Cam-UNAM SEDA scheme for analysing dialogic teaching-and-learning interactions:
SEDA Code Clusters ©2016
This document contains descriptions of the 8 clusters categorising the 33 communicative acts comprising the Cam-UNAM Scheme for Educational Analysis (SEDA). A detailed, full version of SEDA containing descriptions and illustrations of codes is available at https://docs.google.com/document/d/1De48c9GoUZKl0JHjquGR0fNhP9NSjZYeUFeL3lYwhTA/edit?usp=sharing and a one-page condensed version is available at Condensed scheme - published Jan. 2016.
Further details of development and illustrated applications of the scheme are in our article: Hennessy, S., Rojas-Drummond, S., Higham, R., Torreblanca, O., Barrera, M.J., Marquez, A.M., García Carrión, R., Maine, F., Ríos, R.M. (2016). Developing an analytic coding scheme for classroom dialogue across educational contexts. Learning, Culture and Social Interaction (http://dx.doi.org/10.1016/j.lcsi.2015.12.001). Further information and updates on our work are available on the project website at http://tinyurl.com/BAdialogue.
The Scheme for Educational Dialogue Analysis is made freely available under a Creative Commons Attribution (CC By 4.0) licence (international): http://creativecommons.org/licenses/by/4.0/. It can be used or adapted under conditions of attribution to the original research team using the following statement:
"The Cam-UNAM Scheme for Educational Dialogue Analysis (SEDA: ©2015) was developed by a research team from the University of Cambridge, UK, and the National Autonomous University of Mexico, led by Sara Hennessy and Sylvia Rojas-Drummond and funded through a grant from the British Academy. The original scheme and list of co-creators are available at http://tinyurl.com/BAdialogue.”
Under the licence terms you must also indicate if changes were made. We request that those using or adapting SEDA share their applications and findings with us. Please contact us also if you would like to trial one of the sub-schemes under development for use by teachers, for analysing peer dialogue or for contexts of technology use. Contact details: sch30@cam.ac.uk; silviar@unam.mx.
Cluster code | Cluster name | Description |
I | Invite elaboration or reasoning | Invite others to:
In 1 and 2 above, turns coded I should be explicitly asking for reasoning, typically through key words such as ‘why?’, ‘how?’, ‘what caused...?’, or speculation/prediction through conditional phrases such as ‘what would/could/might happen if...?’ Includes inviting to explain or speculate based on one’s own or other’s ideas. In 3 and 4 above, invitations should make explicit reference to prior contributions. If not immediately prior, use C also as appropriate. When asking for reasoning or elaboration later on, for example within an upcoming activity, use C instead. Excludes: asking for simple repetition of contributions. |
R | Make reasoning explicit | Make reasoning explicit through
or
Turns coded R should indicate a clear attempt at reasoning, typically (but not necessarily or sufficiently) through key words such as ‘because’, ‘so’, ‘therefore’, ‘thus,’ ‘in order to’, ‘if...then’, ‘not...unless’, ‘it’s like...’, ‘imagine if...’. The attempt need not be ‘successful’ - that is, reasoning need not be judged good in order to be coded. It should be remembered that when engaging in reasoning speakers will often be tentative and less than clear in their expression. Includes explaining or speculating based on one’s own or other’s ideas. Turns coded R need not relate directly to a previous comment, as reasoning may stand alone as a dialogic feature. Where they do relate, additional clusters can be coded as appropriate. A counter-argument (proposition + reason) or argument with expressed (dis)agreement can be coded R and P. |
B | Build on ideas | Make a relevant contribution to the dialogue by building on, adding to, reformulating or clarifying one’s own or other's contributions Includes: judging ideas to be similar or different to each other without evaluating them, and without giving reasons. If reasons are given, use R instead. When referring to comments, ideas or resources from outside the immediate dialogue either in time, place or person, use C instead. Excludes: repetition of one’s own or other ideas; contributions should add something either in terms of content or in the way ideas are expressed. |
E | Express or invite ideas |
Utterances are coded E when they are judged relevant to the dialogue but not explicitly linked to previous contributions. The response to the participant’s utterance should not be used retrospectively to infer the intent of his/her contribution. Utterances under 1 above should appear to invite relevant contributions without the explicit suggestion that respondents should build on what has been said before - even if this is implied in context. For example, “what do you think?”. If direct mention is made of prior contributions, ideas or contexts, use IR or C as appropriate. Utterances under 2 above should appear to contribute a new and relevant idea or example without an explicit link to a prior contribution or context. Where clearly and explicitly building on, responding to or evaluating previous contributions, use B R, C or P. |
P | Positioning and coordination of ideas | Take a position/ stance in the dialogue by:
Excludes: expressing agreement with one position or idea when no contrasting ones have been offered. Includes: agreeing with one position or idea after a different one has been mentioned, even if the disagreement with that different position is implicit; however, simply saying ‘I agree’ is not enough in this case - the particular position or speaker must be identified (e.g. I agree with Simon). This code excludes identifying similarity or difference between ideas without judging their value - in this case, use R instead if a reasoned distinction is made; if not, use B instead. |
C | Connect | Make explicit links to ideas / positions / arguments / artefacts / prior contributions or knowledge beyond the immediate dialogue that:
If the idea/ position/ argument/ artefact referred to has very recently been mentioned, use B instead. If it has to be explicitly retrieved/ recalled, use C. |
G | Guide direction of dialogue or activity | Take responsibility for shaping and directing dialogue or activity by:
Or through scaffolding strategies such as:
If a proposal for possible courses of action or inquiry extends beyond the lesson, use C instead. |
RD | Reflect on dialogue or activity | Explicit self or group evaluation or metacognitive reflection on purposes/ processes/ value/ outcome of learning or activity, also including:
|
Guidance notes
1. All acts and clusters (including ‘Reflect on dialogue or activity’) are only applied when they occur in the flow or context of dialogue.
