Wednesday, 14 September 2022

NDU press announcement: Academics vs. Aliens

 Academics vs. Aliens: Selected Essays on Social Science Research, Defense Education, and the Power of Partnerships > National Defense University Press > News Article View (ndu.edu)

Our book is out on the web!  Download your copy now.

Academics vs. Aliens: Selected Essays on Social Science Research, Defense Education, and the Power of Partnerships

edited by Gwyneth Sutherlin

Defense education has undeniably been at the heart of the Minerva research project, Defense Education Civilian University Research (DECUR) Partnership. Minerva, an organization focused on basic social science research, launched this pilot program in 2019 with the aim to:

Improve knowledge brought into the classroom; increase professional military education (PME) institutions’ ability to inform and help contribute to civilian social science dialogue; encourage and facilitate better connections across PME institutions for those experts with complementary intellectual and research interests; enhance civilian scientist awareness of critical social science challenges confronting the DOD; and encourage civilian academics to collaborate and engage with PME instructors and their students at PME institutions.1

The approach we took for the Understanding Chinese Influence project is called problem-based learning, which places the students at the center of the research. This framework allows practitioners to keep rigorously testing their ideas as their knowledge evolves. Learning in this manner with a host of partners who are experts in many fields is thrilling, but it is also challenging. Everyone speaks a different jargon and wants to pull the project in a different direction.

The book is divided into two sections of student essays that discuss the main goals of the program: the roles of partnership and social science education in PME. Each section is introduced by a reflection from one of our esteemed partners who worked on the project over the past two years. Eleven masters’ students contributed essays informed by their experience that comment on the broader topics of scientific innovation through collaboration, the role of social science research for national security, and how they would like to see PME take creative advantage of programs like Minerva DECUR.

Note

1 “DECUR Partnership,” Minerva Research Initiative, available at <https://minerva.defense.gov/Programs/DECUR-Partnership>.

Monday, 4 February 2019

How to make a dialect coach for a machine...and where we need it.

For a few years, I have wanted to start a guerrilla book club of sorts. We would send novels and poetry to software designers in the hope of introducing them to new cultures and languages (human languages).  This relies on them to extrapolate from the authors' artistic methods to articulate the human experience and use a sort of mimesis to create new, nuanced algorithms.

Current projects in ML/AI are trying to understand how and why humans are persuaded, why they trust, blame, coalesce and fragment around ideas or leaders, and then follow these patterns of behavior as though they were governed by rules we might one day anticipate.  We have a long way to go before any of this is possible.  In the meantime . . .

One of the most measurable things that writers have done that is transferable to current problems in NLP and ML involves using dialog with dialect.  When Mark Twain created characters that spoke distinctly American versions of English, how did readers recognize and place these local variations?  How do Japanese readers of Murakami recognize the Kansai dialect?  How do we instantly recognize that a character is from a certain region, city, class, or sub-culture?  The author purposefully and methodically transliterates the sound qualities of that language variant onto the page, and we are able to 'hear' it because we share the same concept, the same expectations for sounds, word choice, and phrasing to indicate a particular sub-group. If an author can plan and execute a dialect, and we can recognize it, then there must be rules. If there are rules, then we can teach a machine to find and follow them.  I am optimistic about this method for improving the acuity of NLP and sub-culture recognition by ingesting and parsing novels with strong dialects in their dialog.  This has been in the works while blog has been 'sleeping.' Updates soon on the research.

This approach to NLP is not as whimsical as it may seem.  A January 25th New York Times article illustrates how applied research in this area, prompted by observing and investigating issues of language and technology, could be beneficial.  In, Speaking Black Dialect in Courtrooms Can Have Striking Consequences, John Eligon wrote that:


"Researchers played audio recordings of a series of sentences spoken in African-American English and asked 27 stenographers who work in courthouses in Philadelphia to transcribe them. On average, the reporters made errors in two out of every five sentences, according to the study. 
The findings could have far-reaching consequences, as errors or misinterpretations in courtroom transcripts can influence the official court record in ways that are harmful to defendants, researchers and lawyers said."
To repeat the finding: 40% of the sentences have errors!?!

