Tuesday, May 31, 2016

Multilingual extension in Africa: Framing the contexts

This post continues the discussion about multilingual extension and public education responses to emergencies where there are many languages from the previous post about the response to the West African ebola epidemic, with a look at general lessons and strategies.

Multilingual marketing & multilingual extension

There is a lot of information available about multilingual marketing aimed mainly at commercial users, and often with a focus on web content, for example on blogs and news sites (Business 2 Community, Business News Daily, ClickZ, Mention, Search Engine Journal), and corporate sites (ADAM, Ecoconsultancy, Lionbridge, MarketingProfs). Part of the pitch in these pieces is that most people base decisions like purchasing on information in their principal language, and most people do not speak English. Hence the need to localize messages, involving primarily translation but also attention to cultural norms and sensibilities. Multilingual marketing strategies allow a message to reach a linguistically diverse population - no mystery there. Most of the focus of such discussion, however, is on parts of the world with lucrative markets and good connectivity - much of the global North plus dynamic urban centers globally.

Although one can certainly tap such advice for information relevant to a range of situations, there is not any comparable font of advice that I'm aware of focused on strategies for extension, development communication, public health education, or messaging relating to crises in the multilingual global South, including Africa. These regions tend to have less disposable income and less connectivity than in the global North. And large numbers of their populations, if not majorities, speak languages that may be at once "less-resourced" in technical and literary senses, absent from school curricula or taught only in the first couple of grades, and given lower in status than that accorded the main European, Asian, and Middle Eastern languages.

The relative lack of prioritization of African language messaging by international entities involved in the ebola response can be seen in this context.

Understanding African linguistic realities

Planning for use of diverse languages in extension and public education, or even just accepting that it is worthwhile, requires understanding the "linguistic terrain," which may be defined as including information about the languages spoken (what they are and how they are related), the linguistic demography (who speaks what in what areas), and relevant sociolinguistic factors (how various languages are used, status of languages, etc.). Understanding the linguistic terrain, in turn, requires attention to specifics, and may be facilitated by models or "frameworks" which put such specifics into perspectives useful for action.

For instance, figures on the number of languages spoken in a country are readily accessed, and not infrequently these elicit in foreigners the impression of a kind of Babel, too complicated to deal with beyond maybe assuring there is local staff able to interpret as needed (else just relying on ad hoc interpretation in the field). Beyond the numbers, the names may be unfamiliar, issues of who speaks what where and when not understood, the close relationship of some sets of languages with different names unknown, and there's always someone ready to offer a rationale for not bothering with African languages.

In approaching the multiplicity of languages in multilingual African countries, the first task is getting past the numbers, to put some identity to the names, and then to frame some understanding of how they are (and could be) used, by whom, where, and when.

Hierarchies, relationships, and priorities

The simplest and easiest framework in which to approach linguistic diversity in Africa - which arguably the default understanding for many - is a hierarchy in which the "Europhone" international languages are higher, and "local languages" are lower. That could be on the basis of several factors - elite status, career and education opportunities, existence of written and digital resources, number of speakers globally - but not others - mother tongue status, cultural content, number of speakers locally/regionally.
Classes of development of a linguistic variety proposed by Kloss (1978),
as quoted in Mioni (1988). Source: Guerini (2006:33)

The diagram at the right, adapted from work by Heinz Kloss, illustrates what such a hierarchy might look like (though I'm interpreting it here for purposes different than it was proposed). In many African countries, a Europhone official languages is #1 in such a ranking (Afrikaans, Amharic, Arabic, Swahili might also be considered here), and most other languages are down the scale (per the diagram, they all would be nos. 4-6).

Such a simple hierarchy may have utility for description, but also has limitations - perhaps notably in confirming biases. It is probably not a good guide for communication strategies.

Hierarchy of  languages in de Swaan's (2001) theory. Source:
Wikimedia Commons, By Jqho1 - Own work, CC BY-SA 4.0
A hierarchical approach proposing a system of relations was Abram de Swaan's global language system theory (developed across several publications culminating in a 2001 book). He considered languages as forming a network of "constellations," each having more central and more peripheral languages. De Swaan's system adopts the core-periphery paradigm of world system theory, and includes arguments from history and a mathematical model for the communicative value of a language (Q-value). A more complete discussion of his work is beyond the scope of this post, but for our purposes, the model has been widely influential.

Source: Calvet (2006:76)
Among those referencing De Swaan, Louis-Jean Calvet (1999, 2006) explored how this approach works and may change in several regions. In one African case - Mali - it is clear that some languages do not fit simply into a hierarchical system of constellations (in a different discussion, we'd be calling many of these "cross-border languages," although the limits of constellations do not follow country boundaries).

