Doing SSNA in multiple languages


#1

For some studies, we will want to engage in multiple countries, and across multiple languages. In what follows, I briefly explore what this means in practice.

The Edgeryders method has three main components.

  1. Convening an online conversation (primary data)
  2. Its ethno coding (secondary data)
  3. The actual SSNA, where a SSN is built and explored starting from the primary and secondary data.

Convening a conversation has to be done independently in each language. Each language needs a category on edgeryders.eu, its own social media person, its own community manager. Some of the engagement moves (for example the large offline events, à la OpenVillage Festival), can and should be unique for all languages.

Coding the corpus, again, needs to be done by ethnographers who are native, or at least highly skilled, in the language of the informant.

Doing SSNA does not pose problems. The data model is robust enough to support both a global view on the data (the SSN for all-countries, all-languages) and a local one (the SSN for one single language). This is simply a filtering.

What this comes down to is working in teams. There needs to be an engagement team, a community management team and a coding team. Team members have a certain operational independence, but they do need to coordinate. For example, visuals or high-quality post could be produced only once, and then translated across the different languages. The coding team, of course, needs to produce and maintain one single ontology, with the local codes (example: travail and arbeit) being children of the general “parent” code (example: work).

Training community managers and ethno codes is not a big deal. We do that. I propose we do not let this become an issue. We have assembled many project teams from scratch, and they all turned out well. We will do so again.

What is a big deal is that this way of doing thing is obviously less scalable than how we did opencare. Costs go up. At least three people need to be hired for each language.

Two more observations.

  1. I suggest always having English (more specifically, Euro English) as an extra language. This is the Internet, and people will participate if they want to, even if they are not based in the countries you want to study. It makes no sense to lock wilful participants out, since they are giving us data for free. English will be the default for people who do not speak French or Hungarian or what have you, but still want to be part of the conversation. It will also add an element of depth: how, say, non-Finns perceive the problem as opposed to how Finns do.

  2. It makes sense to do this work as a continuous activity, not a phase. We deploy on day one f the project, launch the convo as soon as possible, and the ethnographers start the coding maybe at month 7. Jan did not seem clear on this – which is reasonable because we never told him.


#2

Thanks @alberto. More on each case as they come