Sentencing and other judicial decisions will be based on the ingest and training of from court documents including transcripts.  We assume these documents are reasonably error free or at least not containing errors that reflect substantially different meanings than a witness or defendant intended.  If a dialect makes this much of a difference, improving how machines can be trained around dialect patterns will certainly be useful.  This is a great example of social science led ML design and evaluation for applied research.  (And maybe a little humanities if the book club takes off.)

Friday, 31 July 2015

Community Policing: at home and abroad



on ANP
 
In this ongoing series on quant peace research, in addition to examining international conflict issues, I want to explore the domestic issue of police militarization.  And due to the fact that policing has been one of the two most successful drivers curtailing terrorist activity in the past (not military action), it seems critical to evaluate how current policing can advance counterterrorism aims or at least partner with local communities affected by the war on terror in a constructive manner. (The other component has been politicalization fyi.)

In considering current US policing, why has a strategy that asserts a military presence or lethal force too frequently won out over one which emphasizes negotiation and community policing? While the jarring and violent negative consequences of these tactics have been receiving increased media attention, the focus of this research is on what can be learned from analyzing other challenging environments that have struggled with the implementation of community policing within a militarized context.  Based on discussions with colleagues who have operational experience in several conflict and post-conflict policing contexts (Nicaragua, Sudan, Kenya, and Afghanistan), I am gathering and synthesizing information about how their experiences are similar and/or different to my local police in terms of training and their day-to-day environment.

The police force in each case study is confronted by their own challenges, environments, and populations; however, even small details and insights can generate improved security.  

In the case of the US, two key goals could be:
1.  Addressing the domestic repercussions of our own post-war status such as returning veterans that may have PTSD or other issues that could involve emergency services (imagine the police called to a situation where a veteran is having a metal health crisis but also armed… is this time for SWAT or negotiation with police and health care workers?)  

In the case of my local environment, 1 in 16 members of the metro population was a veteran according to 2010 census data.  (Imagine your own household and neighbors to either side... is that close to 16 individuals?)


















2.  In relation to terrorism dissolution (tbd in a subsequent post), one of the two things that has been proven successful in combating terrorism is community policing... as opposed to militarized policing.  Is this emerging challenge integrated in training or a broader mission strategy?  How about in other countries that are currently at war as well as dealing with terrorism?  How does each case study's approach support ongoing and future counterterrorism efforts?


Several of these countries are dealing with violent conflict and terrorist threats or, as in the case of Nicaragua, are post-war environments, and exemplify a system that was developed with violence and crime prevention in mind.  (How is crime prevention measured you might ask.   Good question and one of the many things I’m reading about.)  Most of the case study countries have enlisted international consultants to train their police forces.  In Nicaragua, for example, they have had a training partnership with Sweden.  In Afghanistan, the US is often enlisted as the international consultant; however, the strategies the US helped develop for the ANP are markedly different from those the US pursues at home .  There might be an obvious answer to why those differences exists, but it is valuable to reflect on what purpose they serve and how successful they are.  

Basically, mountains of reports and government data to read, potential field visits, and some exciting synthesis potential for quant peace analysis! (What did you do on your summer vacation?)

To begin with some stats about the local environment that police are engaging with here in Indianapolis, IN
                                                                    City                               State






Population,2010
     820,445
6,483,802



Veterans, 2009-2013
50,432
455,105









White alone
58.6%
     81.5%




Black or African American
27.5%
9.1%


Asian
2.1%
1.6%


Hispanic or Latino*
9.4%
6.0%

* Hispanics may be of any race, so also are included in applicable race categories. 









Language other than English spoken at home, % age 5+,   2009-2013
12.6%
8.2%  


Bachelor's degree or higher, percent of persons age 25+, 2009-2013
27.3%
23.2%
Persons below poverty level, percent, 2009-2013
20.9%
 15.4%

What can we infer from any of these numbers?  How does this environment and population compare with the other case studies?  More to come in the next posts....

Monday, 4 May 2015

Quant Peace Research Part 2: Secret (or Unrecognized) Agents


This is the second post on quantitative peace and conflict analysis.  Here, I want to focus on agents and agency or the actors who have the power in key situations.  My examples draw primarily from recent research and from a paper under development on cultural concepts of culpability, agency, doubt, and justice in the context of post-conflict reconciliation.  I just delivered the paper at a conference on Human Welfare in Conflict at Oxford's Green Templeton College and got some terrific feedback from audience members, particularly from Somaliland, the Philippines, and the Ukraine who all basically said it was 'mindblowing.' (I think that's good?) 

