Social network (sociolinguistics)

In the field of sociolinguistics, social network is a term used to describe the structure of a particular speech community. Social networks are composed of a "web of ties" (Lesley Milroy) between individuals, and the structure of a network will vary depending on types of connections it is made up of. Social network theory (as used by sociolinguists) posits that social networks, and the interactions between members of each network, are a driving force behind language change.

The structure of a social network

Participants

The key participant in a social network is the anchor, or center individual. From this anchor, ties of varying strengths radiate outwards to other people with whom the anchor is directly linked. These people are represented by points. Participants in a network, regardless of their position, can also be referred to as actors or members.

Relationships

There are multiple ways to describe the structure of a social network. Among them are density, member closeness centrality, multiplexity, and orders. These metrics measure the different ways of connecting within of a network, and when used together they provide a complete picture of the structure of a particular network.

A social network is defined as either "loose" or "tight" depending on how connected its members are with each other, as measured by factors like density and multiplexity. This measure of tightness is essential to the study of socially-motivated language change because the tightness of a social network correlates with lack of innovation in the population's speech habits. Conversely, a loose network is more likely to innovate linguistically.

Density

The density of a given social network is found by dividing the number of all existing links between the actors by the number of potential links within the same set of actors. The higher the resulting number, the more dense a network is. Dense networks are most likely to be found in small, stable communities with few external contacts and a high degree of social cohesion. Loose social networks, in contrast, are more liable to develop in larger, unstable communities that have many external contacts and exhibit a relative lack of social cohesion.

Member closeness centrality

Member closeness centrality is the measurement of how close an individual actor is to all the other actors in the community. An actor with high closeness centrality is a central member, and thus has frequent interaction with other members of the network. A central member of a network tends to be under pressure to maintain the norms of that network, while a peripheral member of the network (one with a low closeness centrality score) does not face such pressure. Therefore, central members of a given network are typically not the first members to adopt a linguistic innovation because are socially motivated to speak according to pre-existing norms within the network.

Multiplexity

Multiplexity is the number of separate social connections between any two actors. A single tie between individuals, such as a shared workplace, is a uniplex relationship. A tie between individuals is multiplex when those individuals interact in multiple social contexts. For instance, A is B's boss, and they have no relationship outside of work, so their relationship is uniplex. However, C is both B's coworker and neighbor, so the relationship between B and C is multiplex, since they interact with each other in a variety of social roles.

Orders

Orders are a way of defining the place of a speaker within a social network. Actors are classified into three different zones

Visualization of snowball sampling technique showing two sampling zones. The first-order zone contains 7 individuals (black nodes). The second-order zone contains individuals that have direct contact to individuals in the first-order zone. The circles indicate the boundaries of the zones.

depending on the strength of their connection to a certain actor. The closer an individual's connection to the central member is, the more powerful an individual will be within their network. Social network theories of language change look for correlation between a speaker's order and their use of prestigious or non-prestigious linguistic variants.

First Order Zone

A first order zone is composed of all individuals that are directly linked to any given individual. The first order zone can also be referred to as the “interpersonal environment” or “neighborhood." A first order member of a network is an actor who has a large number of direct connections to the center of the network.

Second Order Zone

A second order zone is a grouping of any individuals who are connected to at least one actor within the first order zone. However, actors in the second order zone are not directly connected to the central member of the network. A second order member has a loose or indirect connection to the network, and may only be connected to a certain network member.

Third Order Zone

A third order zone is made up of newly observed individuals not directly connected to the first order zone. Third order members may be connected to actors in the second order zone, but not the first. They are peripheral members of the network, and are often the actors with the lowest member closeness centrality, since they may not have frequent contact with other members of the network.

Social Networks and Language Change

Fieldwork Practices

Social networks are used as analytical devices that can explain linguistic variation in a more sophisticated way than making reference to large categories such as social class. Instead of focusing on the social characteristics of speakers, social network analysis concentrates on the relationships between speakers, then considers linguistic change in the light of those relationships. In an effort to depart from variationist sociolinguistics, the concept of the social network has been used to examine the links between the strength of network ties and the use of a linguistic variant.

The concept of social network is applicable at both the macro and micro levels. Social networks are at work in communities as large as nation-states or as small as an online dating service. They can also be applied to intimate social groups such as a friendship, family unit, or neighborhood. Because social networks are by their nature highly complex and interconnected, sociolinguists usually only study small networks to make the fieldwork manageable, since even the smallest of networks has a potential for large numbers of connections between actors. In fact, even when studying small networks, sociolinguists use the metrics outlined above to get a broad picture of the shape of the network, rather than mapping it out one connection at a time.