In general the constellation approach seems most useful when analyzing the global or broad regional systems, and may not be helpful in communication strategies that require dissemination of messages in what are framed as peripheral languages.

Rough long-tail distro of 1st+2nd languages in Mali
A different approach that I have discussed previously (in general and in the case of Mali) uses the observation that numbers of speakers of languages as a rule follow a long-tail distribution. This is true at any scale, but the languages at the head of the distribution at more local levels may be different than those at a regional or country levels. This is not necessarily better than the constellation approach, but it does offer a different understanding of data that suggests different priorities. Arguably when a crisis such as ebola is most critical in certain localities, awareness of the language distributions on those local levels might be important for effective messaging there.

Types of African language repertoires in Mioni (1988).
Source: Guerini (2006:30-32) (combined 3 figures)
Since linguistic demographics and language policies differ by country it is also important to look at the organization of languages at that level. Alberto Mioni (1988) proposed three language profiles common in African countries: 1) a typical one with a "high language" (HL), usually an "exolanguage," what I have been calling Europhone languages, a national lingua franca, and local languages; 2) having two HLs, a developed national standard and an exolanguage, along with local languages; and 3) with local standard unavailable/undeveloped, and just the HL and local languages. These are simplifications, and like the hierarchical models above, they broadly minimize smaller languages that may be important locally, but they may provide a context for application of another analysis like the long-tail.

The reason for reviewing the above is that taken together, they might be used as alternative "framing" of a given sociolinguistic situation to arrive at a better understanding for planning purposes. That is, while studying the specifics of what languages are spoken in a country is essential (especially for foreigners working there), having diverse conceptual tools to understand the communication environment may be key to appropriate and successful messaging.

Leveraging the global networks of local languages

Due to emigration, there are communities of speakers of many African languages in diverse parts of the world. In other contexts (such as African language Wikipedias) it has been pointed out that these communities may be resources for development of materials.

In the case of epidemics that take on global dimensions, it may be possible to link multilingual messaging approaches between countries where a language is spoken natively, and communities of speakers in other countries (emigrants, diaspora). When fear of a wider ebola epidemic was at its height, the issue of translation was brought up in the US, for example, but apparently without making this link.

Taking multilingual extension to another level

The above is hardly comprehensive, but seeks to outline conceptual approaches to use in strategies for messaging in multilingual Africa, in emergencies like the ebola epidemic, but also general development communication. The previous post on lessons from the response to ebola, reviewed some specific issues. Taken together they are intended to promote discussion leading to more effective and appropriate use of African languages in a range of public education and development contexts.

  • Louis-Jean Calvet, 2006, Towards and Ecology of World Languages, Polity Press. (Translated by Andrew Brown from 1999, Pour une écologie des langues du monde, Plon)
  • Federica Guerini, 2006, Language Alternation Strategies in Multilingual Settings: A Case Study: Ghanaian Immigrants in Northern Italy, Peter Lang.
  • Heinz Kloss, 1978, Problems of Language Policy in South Africa, Braumüller.
  • Alberto Mioni, 1988, "Standardization Processes and Linguistic Repertoires in Africa and Europe: Some Comparative Remarks," in Peter Auer and Aldo di Luzio, eds., Variation and Convergence: Studies in Social Dialectology, Walter de Gruyter.
  • Abram de Swaan, 2001, Words of the World: The Global Language System, 1st ed., Polity Press.

Thursday, May 12, 2016

Multilingual extension in Africa: The response to Ebola

Various earlier posts on this blog have discussed the importance of African languages in extension and development from general and tactical points of view (with particular reference to agriculture). Also during the course of the ebola epidemic in West Africa (which peaked in 2014-5), a number of posts looked at specific efforts to undertake health education and ebola awareness in various languages, and proposed systematically sharing, vetting, improving, and providing for re-use of translated and original materials in African languages.

This post looks from several different angles at multilingual messaging, including translation, over the course of the ebola epidemic. It will be followed by another post considering what strategies one might identify for future responses to emergencies - as well as less urgent public education needs - in multilingual contexts.

First, however, a quick note concerning summaries or evaluations of the linguistic dimension of the ebola response. As far as I am aware, there is no comprehensive overview or analysis. Reports from major organizations involved in the ebola response barely mention language (at the end of this post is a quick review). Perhaps the only report specifically on language and translation in the response was done by the international NGO Translators Without Borders, which was prominent in arranging translations of ebola materials into many languages (not just in West Africa): "Words of Relief  – Ebola Crisis Learning Review" (29 May 2015).  For a little bit broader view, but without pretense to completeness, I posted a summary largely reflecting content on this blog: "Ebola messaging in African languages - taking stock and looking ahead" (12 Dec. 2015, on LinkedIn). Both of the above proposed some lessons for the future.