Driving questions: Does the ICT format presume a western concept of agency or perpetrator/victim relationship in the underlying narrative structure governing the interface and information management design?  Does the format of the ICT homogenize non-western concepts of culpability in order to fit within its narrative constraints?  How can we interpret data collected with this design flaw?   
This post continues to examine results from the experiment that compared three recalls by Acholi-English speakers who had watched a video of a slightly violent, but mostly chaotic street scene.  In terms of ascribing agency-- who was responsible or culpable for the violence and chaos-- the experimental model brought into relief the contrast between the Acholi conceptualization of agent and that which was inherently designed into the ICT format.  The result of the experiment was that participants tended to frame the event as a fight involving several people during their initial oral Acholi recall, but during their recall via ICT, they often shifted the agency from a group to an individual.  

This shift occurred in two ways.  First, there was a change from 'they' to 'he'.   Second, the selection of schema relied to some extent on the participation of bystanders.  The role of bystanders, or the group/relational conceptualization of agency around the main action dissolves in many cases in the ICT format.  This aspect deserves more study.  It is possible that there was a narrative structure (the phrasal order or connection) that was interrupted which integrated the role of bystanders into other concepts within the narrative.  For example, if two interconnected concepts are separated when interviewing a witness, sometimes the witness will become confused about context and give a misleading statement.  In this way, the ICT format did not anticipate the necessity to link the role of bystanders with event framing (although there were closed an open questions which addressed the concept). 