In order to apply any sort of quantitative analysis to the data, Milroy proposes that a network strength scale must be allocated to each speaker. The allocation of a network index score allows the network patterns of individuals to be measured and possible links with linguistic patterns to be tested.

[Write more about fieldwork, specifically network strength scale?]

Social Network Theory

Because social networks investigate the forces that impact individual behavior, rather than simply attributing linguistic difference to social class, a theory of language change based on social networks is able to explain linguistic behavior more deeply than variationist sociolinguistics. The two major findings of social network theory are that dense (highly interconnected) networks are resistant to change, and that most linguistic change is initiated by weak links--people who are not centrally connected to the network in question. Though most sociolinguistics working on social networks agree on these findings, there has been extended debate about which actors in the network are the primary drivers of linguistic change. The results of this debate are two theories, the strong-tie theory, and the weak-tie theory.

[Write more about social networks and language change?]

Strong Tie Theory

The strong tie theory, or agentive theory, has long been thought of in classical sociolinguistic theory as a driver of change, even prior to social network theory. In the context of social network theory, agents are the people who are most connected to others in the network, and whose speech style is often imitated by people within the network. These agents also regulate other people’s language usage inside the group, and therefore ensure the spread of their preferred variant form throughout the network, as group members are more likely to adopt the style of those members in the group who have status. Strong tie networks are believed to be resistant to linguistic innovation and themselves serve as mechanisms that enforce the norms associated with a particular social network. The strong tie theory holds that linguistic changes and norms within a community are spread by those who are most connected to the community. People who have the most connections, or prestige, within the community spread these changes, by using their status as a way to enforce norms, but also engage in dialogues with well connected people in other social networks, creating a value like diffusion of speech norms. Labov's study of Philadelphia speech communities provides an example of this.

Weak Tie Theory

Sociolinguists have more recently begun to focus their studies on “weak links”: networks that are not closely connected by shared socioeconomic status, or individuals who are not closely tied to any group, such as people who move frequently or live in isolated areas. The weak tie theory, first proposed by Milroy and Milroy in their 1983 study of a village in Belfast, language change is seen as being propagated by the people who are second order members of the network. “Weak ties provide people with access to information and resources beyond those available in their own social circle” (Granovetter, 1983). Loosely connected individuals are viewed as less likely to conform to group language practices than integral members and thus bring those members of the group into contact with variations that they might otherwise never have encountered. A weak tie social network theory postulates that linguistic variables are spread by means of weak, uni-dimensional social links between non-central individuals. Therefore it is the case that language change will have the propensity to be faster in larger communities rather than in smaller communities. Support for this theory is found in William Labov's study of "lames' in Harlem, and in Lesley Milroy's 1987 Belfast study.

Computational Modeling

In recent years, computer simulation and modeling have been used to study social networks from a broader perspective. Because previous social network studies were focused on individual connections, they tended to limit the size of the network in order for the researcher to work personally with subjects. With the rise of computer modeling, sociolinguists have been able to study the linguistic behavior of large networks of individuals over long periods of time without the inconveniences associated with

individually working with thousands of subjects.

Advances in computer simulation and modeling technology have been used to study social networks on a larger scale, both with more participants and over a greater span of time. Previous social network studies had to examine individual connections in great detail, and so had to limit the size of the networks involved. Linguists working in the field were also unable to accurately pinpoint the causes of linguistic change because it tends to occur slowly over a long period of time, on a scale beyond the scope of a single research project. With the rise of computer modeling, sociolinguists have been able to study the linguistic behavior of large networks without the huge expenditure of time required to individually work with thousands of subjects long-term.

Linguistic Studies

Support for the Strong-tie Theory

The Jocks and Burnouts Study

This study revealed that people chose to follow those students who had the qualities they desired, namely the toughness exemplified by strong urban kids. This desire for qualities indicates an agent led change, or people imitating the manners . Eckert’s studies in indexicality indicate that agents lead language change by spreading norms through network members. In Eckert’s study of speech norms in Detroit high schools, she notes that suburban youth adopted the speech traits of urban youth (including a dipthongized and lowered [i]). They did so to adopt the traits they found desirable in urban youth. The suburban youth adopted these norms to project an image of toughness, wanting to act like urban youth, who were agents of change within their networks.

