Monday, 24 June 2013

Yu's on First

No matter how clearly we think we've written something, we've all had someone who's read our email, our essay, our simple message and come back with a completely different reading than we intended.  The miscommunication, the multiple interpretations come at the semantic level in my experience, the result of a disagreement on degree or connotation or tone.

Similarly, we've all misheard things.  "Have I got cat spies?" "No. Have you got a half size?"  Ah.  Our brain fills in as best it can around the morphemes it missed even though we know what we came up isn't semantically solid.

Here in Gulu, I can only describe the mode of misunderstanding as feeling a bit like a hybrid of these two-- on the one hand, the problem I'm seeing happens between writer and reader, but on the other hand, the nature of the problem isn't as much semantic as it is denotative. 

In my experiment, I ask participants to fill in one of two surveys in Acholi. (surveys designed with native speakers)  I was prepared for the semantic level misunderstandings surrounding some questions, but I was not prepared to have denotative deviation, something along the lines of my writing the word 'carrot' in English and having someone interpret it as 'marble.'  That's basically what has been happening, a dispute at the foundational level of word meaning.  I'll give a second example in case that wasn't clear.

I throw the ball.  (what is written on the page)
I throw the fight. (what you, a native speaker of this language understand from what you've read)

I've never had this kind of misunderstanding before.   

Not every participant has the same misinterpretations, so it's not a mistake, typo, or spelling variation.  I don't entirely have an explanation yet, but in one instance a participant will give me a lecture about the inaccuracy of my survey to have included such a bizarre term and in the following session the individual sees no problem.

I understood from my research and from other linguistic environments, that with an oral language, meaning is tied to listening and speaking.  This experiment brought this process into relief for me as a native speaker of a chirographic language because I was able watch the meaning mechanism weaken on the page, dissolve without the context of sound.  In the short-form writing mandated by mobile applications, meaning becomes fuzzy, unpredictable, contestable. (I wrote about the 6 meanings of gwok in a previous post.  Coo has 9)

One of the series of questions I asked participants before we began the experiment was about language use in daily life, at work, and with their mobile device.  Did they mostly talk on their phones, talk and text, how did they use their phone?  Most respondents were college educated.  Many were multilingual.  The most common answer to, 'How do you use your phone?' was that they rarely texted or not at all.  They preferred to call even though texting is cheaper.  Everyone texted in English.  And they told me this with a tone that said this was a foolish question.  A few later added that they could mix languages to joke with friends in Acholi, or they can send a text to their grandmother in the village who does not speak English.  I have no quantitative data to look at how frequently this happens.  There are other studies which look at the frequency and context of language mixing.  A participant from a micro-finance organization described why he used English to text:
“We wish to text in Luo, wish to maybe send voice mails in Luo, but [the mobile phone] has been invented in another country, so in English we have to now do that. [me: Why couldn’t you text in Luo?]  In Luo?  It is not easy. You know our language is very short. chuch chuch chuch [noise of texting action].  So to formulate is not easy.”
He was describing the problem of the misunderstanding I had encountered.  It is avoided in speech because Achoil (sometimes called Luo) uses a lot of repetition as well as intonation to convey meaning.  This kind of thing isn't easily adapted to mobile applications where the design focus is on minimalism and streamlining of interface text.

Have now wrapped up my three months in Gulu with transcriptions and translations mostly completed.  On to data analysis.  Will be sharing first impression from the field at Africa Writes, July 5th at the British Library. 

Tuesday, 11 June 2013

Hardware Heartbreak

Speaking to people is the easy part.  Getting the mobile devices, laptop, digital recorder, camera, and 3G modem to cooperate-- that's the circus of modern fieldwork.