Another issue examined here involves the response to Question 4: Who was the Attacker?  as well as the open format SMS responses related to the same frame.  The issue was that participants conflated the person being beaten (victim) with the person giving the beating (attacker).  This was due to their culturally learned schema, selected from cues such as the role of bystanders as well as the behavior of the person being beaten. 
Is the problem of identifying the attacker a conceptual transfer issue because Acholi’s object pronouns do not map conceptually into English pronouns?  This is a problem of categorization in which one language’s categories are more or less numerous than another’s and perhaps not governed by the same conceptual qualities.  (Think of English you vs. French tu/vous.) The results of the experiment revealed that the participants tended to frame the event in stage 1 in terms which did not focus on a specific actor as the primary agent; however, the event conceptualization supporting the ICT-formatted questions presupposes a responsible or culpable agent.  The stage 1 language, specifically the pronouns and the subject carried within the verbs, proved to be difficult for some participants to navigate in English during stage 3.  How did this conceptual challenge come across in stage 2 when the participants recalled their narratives via mobile technology format?.... First, a bit more about the concept of 'agent.'
Boroditsky (2010) explains the value of looking at the role of the ‘agent’ in multiple languages.  This has not been considered for Nilotic languages such as Acholi.  She describes:
In studies conducted by Caitlin Fausey at Stanford, speakers of English, Spanish and Japanese watched videos of two people popping balloons, breaking eggs and spilling drinks either intentionally or accidentally. Later everyone got a surprise memory test: For each event, can you remember who did it? She discovered a striking cross-linguistic difference in eyewitness memory. Spanish and Japanese speakers did not remember the agents of accidental events as well as did English speakers. Mind you, they remembered the agents of intentional events (for which their language would mention the agent) just fine. But for accidental events, when one wouldn't normally mention the agent in Spanish or Japanese, they didn't encode or remember the agent as well.
English emphasizes the agent when events are intentional.  When the presence of an agent appears in the Acholi-ICT, this was most likely evidence of conceptual transfer.  Participants frequently struggled to identify the perpetrator with any specificity.  They answered simply lacoo//a guy or dano//a person (participants 3, 9, and 24); lakwo// the thief (Participant 6); gipol// the many (Participant 28); and dano ma labal// the trouble causer (Participant 5).  These were difficult to convert conceptually into English in stage 3.  Principally, they did not fit easily within the concept of the ICT information structure which anticipates more specific identification of an agent or agents; the categories available within the application’s design anticipate a certain structure of information built around cause and effect (the same as in English).  The Acholi narrative does not structure the interaction between all the individuals in the video with the same causal connections or with the same notion of culpability.  This was reflected in their choice of schema and subsequently, in their language choices. 
Agent or agency has another meaning.  Barlas and Obhi (2013) offered a definition from neuroscience that captures the essence of what social scientists investigate.  They asserted, “The sense of agency is an intriguing aspect of human consciousness and is commonly defined as the sense that one is the author of their own actions and their consequences.”  Barlas and Obhi (2013, abstract) conducted a study in which they, “. . . varied the number of action alternatives (one, three, seven) that participants could select from and determined the effects on intentional binding which is believed to index the low-level sense of agency.” (ibid) From a neuroscience perspective, these researchers investigated the effects of choice on sense of agency.  Participants’ sense of agency increased when the number of choices they were presented with increased.  This has implications for ICT design used in the context of democratic participation.  It would seem that streamlining could potentially diminish the agency of individuals in the context of ICT use.  The goal of this research is to develop a new variable (cognitive/cultural) to increase the usability of ICTs for users outside the western context. 
Consider the following two examples which included an SMS and a smart phone path: In stage 1, only awobi//guy or young man was used initially then the concept of the person was integrated into the verbs and object pronouns in Acholi.  However, in stage 3, the English account required the speaker to make a choice in nearly every sentence to be more specific with regards to the subject and object.  This exposed the incongruence between the Acholi and English conceptual categories for persons.  In stage 2a and 2b, the Acholi concept of agent resisted the format suggested by the ICT information structure because there was no clear mental match.  This is similar to the category mapping issues described in section 3, for example, in Spanish there is a plural and singular form of ‘you’ that both map to one word in English.  This can cause pronoun problems for English speakers in Spanish.  For Acholi, the actor and indirect object are connected to the verb.  When an Acholi speaker must identify the actor and object in a phrase each time, he or she may reach for synonyms that roughly map to the concept in Acholi, but it is not a one-to-one concept match just as when an English speaker wants to say ‘you’ in Spanish.
Participant 19
Stage 1: Gin ma aneno. Aneno awobi ma nen calo okwalo gin mo, so tye ka ngweci then dano obino gitye ka goyo ne ey kitye ka penyo ne pingo okwanyo gin eno ni, so, dano madwong obino opong ikome ki dano tye ka dongo ne gitye ka goyo ne ki mayo ngo ma onongo tye icinge ma en okwanyo, yeah so en bene obedo ka ngwec tye ka ringo cen but dano tye ka lubo kore madwong kitye kawoti dongo ne ki goyo ne.//Things that I saw. I saw a guy who maybe stole something, so he was running then people came they were beating him ey they were asking him why he stole that very thing, so, many people came gathered around him and people were beating him they were hitting him they removed what was found in his hand that he had taken, yeah so also he was running away but many people are following him they are at the same time hitting him and beating him.