The Philadelphia Study

Labov, in his 1986 study of Philadelphia speech communities (used as a precursor to social networks) created along racial lines, shows that the agents of linguistic change were the leaders of the speech communities. People with high levels of prestige in the linguistic led the use of these forms, and enforced them as norms within the community. Members of this network then used the forms normalized within the network outside of the network, and continuous usage led to wide adoption of these speech norms.

The Japanese Schools Study

Takeshi Sibata's 1960 study of elementary school children provides strong support for the view that insiders, or leaders, in a social network facilitate language change. He interviewed several elementary school children, and taught them some invented words he created for the purpose of this study. These words held no actual meaning, they were just created to observe adoption patterns. After teaching the students these words, and telling them to teach the other students these words, he came back a week later to observe the results. A few children, those who were popular, friendly, cheerful, active in class activities, were the main people who spread those words.

The Harlem Study

In Labov's 1966 study of African American Vernacular English in South Harlem, he discovered that it was most often people who had second order linguistic membership in African American social networks who caused changes to be propagated to mainstream communities. These people were “lames”, positioned on the end of the social network, and were people who weren’t held in high regard by the leaders of the speech network. They did, however, have connections to other networks, and were thus able to introduce linguistic variables used by members of the social network they are of which second order members.

Labov’s Harlem study served as the basis of the ‘’Weak Tie Theory’’ proposed by Milroy and Milroy's 1980 Belfast study.

Support for the Weak-tie Theory

The Belfast Studies

Belfast: The Original Study

This Milroy and Milroy study examined vernacular English as it was spoken in inner-city Belfast in the 1970s, in three working class communities in Belfast: those in the Ballymacarrell area, the Hammer area, and the Clonard area. Milroy took part in the life of each community as an acquaintance, or 'friend of a friend', investigating the correlation between the integration of individuals in the community and the way those individuals speak.

Map of central Belfast

A five-point network strength scale used in the Belfast study:

(1) Membership of a high-density, territorially based cluster.

(2) Having substantial ties of kinship in the neighborhood (more than one household, in addition to his own nuclear family).

(3) Working at the same place as at least two others from the same area.

(4) The same place of work as at least two others of the same sex from the area.

(5) Voluntary association with work mates in leisure hours.

• According to Milroy, condition one is an indicator of density, while conditions two to five are indicators of multiplexity.

Each individual studied was given a network strength score based on the person's knowledge of other people in the community, the workplace and at leisure activities to give a score of 1 to 5, with 5 being the highest network ‘strength score.

Each person's use of phonological variables, (ai), (a), (l), (th), (ʌ), (e), that were clearly indexical of the Belfast urban speech community, were then measured. The independent variables for this study were age, sex and area. These linguistic variables made up the dependent variable of the study, and were analyzed in relation to the network structure and background of each individual speaker. Deviation from the regional standard was determined by density and multiplicity of the social networks into which speakers are integrated.

What the Milroy's found was that a high network strength score was correlated with the use of vernacular forms, such that the use of vernacular variants was strongly influenced by the level of integration into a network. Close-knit networks are important for dialect maintenance.

Belfast: Subsequent Study

this study examined the variable [u], and it’s relation to a working class identity and usage. The findings from this study show that people with the weakest tie to this identity use the variable [u], the most, possibly as a way to strengthen their ties. Milroy’s 1987 study explored the usage of a rounded or unrounded realization of [u] in Belfast. An unrounded [u] is generally seen as a trait of the working class. In Ballymarket, one of the villages they surveyed, this unrounded [u] was most often used by young males and females, who don’t necessarily identify as working class members. They hold no strong tie to the working class networks, but use the variables to project an image of working-class toughness. These young people, who often interact with people of other social networks, and thus spread the [u] realization through their social networks, creating large scale change, causing the adoption of unrounded [u] in most of Belfast. These results provide support for the weak tie theory of language change.

The Internet Chat rooms Study

In Berg’s 2006 study of digital social networks as sociolinguistic social networks, he notes the values of social networks as linguistic corpuses and linguistic networks. Although they do not necessarily have explicit first order or second order memberships within the network, there is greater usage of certain variables within certain sub-communities. In this paper specifically, the linguistic variables seen in the #India IRC community, are mostly used by young people within a youth sub-community. These variables aren’t seen across the networks a whole, just in sub-communities. These innovations are coming from certain sub-communities, and remaining in peripheral sub-communities, different from the pattern of change extending out of sub-communities into large scale communities in traditional, non computer mediated social networks in linguistics.

