October 11, 2013, by Brigitte Nerlich

Do online user comments provide a space for deliberative democracy?

This is a guest post by Luke Collins who is working with Brigitte Nerlich on an ESRC funded project dealing with climate change as a complex social issue. Yesterday, he gave talk about his research to an interdisciplinary audience attending the Institute for Science and Society/STS PG seminar series.

The internet has enabled traditional newspaper organizations to distribute their publications online, as well as offering the opportunity for the readership to respond, most commonly in the way of user comments. Some hope that online journalism creates spaces for more deliberative and democratic engagement with news stories. Though journalists may remain the ‘authority’ on online content, with online resources we find the greatest potential for that shift from journalism as a ‘lecture’ to a ‘conversation’ and the opportunity for discourse as a fundamental principle of democracy.

Some scholars, however, question the impact of reader comments on realising the ‘deliberative democratic potential’ of online discussion. Furthermore, rather than fostering critical debate, some suggest that the space for freedom of expression has led to polarized and extreme views. Painter observes that particularly in the U.K. and the U.S., “climate change has become (to different degrees) more of a politicised issue, which politically polarised print media pick up on and reflect”. Popular Science has elected to shut down user comments altogether. There is also, of course, the issue of moderation. This suggests a need to examine the potential for deliberation in online discussion threads and whether they really do offer spaces for democratic debate.

User comments

The wealth of comments which follows articles poses methodological challenges social scientists and linguists studying them. Previous research looking at online user comments has generally been based on manual content analysis and as such has been limited in the scope with which it can represent online debates. Corpus analysis is a systematic and automated process based on the statistical analysis of word frequencies which allows us to process larger datasets more quickly and more objectively. In turn, this allows researchers to explore much broader questions in relation to online journalism across larger datasets. Some scholars dealing with climate change issues have begun to make use of corpus linguistics to study online reader comments.

The corpus analysis tool WMatrix3 that I used for this analysis has a built-in semantic categorisation function, which allocates each word of the data to a category based on its semantic meaning. The software tool is then able to determine which are the most key categories based on a statistical comparison with a normative corpus provided by the British National Corpus (BNC) as a representation of ‘normal’ language use. This allows us to make broad comparisons between user comment threads about what are the key themes of the discussion.

Taking, for example, three articles on climate change from both The Guardian and The Daily Mail online which received the highest number of comments, the graph below shows which semantic categories made up the ‘top 10’ for each thread:



In all three discussion threads from The Guardian and all three from The Daily Mail the categories of ‘Evaluation: True’ (incorporating words such as ‘proof’, ‘evidence’, ‘fact’, ‘truth’); ‘Weather’ (words such as ‘climate’, ‘weather’, ‘snow’, ‘wind’); and ‘Science and technology in general’ (‘science’; ‘scientific’; ‘thermodynamics’) were prominent:


Those categories that were prominent only in the discussion threads taken from The Daily Mail were ‘Substances and materials: Gas’, which largely consisted of references to CO2; ‘Geographical terms’ and ‘Temperature: Hot/On Fire’, which together accounted for frequent use of the term ‘global warming’; and ‘Evaluation: False’, which in contrast to the category incorporating words of ‘truth’, ‘proof’ and ‘evidence’, was made up of terms such as ‘lies’, ‘hoax’, ‘false’, ‘deception’ and ‘misleading’. This category was not as prominent in the discussion threads taken from The Guardian.


The implication is that much of the debate is based around the available scientific evidence for anthropogenic climate change, which is perceived by some to be manipulated and part of a ‘hoax’. Those categories which were prominent only in the threads taken from The Guardian were ‘Existing’ (including the words: ‘is’, ‘are’, ‘being’) and ‘Negative’ (‘not’, ‘nothing’, ‘neither’) which represent a pattern of claim and counter-claim/rebuttal, such as ‘This is not credible’. This in the very least is dialogic in that it shows engagement with claims that have been made in the discussion. As flat-out denial however, there is little potential for mediation. This level of engagement was also manifest in the prominence of the category ‘Other proper names’, which was largely made up of the monikers used by users in the thread as commenters made reference to specific users and their comments.


One important thing to consider in such discussion threads is the continual presence and contribution of particular users and consider the 1-9-90 principle. In the discussion threads referred to here, it was often found that 8-10 users would be responsible for 20% of the comments and that these same users were the most prolifically referred to in the comments of others. Two-thirds of users made only a single comment. This is indicative of a limitation on the ‘democratic potential’ of such online spaces, which appear to be dominated by a small number of ever-present users. However, researchers have found that high posters showed a “marked preference for a dialogical mode of address”, which would suggest that such users encourage deliberation, or in the very least that they acknowledge the contributions of others.

‘Dialogic expansion’ and ‘dialogic contraction’

To examine the potential for deliberative and democratic engagement we can refer to the language of intersubjective stance and the ways in which individuals invite or inhibit deliberation through aspects of their discourse. We can refer to ‘heteroglossic engagement’ which describes utterances which engage with dialogic alternatives. Recognising that individuals communicate their stance – or their association around a particular idea or principle – through discourse we can distinguish between the strength with which they associate themselves with that idea (modality) and the potential for alternative ideas to be considered. At one end of the spectrum we find the bare assertion: the statement as if ‘fact’ that makes a simple claim in simple terms. More often however, utterances consist of discursive features that – to some degree or another – recognise that the statement exists amongst a multitude of alternative positions. Within the ‘heteroglossic’ there are those linguistic resources which are seen to be ‘dialogically contractive’ and those which are ‘dialogically expansive’.

Whether a statement is ‘dialogically contractive’ or ‘dialogically expansive’ is largely determined by stance indicators, including modal verbs ‘can/could’, ‘may/might/must’, ‘shall/should’ and ‘will/would’. We can see how the more uncertain terms, ‘may’, ‘could’, ‘might’ more effectively encourage the consideration of alternatives than the assertive ‘will’, ‘shall’, ‘should’. The use of modal verbs in this way not only indicates the individual’s attachment to a particular stance and the potential to ally themselves to new alternatives, but also welcomes alternative voices or propositions from other interlocutors.

A statement such as ‘if it could be proven that’ is speculative, it entertains an idea and as such sets a precedent for other contributors to discuss in terms of possibilities. This kind of expansion is necessary for the introduction and development of new ideas, of alternative viewpoints and for learning.

In contrast, a matter-of-fact statement such as: “Extreme (cold) weather won’t falsify AGW”, regardless of whether it is agreeable or not, offers little encouragement to deliberate. This kind of analysis demonstrates that there are multiple opportunities for commenters to negotiate the expansion or contraction of the debate, that is, to open it up or close it down. In reality, individuals will offer both dialogic expansion and dialogic contraction alternatively across sentences and we will always want to make assertions in certain terms. But to encourage a culture where uncertainties are part of the discussion at the level of discourse encourages interaction and deliberation.


Posted in Climate ChangeLanguagepublicsScepticismScience