a population of 3420 and experiences little migration (Straatinfo, 2011). Often people live here for their entire lifespan. Therefore Dalen provides a stable speech community and

thus an ideal sample of participants for this research. Furthermore, the researcher is acquainted with some of the locals. Through these contacts the researcher established their residents as a sort of ʻbase of operationsʼ.

4.8 Variables

The independent variable in this research is the age of the target group. The response of the participants is the dependent variable. Also two control variables were included. These were gender and the context.

Figure 9: Location Dalen

(35)

Table 3: variables

Independent variables Value Level of

measurement

Age respondents 20 - 40

41 - 60 61+

Ordinal

Age models 20 - 40

41 - 60

Ordinal

Gender models / respondents

Male Female

Nominal

Context models Asking directions

Professional setting

Nominal

Dependent variables Value Level of measurement

Response U

Je

Nominal

4.9 Prediction

Predictions can be made according to theory and by the observations of Wagenaar and Vermaas. It is challenging to predict the behavior of the participants despite the theory. Several variables are at work simultaneously.

An overview per category of variables will be provided in this section. This shows the predicted outcome of individual variables but not the interaction between variables.

4.9.1 Age group

It is expected that the oldest generation will respond mostly with je as they are addressing younger people. The second age group will probably address theyʼre own generation with U. But it might also occur that they respond to their own generation with je since they are of the same age group. But Vermaas (2004) observed that this behavior occurs more with familiarity.

However, since the models are all strangers we expect a slightly higher

tendency towards U. The youngest age group is expected to respond with U

towards the older age group. It is expected that they show more camaraderie

towards their own age group thus address them with je. We can attribute the

(36)

observed results to age-grading when the data matches these predicted outcomes.

Figure 10: Age group predictions to say je

4.9.2 Situation

It is expected that the participants will have a stronger inclination of addressing with U towards people in a professional setting. Still, it is unknown how this variable will interact with the age variable.

Figure 11: Context predictions

4.9.3 Gender

Vermaas (2004) has observed that the middle and youngest generations in

her investigation indicated no significant difference in addressing someone

from the same or opposite sex. Therefore, it is expected that there will be little

difference in how men and women address one another.

(37)

5. Results

This chapter presents the data collected by the questionnaire. The first section provides a rundown of the distribution of age and gender of the participants. These sections verify that the respondents are normally distributed and are thus a representative sample for this research.

The ensuing section will report on the data itself. It is disseminated according to four categories: the aggregated data, gender, context and age group.

5.1 Distribution of age

The participants varied between the ages of 20 to 87. The researcher ensured a normal distribution in the research population in terms of gender and age by pre-selecting participants accordingly.

The respondents were divided into three age categories. The first category consists of respondents between 20 and 40 years old. The second category consists of people between 41 and 60 years old. The third category consists of people between 61 years old and older.

The following table shows the distribution of participants according to age.

Table 4: Age group distribution

Age group 1

20-40

2 41-60

3 60+

Number of cases 49 52 47

The participants are adequately distributed in age with a 10% difference

between the highest and lowest frequency of participants of the age

categories. A chi-square goodness-of-fit test (χ

²)

was performed to calculate a

normal distribution of the age groups. This test maintained a significance level

(α) of 0,05. The results of the calculations show χ

²

= 0,257, with a p-value of

0,880 (χ

²= ,257, p < ,880). Thus there is no significant difference between the

normal distributions of respondents across the age categories.

(38)

5.2 Distribution of gender

Figures by the central bureau of statistics show that the population of the Netherlands is divided in 49% male and 51% female (Statline, 2011). The following table describes the distribution of gender of the respondents.

Table 5: distribution of gender

A chi-square goodness-of-fit test (χ

²)

was performed to calculate a normal distribution of gender. This test maintained a significance level (α) of 0,05.

The results of the calculations show χ

²

= 0,824, with a p-value of 0,364 (χ

²= ,824, p < ,364). Thus there is no significant difference between representation

of gender (male: 45.3%, female: 54,7%) in the sample of this research and previous national statistics (Male: 49%, female: 51%).

As the gender and age groups follow a normal distribution, provides this sample of participants is a good test population for this research.

5.3 Aggregate data

The following table describes the aggregate responses for each of the photos.

The frequencies represent the sum of the answers given for each photo.

Frequency Percent

Male 67 45,3

Female 81 54,7

Total 148 100,0

(39)

Table 6: cumulative frequencies

Photo 1 Photo 5

Frequency Percent Frequency Percent

U 61 41,2 119 80,4

JE 87 58,8 29 19,6

Total 148 100,0 148 100,0

U 112 75,7 96 64,9

JE 36 24,3 52 35,1

Total 148 100,0 148 100,0

U 86 58,1 20 13,5

JE 62 41,9 128 86,5

Total 148 100,0 148 100,0

U 135 91,2 97 65,5

JE 13 8,8 51 34,5

Total 148 100,0 148 100,0

Overall the participants provided more responses with U (61,3%). Response

with je composed 38,7% of the score. The fourth photo gained an

(40)

overwhelming 90,8% response with U. Photo number seven on the other hand was 86,6% of the time addressed with je. Photoʼs two, six and eight showed moderately more responses in U, ranging between 64,1% and 75,7%.

Photoʼs one, three, five and eight however gained unequal results ranging between 41,2% and 80,4% addresses with U.

5.4 Data per gender group

The participants involved both men and women. Also the pictures consisted of male and female models. An examination was made whether gender is an important factor for the linguistic variable. When the gender groups present great variety in their answers, then the variable must be an important factor for the T/V custom. Table thirteen displays the frequencies of U and je per gender group. Table fourteen displays the results as percentages.

Table 7: Frequencies of gender-related response

Gender Gender

Frequencies

Man Woman Total Man Woman Total

U 25 36 61 54 65 119

JE 42 45 87 13 16 29

Photo 1

Total

67 81 148

Photo 5

67 81 148

U 56 56 112 44 52 96

JE 11 25 36 23 29 52

Photo 2

Total

67 81 148

Photo 6

67 81 148

U 37 49 86 11 9 20

JE 30 32 62 56 72 128

Photo 3

Total

67 81 148

Photo 7

67 81 148

U 59 76 135 40 57 97

JE 8 5 13 27 24 51

Photo 4

Total

67 81 148

Photo 8

67 81 148