The Facebook Study

This theory shows that people choose to follow dominant language forms in their countires’ regardless of their modes of communication. This study is a step in a new direction for social network theory, examining written communication. In Perez-Sabater’s 2012 profile of Facebook users, she discusses native English speakers usage of Facebook on university Facebook pages. She categorizes these posts as a model of computer-mediated communication, a new communication style that combines features of writing and speech, and compares the usage of Facebook by native and non-native English speakers. Facebook posts, she notes, generally have a degree of formality, no matter whether native or nonnative English speakers, but native English speakers often have a higher degree of informality. She compares the informality and abbreviations used by native English speakers on these Facebook pages, which are mostly formal which often include abbreviations, a mix of capitalizations, and the usage of non-separated salutations, to those of non-native English speakers. Native English speakers often use peoples names in their posts to indicate if they are replying to a specific individual, while non-native speakers do not. In addition, non native speakers cited in the study use separated letter-style greetings and salutations, indicating linguistic insecurity (a lack of confidence in their English skills). Perez-Sabater concludes that computer-mediated communication does not always tend toward informality, and that social networks serve as a platform similar to non-virtual social networks, while also noting that further information must be found.

Bridging the Two Theories

The Leaders and Loners Study

One key study that employed computer simulations was Fagyal, Swarup, Escobar, Gasser, and Lakkarajud’s work on the roles of group insiders (leaders) and outsiders (loners) in language change.

File:Diffusion curve of linguistic variants on a condensed time scale.tif

Diffusion curves displaying the number of agents adopting each of the competing variants in the network over time

In this study, the researchers simulated a social network of 900 participants, called nodes, which were connected into a network using a matrix algorithm. They then randomly assigned a linguistic variant to each node. On each cycle of the algorithm, every node interacted with another node, and the variant assigned to each node changed randomly depending on which variant the other node had. This cycle was repeated 40,000 times, and at the end of each cycle, the variant connected to each node was recorded.

The results of the Fagyal et al. study indicated that “in a large, socially heterogenous population,” one linguistic variant eventually became the community norm, though other variants were not entirely eliminated. However, when the researchers manipulated the network to remove either loners or leaders, the results changed: without loners, one variant rapidly caused the loss of all other variants; and without leaders, no single variant became the norm for a majority of speakers.

These findings allowed the researchers to address the major debate in social network theory: whether it is leaders (or centers) or loners who are responsible for language change. In their findings, the presence of both leaders and loners was essential, though the two types of agents played different roles in the process of change. As the researchers write, “centers’ most essential role is to serve as conduits for innovations to propagate to others in the network.” Rather than introducing entirely new forms, leaders accelerate the adoption of forms that already exist within the network. Conversely, the researchers describe the loners’ role this way: “when loners are a part of a population structure that allows their influence to reach centrally-connected hubs, they can have a decisive impact on the linguistic system over time.” Previously, researchers had posited that loners preserved old forms that had been neglected by the larger community. Fagyal et al. complicate this claim by suggesting that the role of loners in a network is to safeguard old features, then reintroduce them to the community. Thus, both leaders and loners are responsible for language change.

See also

• Lesley Milroy
• Penelope Eckert
• William Labov
• Mark Granovetter
• Indexicality
• Sociolinguistics
• Social network analysis
• Social network
• Computational linguistics
• Fist bump (=ƎE=)

Further reading

• Bergs, A. (2005). Social Networks and Historical Sociolinguistics: Studies in Morphosyntactic Variation in the Paston Letters. Berlin: Walter de Gruyter.
• Chambers, J.K., et al. (2002). The Handbook of Language Variation and Change. Oxford. Blackwell.
• Chambers, J.K. (2009). Sociolinguistic Theory. Linguistic variation and its social significance(3rd Ed.). Oxford:Blackwell.
• Granovetter, M. S. (1973). "The Strength of Weak Ties". The American Journal of Sociology 78 (6): 1360–1380. doi:10.1086/225469. JSTOR 2776392.
• Labov, William (1966). The Social Stratification of English in New York City. Washington D.C.: Center for Applied Linguistics.
• Milroy, L. (1987). Language and Social Networks. New York: Blackwell.
• Trudgill, Peter (2002). Sociolinguistic Variation and Change. Edinburgh: Edinburgh University Press.
• Wardhaugh, Ronald (2006). An Introduction to Sociolinguistics. New York: Wiley-Blackwell.