Title: Discourses, power and communities: exploring the impacts of social media technologies on the theory and practices of informal learning
Author: Peter Evans
Organisation: The University of Edinburgh
Address: Institute for Education, Community & Society
The Moray House School of Education
The University of Edinburgh
Holyrood Road
Edinburgh
EH8 8AQ
United Kingdom
Email:
Stream: Critical theoretical and methodological issues in HRD
Submission type:Refereed paper
Abstract
This paper sets out an investigation of the social practices and community-forming activities associated with professional development activities in social media environments. While claims are made on the non-hierarchical nature of these social media and informal learning environments, as with any social practice, they include issues of power relations. This study focuses on the emergence and evolution of power relations within open online environments for learning. The study explores how competing projections of power emerge and are “processed” in a specific digital social learning environment to impact on community creation through collaborative meaning-making actions. The study examines whether such informal learning environments are sites for “restrictive” or for “expansive” learning reflecting similar discursive power relations specific to those found in other, more formal, learning environments.
The study has implications for practitioners in terms of the design and facilitation of learning interventions using social media technologies. In addition, this study does also point to the usefulness and challenges of Actor Network Theory and Discourse Analysis as research approaches for the study of social interactions in social media environments for learning.
Key words: social media; technology; social learning; learning community
Discourses, power and communities: exploring the impacts of social media technologies on the theory and practices of informal learning
Introduction
This paper reports on an investigation of the social practices and community-forming activities associated with professional development activities in social media environments. While claims are made on the non-hierarchical nature of these social media and informal learning environments (Bingham and Conner 2010) yet, as with any social practice, they include issues of power relations (Huzzard 2004). This study, therefore, focuses on the emergence and evolution of power relations within open online environments for learning. The study explores how competing projections of power emerge and are “processed” in a specific digital social learning environment to impact on community creation through collaborative meaning-making actions. It also examines whether such informal learning environments are sites for “restrictive” or for “expansive” learning, reflecting similar discursive power relations specific to those found in other, more formal, learning environments (Fuller & Unwin 2004).
The study research site is a regular open Twitter “chat” event for those working in or interested in learning and development as a profession. The use of social media to enable collaborative and peer-to-peer professional development activities has become increasingly common in recent years (McCulloch, et al 2011; Bingham and Conner 2010). Twitter is described byLerman & Ghosh (2010) as
… a popular social networking site that allows registered users to post and read short (at most 140 characters) text messages, which may contain URLs to online content, usually shortened by a URL shortening service such as bit.ly or tinyurl. A user can also retweet or comment on another user’s post…
Specific “conversations” and discussion events are organised through the convention of hashtags (#) (Bruns 2011). Professional learning events hosted on Twitter include: #ARchat (business analysts); #brandchat (branding); #edchat (technology and education); #hcsm (healthcare communication and social media); #imcchat (integrated marketing communication); #innochat (innovation); #lrnchat (corporate and academic learning); #kmers (knowledge management); #pr20chat (PR and social media); #sbbuzz and #smbiz (small business); #scriptchat (screen writing) and #talentnet (recruitment industry). There are in excess of 100 of these professional community events scheduled to take place weekly along with many others on non-professional subjects (Swanwick 2010).
The research approach
The research approach draws on social constructivism (Vygotsky 1978) as the dominant theoretical perspective in educational research (Phillips 1995; Fox 2000) which is also becoming increasingly prominent in management research (Alvession & Skoldberg 2009; Cunliffe 2008). However, seeing learning as essentially a practical act draws in notions of pragmatism (Cook & Brown 2005) as well as practice (Bourdieu 1977; Antonacopolou 2006). However, the research focus on interactive digital environments, involving complex interactions of people, artefacts, language, collaboration and control (Nicolini et al 2003; Guzman 2009; Geiger 2009; Tuomi 2000) suggests a socio-material practice-based research framework. Early analysis of the research data indicated the dynamic and unstable nature of the social interactions (Sorenson 2007) under investigation. This in turn, suggested a research framework that acknowledged the dynamic interactions between the human and material. Thus the research makes use of Actor Network Theory as a socio-material and practice based framework operationalised through case-orientated research design using Discourse Analysis. Actor Network Theory (ANT) (Latour 2005) provides a “lens” for the interpretation of the discursive data. Given the focus in the study on interactive digital environments that can be labelled as Web 2.0, a practice-based approach that is concerned with the complex interrelations between people, artefacts, language, collaboration and control seemed appropriate (Nicolini et al 2003; Guzman 2009; Geiger 2009).