Even lists of resources and websites, mainly done for reference or reporting during the period of the epidemic, are not comprehensive (for example, here). As I'll discuss below, one reason has to do with the scale or "layer" at which these operate, meaning that local activities either tend to be missed by international activities (which do the reports we see), or just aren't seen as falling in their scope. But it may also just be that information didn't travel or networks weren't connected. When researching ebola projects recently, for instance, I discovered an analysis of an interesting multilingual cellphone messaging project in Guinea called "Polly" from a group at Carnegie Melon University - something I personally had not seen mentioned elsewhere. Are there others?

Image by Chris55, Wikimedia Commons, CC BY-SA 4.0

Ebola messaging in African languages

It is useful to disaggregate as much as possible several interrelated aspects of the process of responding to a crisis like ebola in a multilingual context: time; space; demographics of languages; the levels or layers of activity (international, national, provincial, local); modes of developing information in diverse languages (translation, interpretation, and creation*); media (print, media, mobile, in person); and organizational policies.

  • Time: Information about the epidemic came most quickly in English and French - official languages and main languages of media, the aid community, and the medical community. There was a lag in getting information in other languages of the affected countries. Major efforts to address this took a while to get going - according to TWB's report (cited above) its translation efforts began in November 2014 (though it seemed to have started some work in August), which was around the peak of the epidemic (per graph of cases above). Ad-hoc interpretation of incoming information on local levels (local radio, community meetings) was probably ongoing from relatively early in the cycle, with more organized efforts later on, though documentation is limited.
  • Space: Three countries were most hard-hit by the epidemic, and within those some regions within them more so than others. Responses obviously tried to address those, but logistics and information (such as for ebola education) tend to move from center to periphery. Data, on the other hand, may move from periphery to center by one or another means. Neither is perfect, but they evidently took time, especially early in the epidemic, even where there was mobile technology. Since the spatial distances were accompanied by language differences (different languages, different varieties of a language, or even different competencies in a same language variety), there was added time and evidently more potential for misunderstanding. All of this is in theory - I have seen no discussions of spatial dimensions of the response (looking in particular for the impact on information flows).
    • Linguistic demography: One dimension of linguistic demography is of course who speaks what where. Another is that on the country level, speakers of some languages are more numerous than others overall. Translations in languages more widely spoken on the national level were apparently done before less spoken ones. This makes sense on a broad scale but where less widely spoken languages dominate in a local areas, these likely were used orally (the backstory to the murder of health workers in Womey, Guinea included mention of local languages; a story from neighboring Côte d'Ivoire offered a peak at interpretation on the village level), but otherwise may not have received much attention. The only project I heard of that focused on translations in "smaller" languages was one organized by SIL in the neighboring Casamance region of Senegal (mentioned here).
    • Levels/layers: International aid and NGOs, national administration and services, and local communities and their local and extra-local networks seemed to operate on different and often only selectively interacting levels, especially early in the cycle. This dynamic meant that what happened or was being said on one level may or may not have had (immediate) effect on, or even in the extreme not been known by the others. And where information and diverse languages are involved, gaps and even contradictions (the latter in the form for example of misinformation about what ebola is or its treatment) seemed to emerge fairly quickly especially in critical early stages of an emergency.
    • Modes: Translation efforts focused on standard texts or posters, especially on the level of international aid and NGOs, but also some national level print media, and governmental and non-governmental entities. Interpretation and creation of information in African languages were more national and local in character - meetings, individual contacts, blogs. In the case of media (print, radio, TV) content about ebola, it is not clear the mode in which it was rendered in African languages (translation, interpretation, or creation*). For example, two international organizations working with media - International Committee of the Red Cross (ICRC) and Catholic Relief Services (CRS) - had projects with community radio stations to broadcast messages in local languages, but in neither case was it clear how that was done (which languages, whether oral translation from Europhone scripts or scripts pre-translated in target languages, what training, etc.). An example of community contacts in local languages to spread information about ebola in Liberia was given by International Rescue Committee (IRC) - the person in the example was a physician assistant and evidently fluent in the target language, so able to put information in his own words ("creation" in this model) to fit the needs and questions/challenges as he encountered them.
    • Media: Ebola messaging in African languages was conveyed on diverse media (print, poster, TV, radio, blogs, video, and mobile apps) as well as via community meetings and individual contacts. I am not aware of any studies of media activity during the ebola epidemic that might include discussion of languages used. An exception is information about mobile media used during the epidemic - "mobile-based interactive voice response technology," and apps like the SAWBO animations with African language voice-overs and the abovementioned "Polly." Also, an observation mentioned in an earlier posting that the dominant languages used in each means of communication tended to differ, such that in Nigeria supposedly the higher tech media used more English and the in person and community communication tended to be in first languages.
    • Organizations and languages: The observation by Nadia Berger and Grace Tang based on TWB's work translating ebola messages that "translation is not always considered a priority by governments and aid agencies" echoes experience more generally in international development. This is a major structural hurdle to more effective use of African languages in response to crises like the ebola epidemic, in health education, and indeed in all aspects of development. Do organizations and agencies have policies with regard to languages in the multilingual societies struck by the epidemic, or was the linguistic dimension of their work handled as an incidental consideration really given less priority in the face of a very physical crisis, despite its relevance to successful public education and control of disease transmission?
    With the latter point in mind, it is worth taking a quick look at whether and how language as a factor did or did not figure in reports of organizations involved in response to the ebola epidemic.