2. A CA is defined by a single participant; it may only include more than one turn in exceptional cases where there is a minor interruption or parallel / off-task talk that does not affect the path of thinking and the original contributor then resumes the same line of argument, or continues an utterance or other CA, with no change of content or function).
3. Utterances/actions not coded with one of the CA may be coded as U for uncoded. When a CA covers more than one turn but with minor interruptions, parallel or off-task talk (see previous note), mark these as U (for uncoded). Note that reading out a set task or question is uncoded.
4. Contributions that repeat an earlier contribution by the same or a different participant are uncoded; only the first instance is coded, unless the participants change, eg a question is repeated to a separate group, initiating a new interaction.
5. Not all of the clusters in the scheme need to be observed in any single episode or sequence for it to be considered dialogic.
6. Coding as dialogic does not depend on subsequent take-up; an invitation to explain may not be taken up, for example. Nor does the coding include inferring intention.
7. Some codes have much more weight if the acts are done by a learner than by a teacher or more expert other. Researchers might mark CA with coloured font if the learner is acting, to indicate that it has more weight or significance. This may also affect the suggested rank order according to dialogicality since a CA may be given different weight if it is carried out by a learner or more expert other; researchers can re-order the codes within a cluster as they see fit, according to research purpose and participants in dialogue.
8. Less explicit and less sophisticated use of evidence in an argument, hypothesis/speculation, etc. (than the current descriptions imply) may be applied for young children or in some contexts, at the researcher’s discretion.
9. It is important to take into account the prior and subsequent (immediate and extended) context of an act before determining how to code it. If there is little or no agreement to code something, if it turns out to be hard to explain, or if it may not be clear to external observers, then it should not be coded.
10. SEGMENTATION: Our aim is to segment utterances where applicable to capture individual CA, although researchers can decide for themselves how to segment and how many instances of the same code to apply to a single turn when an utterance contains multiple instances. We try to decide if the same line of thinking is being followed (e.g. through successive elaborations of a single idea or explanation), in which case it is one code, or if there are different discrete ideas/reasons being expressed, in which case we code multiple times. It is also important to mark utterances that belong together but got split with an interrupting or overlapping intervention. If they are considered together, the code they get is only counted once.
For measuring inter-coder reliability researchers may wish to consider the total codes applied to a whole turn, or a lead researcher may segment the transcript before coding begins, perhaps allowing other coders to adjust it where they perceive it is needed. The procedure for segmentation will be particularly crucial for researchers who wish to count frequencies.
11. Researchers can choose to use some or all clusters, depending on the research focus and context. In particular, codes in the E cluster (‘Express or invite ideas’) may be very significant in terms of progressing a dialogue but may lack a specific dialogic intention to help the addressee understand someone else’s thinking (as embodied in elaboration and explicit reasoning); the cluster includes short, unlinked responses during a brainstorm, for instance. Likewise, a teacher or other participant may not take up any of the resulting ideas expressed in response to invitations for opinions/beliefs/ideas. The E cluster may suit many researchers’ purposes but possibly not those who wish to distinguish (and count frequencies of) clear markers of productive dialogue.
It is also notable that some codes can be very significant in terms of progressing a dialogue but may lack a specific dialogic intention to help the addressee understand someone else’s thinking (as embodied in elaboration and explicit reasoning); E2 (‘making other relevant contribution’) includes short, unlinked responses during a brainstorm, for instance, and it can occur with similar frequency in more and less dialogic lessons.
12. APPLICATION OF SINGLE/MULTIPLE CLUSTERS
Once a transcript is segmented as far as possible, researchers within a team need to agree on whether and how to code CA that appear to relate to more than one cluster, namely whether to apply more than one or to prioritise in some/all cases. Teams can make their own decisions on this, depending on context etc. The scheme’s originators apply multiple codes in order to capture the range of dialogic interaction and to avoid arbitrary hierarchical ordering of clusters, although this of course means that categories are not mutually exclusive and inter-coder reliability analyses are more complex. For information, some common cluster combinations that we apply are listed below.
Cluster code | Cluster name | Description |
R | Make reasoning explicit | If turns coded R do relate directly to previous comments, they may additionally be coded P, C, RD or G depending on the nature of the relation. For example, a counter-argument (proposition + reason) or argument with expressed agreement should be coded R and P. |
I | Invite elaboration or reasoning | If the invitation is to build on ideas/ opinions from outside the immediate dialogue either in time, place or person, you may also use C. |
B | Build on ideas | When responses are given as informative feedback to others, also code G. |
P | Positioning and coordination of ideas | Where reasons are given, distinctions drawn, or different implications explored in evaluating ideas/ perspectives/ arguments, also use R. When ideas/ perspectives / arguments are brought in and evaluated from outside the present time, place or person, also use C. |
C | Connect | If evaluation/ agreement/ disagreement with the ideas/ positions/ arguments/ artefacts recalled is invited, also use P. |
G | Guide direction of dialogue or activity | If a proposal for action or enquiry also includes a hypothesis, proposal or justification, also use R. When informative feedback is offered, B, E or R may be used in addition to indicate the form of feedback. Likewise when focusing is evident, other clusters, typically B or I, may be used in addition to qualify it. |