Actor Network Theory
This study made use of three key aspects of ANT: translation; network assemblages and symmetry. ANT has been described as a sociology and translation (Latour 2005) whereby translation refers to the interpretation and reinterpretation of knowledge or meaning as a means of enrolling actors into a particular network (Mitev 2009). Processes of translation work to both generate as well as order and stabilise networks (Fenwick & Edwards 2010, p9). ANT’s interest in network assemblages is less concerned with the size of networks but rather with the dynamics of influence in and on networks (Fox 2005), which underpins a central concern of ANT with power as persuasion (McLean & Hassard 2004). Finally, symmetry is the avoidance of subject-object dualism that privileges the human while avoiding technological determinism (Mietten 1997). So ANT understands “actors” as being either human or non-human active participants within networks.
Discourse analysis
Discourse analysis (DA) is concerned with studying “language in use” (Nunan 1993, p7) and can be seen as operating at a number of levels (Fairclough 2003; Alvesson & Skoldberg 2009). Heracleous (2006) identifies two levels of discourse: communicative action based on interactions between individuals to, for example, share experiences or build relations, and deeper discursive structures that ‘guide’ communicative actions. Individual utterances interact with one another to build patterns of sequences of exchanges that can be characterised as conversational floors (Simpson 2005, p337).
Dennen (2008) states that a focus on structural aspects of discourse can be used in the analysis of non-discursive factors such as group dynamics. However, a structural approach to analysis does not provide for effective understanding of collaborative sense-making. Dennen suggests the use of forms of dialogue and content analysis that are highly interpretive.
Bragd et al (2008) argue a discursive community generates common meanings through discussion so each utterance can be treated as being created by a group of people rather than as the isolated acts of individuals (Dennen 2008).Thus, discourse is a mechanism that enables people to ‘feel’ part of a community by contributing to a particular discourse with particular uses and particular terms that are commonly understood. So a community is generated around some level of discursive structure. Arguably, discursive communities enable learning that seeks to reinforce common understandings among the members as well as highlighting perspectives that differentiate members from ‘others’ outwith the community (Bragd et al 2008).
The events examined in this studywere selected at random from one of these regular Twitter chat events for people working in, or interested in, learning and development. The two selected events took place in October 2009 (Event 1) and January 2010 (Event 2). The events involved respectively 54 participants (N=54) in the former and 72 participants (N=72) for the latter Each participant has been anonymised as far as possible (Androutsopoulos 2008; Eysenbach & Till 2001) in agreement with the event facilitators. During the event, on October 2009, 922 tweets were made in the archived transcript giving ameanaverage of 10.2 tweets per minute. In the event of January 2010, 773 tweets were made at a mean average of 8.6 tweets per minute.
The research site
Each event is organised around a particular theme. Event 1 was focused on the use of metrics in learning and development while Event 2 was focused on crowdsourcing (the process of problem solving by outsourcing the activity to an essentially undefined network of people, the ‘crowd’) in learning and development.
However, the data boundaries of these online events cannot be clearly prescribed. Schneider and Foot (2005, p158) use the term web sphere to denote:
not simply a collection of web sites, but as a set of dynamically defined digital resources spanning multiple websites deemed relevant or related to a central event, concept or theme.
For example, at the end of Event 1, a number of participants advertised other events that may be of interest to the participants and a few included urls to personal professional sites, requests for help and responses to questions raised:
Extract A, Event 1
8:46:36 / @S / @G I found this to be useful – I’ll be doing a deep dive on it this weekend.9:56:01 / @C / Mike from Connecticut
9:55:41 / @M / M, – You can help me by keeping me engaged, learning — and hitting my session at #dl09 :)
9:57:02 / @AB / Gina, outside Boston, Grad student and course developer sometimes instructor for #EMC, looking @ snow on the radar !!!!
Extract B, Event 2
8:45:47 / @r / MTurk – #lrnchat9:58:19 / @tt / Christy, great white north. Currently lost in thoughts about learning & the computational internet...