    Language in reports on ebola responses?

    Reports on ebola response efforts by Medecins Sans Frontiers, World Health Organization, and IRC did not mention languages or translation, nor did an analysis of the international ebola response by the UK Overseas Development Institute's Humanitarian Policy Group.

    The U.S. Centers for Disease Control (CDC), to its credit, did in its report mention language as a factor: "The Road to Zero: CDC's Response to the Ebola Epidemic." (Recall that CDC's ebola information resources included a few in Malinké - a variety of Manding). The two specific mentions merit a closer look:
    • A CDC worker with "intermediate" French worked with people Guinea's forest region, of which perhaps only one per village might speak French, to "translate important  health messages into the local language" and train local workers. An impressive effort, but the kind of situation where available language support - people with better French and acknowledged erudition in the local language(s) - could help assure accuracy of translations and messages in the local language. Also, since that part of Guinea has several local languages, it is not clear which was the focus of efforts
    • A training of midwives and cleaners was held in the Malinké language "since they do  not speak French, the official language of Guinea." Looked at another way, a big reason to use an important first language and lingua franca like Malinké would be that this is understood better, and that an advantage of training using such languages is that it could arguably better prepare trainees for work with patients who may also speak them better than the Europhone official language, French. (Another possibility might be a bilingual format with relevant African and Europhone languages, arguably better preparing trainees for contexts they will encounter.) 
    Reports or website content of all but a few of the 67 organizations on USAID-CIDI's list of NGOs involved in ebola response also did not discuss languages or translation (to be fair, a few of those which didn't were only involved in logistics of shipping supplies). CIDI's list is extensive, but notable among the omissions is TWB, certainly the most prominent international NGO facilitating translation of ebola materials. The exceptions which did mention language include:
    Another organization on the CIDI list, Mali Health Organizing Project, used Bambara in its ebola response, as one would expect (for example in a factsheet), but did not mention this linguistic dimension in its report or website.

    In all of these instances it would be helpful to have any materials developed/used for review (were they saved?) and incorporate them in archives for potential reuse (per 2Ds & 4Rs).

    Perspectives of health workers & patients?

    Another level at which language is important in such an epidemic is that of the interaction between health workers and the patients. For a glimpse of this issue, see two personal accounts by Western nurses about the language gap between them and patients (in Nurse.com and PIH.org). It would not of course have been possible to train foreign health workers in any of the local languages they would encounter, given the time factor, but one wonders if a couple of words or phrases might have made the difficult experience a bit more bearable for caregiver and patient (especially the younger ones).

    * Note on these 3 terms. On the first two, there seems to be a gap in the terminology: "translation," strictly speaking, refers to text, as distinct from "interpretation" being defined as oral and in person (physically or virtually). This leaves out what we might call oral translation, or interpretation of a source either received as text (announcer on radio reading in one language but giving the meaning in another) or heard in one language (in a meeting or on media) but transmitted orally in another language at a later time. These kinds of cross-language sharing are common in multilingual societies with rich oral traditions, and I suggest in this context they be considered as forms of interpretation (keep in mind the range of meaning of "interpretation" as an art). "Creation" is intended to cover instances where the writer (for example of a blog) or speaker (many possible scenarios) is conveying their understanding of a subject in their own words - which happen to be in a language different than that of the main international media, medical research, etc. Of course the lines among these three are not rigid, but I think it is useful to retain them as more or less archetypes of different ways information about ebola was communicated in African languages. (The neologism "transcreation" has specific applications, but might constructively be adapted for use somewhere within this space.)