Everyone, research participants that is, seems delighted to take a short break from their workday and find out what I'm up to.... balancing my laptop with a Nollywood-style video playing, repositioning the set-up on a plastic chair to find the sweet spot for the 3G modem, trying to nonchalantly hold the digital voice recorder so that participants forget I'm recording (which means I sometimes forget myself and cover the mic with my hand), and pausing from pen and paper note-taking to pass a mobile phone where they fill out a brief survey about the video they've watched.  It's the first time many of them have used a touch screen, so I crouch and lean over their shoulder squinting in the lunchtime sun. (Here at the equator, the optimal time to use any device outside.)  I am conducting an experiment, which I'm told is very risky for a thesis methodology in the social sciences.  With five devices in this scheme, the chances of suffering 'technical difficulties' are terribly high.  Not to mention the chances of me just dropping one of them in the dirt.  The experiment is simple; the hardware is the challenge.

The who
the what
the how of it all.

Ideally, my participants are individuals who are engaged in their community in such a way that they might want to gather information or make reports to address a problem, take action, or make a policy.  The kind of people who might use these mobile applications in the future.  This could include ngo staff, social workers, local government staff, community development project members, and ICT students.  In addition, they are bi-lingual in English and Acholi (although they may speak other languages), but they have not spent more than 6 months outside of Uganda which may contribute to acculturation.  Age was not a primary consideration, but an effort was made to speak with an equal number of men and women.

It's quick and painless.
I show a one-minute video, then ask people to tell me what they've seen.  First, in Acholi, then through a series of questions on a mobile device (also in Acholi), and finally a third time in English.  I chose the video to approximate a scene of conflict because I am interested in how software applications can ultimately be adapted as a tool in conflict resolution.  The video is not particularly violent, but it shows a scuffle on a crowded street involving many people.  It is unclear to the viewer why the fight started, and this ambiguity means s/he must draw on schema, past experience, intuition to understand what has transpired. (This connects to my previous post about narrative structure and cultural variations in interpreting narratives surrounding events.)

I will be comparing the three versions using cognitive linguistics-- a combination of looking at thought and language focusing on how we decide (although perhaps not consciously) to articulate concepts when we have access to multiple languages in our minds.  I am looking for English concepts that appear in the Acholi versions-- especially the one produced with the mobile device.  This would be evidence of conceptual transfer and transfer can happen in either direction.  At the level before language is produced, even when the participant is intending to use Acholi, they might be engaging with English concepts.  My hypothesis is that the mobile technology triggers this engagement.  Changing the interface language to Acholi isn't enough, participants will convey an anglicized narrative when using this ICT.  By doing an experiment I can measure the instances of transfer.  Of course there might not be any or they could occur in a manner I don't expect.  That's the fun and the risk of doing an experiment.

It is only an initial step to describe in quantitative terms the limitations of current technology to capture narratives in languages that are extremely different from the languages that the software was designed for.  And also point towards avenues for addressing these limitations.  That is the next step, to contemplate what it could look like... a step for software engineers here to consider. 

Why does this matter?
There is already research that looks at how we remember events differently in different languages.  We connect these memories to sensory and emotional information through language.  If we are forced to recall events in another language, the narrative we give may be different.  If you stack up enough of those altered narratives as reports about a conflict, a crime, a human rights abuse, the final picture will compound these distortions.  What would happen if the narrative could be collected in another way?  A method that does not yet exist, but that reflects not only the language of the individuals sharing narratives, but their notion of what constitutes information worth collecting, and still further, how to piece that information together, how to organize it. An alternative to the current logic governing ICTs.

Beyond uses for conflict management, indigenous software that boosted regional or domestic use would be economically significant in places like Uganda.  If it organized information in a way that was incompatible to current software, i.e., the new couldn't immediately talk to the old, then that creates a market for still more software to facilitate interoperability when it needed to happen.  And purposeful inaccessibility could make systems more secure.

But this is all a long way off.  This is the potential, the reason I find the topic so interesting.  My immediate concern is contextualizing results such as when an individual with a degree in computer science and a job in crowdsoucred ICT work prefers to have me type the text on the mobile device or to speak their answers rather than write them in the provided textbox.  Is user-repellent the opposite of user-friendly?