Structure of the discourse
Thediscussion events can be considered as a cooperative learning event. However, where analysis of class room discourses have tended to look for a discursive structure of initiate – response – evaluation/ feedback (Bloome et al 2005) events that lack an explicit pedagogical focus tend to be less structured (Belnap & Withers 2008, p8).
However, all discourses have common structural building blocks based from the single utterance or move, or specifically in this case, a tweet. Such moves indicate a particular function (Fairclough 2003) such as asking questions and providing evaluations, information or facts. Two or more moves from different participants become sequences of exchanges (Belnap & Withers 2008). However, as Bloome et al (2005) suggest, these structures are not necessarily always clear as different functions and participants become involved and so divergent patterns emerge from the discourse.
In a study of unstructured learning events Belnap & Withers (2008) provide 16 functional categories for moves. The categories of particular interest here are the building blocks of sequences including: (a) suggestions directly addressing the dominant task and (b) propositions contributing to the development of the discussion. Moves and exchanges can be linked through the use of (c) modifications and (d) clarifications. The validity of statements can be addressed through: (e) justifications; (f) invalidations; (g) confirmations; (h) qualifications; (i) restatements and (j) simple responses as basic acknowledgements of statements often used to indicate acceptance. Restatements (i) are often given as retweets and play an important function in twitter discussions. Retweeting is the practice of ‘forwarding’ the message of another and is a common practice on twitter in general (Boyd 2010). Boyd (2010) identifies a range of reasons for retweeting that appear pertinent to the specific event such as: spreading a tweet to others; indicating support or homage; validating the comments of others; to attract new followers or gain prominence from more visible participants. For a specific discursive event, retweeting can perform additional functions such as repairing or reinitiating a sequence that had stalled, or to maintain a collective focus on the formal topic. At another sample of these events, a specific request is made for a retweet of the formal question as the participant stated: “I need some level of structure”. In the text of Twitter, retweeting the tweets of specific authoritative participants can act as a way to legitimising ones own comments, which is similar to what Fairclough (2003, p100) termed mythopoesis with authorisation. (Belnap & Withers 2008).
Sequence building
As can be seen in Table 1 from Event 1, sequences tend to build up over a number of short exchanges. The sequence is initiated by a direct question from the moderator: “How can crowdsourcing assist workplace learning? Personal examples?”. It is notable that the initiation by a direct question from the event moderator receives only one direct Suggestion but appears to provide an umbrella for a series of propositions contributing to the broad topic that appear to generate further exchanges.
Table 1
8:45:18 / Initiation8:46:34 / Suggestion A
8:47:07 / Proposition B
8:47:32 / Proposition C
8:47:59 / Proposition D
8:48:13 / Restate (Retweet)
8:48:22 / Restate (Retweet)
8:48:36 / Proposition E
8:48:50 / Restate (Retweet)
8:48:55 / Extension/ qualification
8:49:45 / Simple response
8:49:49 / Restate (Retweet)
8:50:34 / Invalidation/ new proposition D2
8:50:35 / Restate (Retweet)
8:51:06 / Qualification
8:51:26 / Restate (Retweet)
8:53:24 / Qualification
8:54:24 / Extension/ new proposition B2
8:55:35 / Restate (Retweet)
8:55:38 / Extension
8:56:16 / Restate (Retweet)
Proposition B2 initiates a new exchange of requests for information on the future progress of an organisational change initiative. This exchange of requests terminated when the original author of Proposition B2 confirmed that further information would be posted to the event blog.
However, the sequence commencing under Proposition D appears to terminate at a restatement (at 8:51:26). But a new exchange sequences was generated earlier with Proposition D2. The Proposition generated two short sequences progressing in parallel:
Table 2
8:50:34 / Invalidation/ new proposition D28:51:39 / Validate/ extension
8:53:12 / Extension
8:53:23 / Invalidation
8:54:34 / Simple response
8:55:00 / Extension
8:57:03 / Qualification
Event 2 tended to display far less developed sequences building and developing on initial propositions. So, there appears to be less interactive sense-making activities and knowledge exchange as evidenced through building on the initial Propositions.
Throughout the discussion events, sequences are displayed in a fragmented manner co-terminously with other sequences, such that each exchange sequence becomes difficult to follow. The duration of Table 1 from Event 1 of 10 minutes and 58 seconds involved 137 individual tweets. Twitter is particularly ‘messy’ in this regard as the discussion event is presented as a single chronological list of tweets without any threading to indicate links to specific exchanges. In particular, where a tweet does not include a reference to a previous tweet, it is often difficult to identify whether the tweet is part of an existing exchange or not. This difficulty may account for the seemingly short duration of each sequence ‘run’. Furthermore, as indicated in Table 1, there are competing temporal frames between the timescale of exchanges and sequences and a more rapid transcript time of the flow of event tweets[1].
So a Twitter event appears to exaggerate many of the key problematic features of unstructured discussions identified by Belnap and Withers (2008, p8) including: sequences extending over many exchanges; overlapping exchanges and sequences; short sequences tending to be cut off prior to a conclusion and sequences re-emerging later in discussions. This would suggest that the lack of event coherence and stability should be more problematic, but the participants appear to develop a range of approaches to deal with this.
Furthermore, while the exchanges themselves are generally of a limited duration, the overall discussion arguably displays a deeper discursive structure (Heracleous 2006). Arguably, it is analysing how particular patterns of discursive structure may emerge or be imposed (Cherny 1999) that is of a particular concern in this case rather than using models of turn-taking (Simpson 2005).
Simpson (2005) refers to conversational floors in computer-mediated discussion in terms of the cohesion and coherence in the discourse (p338). Here, the floor is collectively generated in that the floor must be both stated and then accepted by other participants (Simpson 2005, p345).
It is noticeable that Suggestion A and Proposition C in Table 1 are both retweeted but then are not taken further, that is, they are not accepted by the wider body of event participants. However, Proposition B generates direct dialogue in the sense of an extension and qualification rather than the simpler retweets and goes on to generate further exchanges.
Table 1, Suggestion A and Propositions B, C and E all come form the same participant and so can be seen as an attempt to ‘capture’ the conversational floor this may be to control the discourse direction or to set the floor as a means to generate discussions relevant to the formal topic of the event. This specific participant has a high profile in the sector and in these specific events and it may be that their attempts to capture the conversational floors are accepted and supported by others as a form of mythopoesis (Fairclough 2003) or acceptance of referential authority (French & Raven 1959). However, in Event 2, this was less clear as Propositions more often were generated by a greater variety of participants and produced less vertically orientated responses that would be expected if mythopoesis were a factor in the sequence structures.
Community and identity
Concepts such as Legitimate Peripheral Participation and the Zone of Proximal Development (Lave & Wenger 1991) did not appear to be applicable as the recognition of expertise authority itself appeared to be highly volatile and unstable. It appeared that the dynamic nature of the “community” placed a focus in the discourses on participants in general seeking commonality and consensus (Fairclough 2003) rather than displaying processes of Legitimate Peripheral Participation of some into a dominant perspective.
Extract C
1 / 8:47:59 / @V / learning is not a metric that businesses care about: positive performance change is.2 / 8:48:13 / @AD / Business metrics are often financial, or engineering based in some way- quite different from learning
3 / 8:48:50 / @AG / : here, here! RT @V: learning is not a metric that businesses care about: positive performance change is.
4 / 8:49:45 / @AJ / @V Agreed
5 / 8:49:49 / @AK / RT@V learning is not a metric that businesses care about: positive performance change is.
6 / 8:49:57 / @AG / funny, if we want to performance metrics, why don’t we identify performance objectives instead of learning objectives?
7 / 8:50:06 / @F / Good way to put it RT @V: learning is not a metric that businesses care about: positive performance change is.
8 / 8:53:33 / @AF / Measuring the performance (instead of learning) does the trainee produce at a higher level, make fewer errors after the training?
9 / 8:54:24 / @AE / @P I’m overhauling our dumb kirkpatrick model in favour of new return on perf & return on learning measures – launch /Jan
The discussion in the extract can be seen as a negotiation on the agreed professional identities of the participants. Line 1 and 2 suggests that learning is in some way separate, in measurement terms from performance and this view gains approval in lines 3 – 5 and 7. Line 6 questions the assertions that learning and development is business performance focused – “if we think X, why do we do Y?”. Line 8 is an attempt at integration by highlighting the alignment of individual learning with the instrumental needs of the business. Line 9 reinforces the supports and extends this with reference to specific actions the individual intends to do. It should be noted that, while lines 8 and 9 can be seen as part of a common discourse on learning and performance, they can also be located in separate albeit overlapping exchanges.