Stage 2b:
1.    Ineno_____? mony, kwo, laro lok// You saw____? Fight, theft, argument
Kwo// theft
2.    Ingeyo nining? //How do you know?
Aneno dano tye ka goyo ne.// I see people are hitting him
3.    Dano adi ma obedo iye?  2, 3-4, pol kato 4 // How many people were there?
4+
4.    Nga ma obedo lamony dano? Which person was the attacker?
Dano ma lakwo.//The person who is the thief
5.    Cwinyi tek i kom lamgam eni? Cwinya tek adida, cwinya tek, cwinya pe tek tutwal// Do you feel sure about this answer?
Cwinya pe tek tutwal //I don’t feel sure at all
Stage 3: Yeah, from the beginning, I saw there was a boy and some gentleman, so that gentleman started slapping that guy like trying to fight him, and I think the guy had stolen something, so that boy was trying to run away from that guy, but many people are joining, and they also started beating that guy, and like to without even finding out what that guy had done so they just joined like a mob justice, they joined and the boy wanted to leave, to go but like to run away but they followed him until he went when there were many people but those people they joined him and there were also trying, and he went because he picked something, there was something in his hand, so they were trying to remove, one person was trying to remove that thing from the boy’s hand, and others were just trying to,  slapped him, hitting him, like that.
In English, there is a difference between the images of a guy running or a boy running from mob justice.  This would be an important distinction shading the narrative with the innocence of a child or the suspicion of a young man.  The choice of describing someone as ‘a gentleman’ or just ‘a guy’ also colors the recollection in a certain light.  This may seem like a matter of translation (a sometimes arbitrary but necessary choice between synonyms when converting between languages), but it is an indicator of an underlying conceptual category governing the speaker’s choices.  While the subject of this experiment was not translation, the mismatch of conceptual categories for ‘person’ in the role of agent was clear from the comparison of these stages.  If stage 2b had been a closed-question survey consisting of only tickbox options, would the participants’ answers have fit neatly?  Or are they conceptualizing the agent’s role by actions and by social or relational cues rather than by identifying responsible individuals?
The pronoun choices in stage 3 from Participant 22 illustrated this point.  He was challenged by pronouns changing between man, young fellow, and boy as well as using the formal ‘complainant’ to describe the opposing figure. During Stages 1 and 3 there was an expression of doubt and concern about the root cause of the situation both using a repetitive narrative structure.  Stages 1 and 3 matched in their event framing; however, stage 2a did not use ‘doubt’ words and very little repetition.  The concept of theft was hinted at but did not fully materialize as a concept frame.  The SMS version disrupted the conveyance of the social or relational construction of culpability and the description of agency, the framing of the event as a theft.  In stages 1 and 3, the oral versions allowed the full context of the crowd to be integrated in the recall.  In the abbreviated SMS version, the role of the crowd (the fact that no one intervened) was mentioned, but without cultural context, this phrase would not be accessible to a future algorithmic amalgamation of the text to extract the frame ‘theft’ as the participant had intended. 
Participant 22
Stage 1: Aneno dano gitye ka lweny aa videyo eni pe angeyo maber ngo mutime pien ki nyuta ma dano ocako lweny dong nen calo tele moni obedo tye ki kit ma aneno kama videyo eni otime iye, aneno calo tye kamongo ma tye i bus park onyo kama motoka dwong iye ci latin awobi moni matidi eni nen calo lakwo mukwalo gin mo ki i jeba pa lawote ci lawote eno ni dong, tye ka lweny ikome pi gamo jami ne ma en okwanyo ki i jeba, ento ki gum marac pol dano pe tye ka niang ngo ma tye ka time, ci inongo ni i cawa mongo laco ni dok ocung ki cen ngat ma pat aye tye ka dongo awobi ni ngat ma bene pe ngeyo ngo ma onongo tye ka time ci bene kit ma gin ne otum kwede pe wangeyo kono gucobo onyo lwenyi pud gi obi mede kede onyo ngo mutime pe wangeyo ki bene pe waneno laloc mo nyo ngat mo ma bino ka juk dano weng bino ka lweny.// I saw people fighting. Aaa this video, I don’t know well what happened because they showed me that people started fighting already maybe some loggerhead was there how I saw where the video happened, I saw as if it is somewhere at the bus park or where there are many cars, then some small young guy maybe a thief who stole something from the pocket of his peer, then that peer he is fighting with him to collect his things that he picked from his pocket, but unluckily, many people are not understanding what is happening, then you find that at some time that guy again stood from afar someone else is hitting, someone who doesn’t know what is happening, then also how the thing ended we don’t know,  could it be that they made up or the fight was still continued or what happened we don’t know and also we didn’t see any leader or any person who came to stop, all people came to fight.
Stage 2a: Aneno dano dong Kitye kalweny. Pe kinyutu lingo lweny man otimme. ento kama lweny man tye iye onyo bus park nyutu ni cente pa laco ni kiyutu. en tye kalweny kom dano man ento pol dano odonyo iye ata. Ngat mop e ojuku gi.// I saw people hitting they were fighting.  They were not showing why the fight happened but where the fight was or bus park showing a guy’s money they showed.  [but] people they were fighting around there but some people [hesitated] to stop it.  No one stopped them.
Stage 3: Ok based on the scene of that interview, it appears that the crime happened in a bus park where people are traveling. Now the video is so abrupt in how it comes, it doesn’t show us the preceding events or what happened.  You like basically see people fighting and then when you try to follow and to make sense of what could be happening then you get to realize that probably, this man had something in his pocket that the young fellow pocketed and ran away with.  So as he went to recover that thing, a lot of other people came to join in, but they did not know exactly what was going on.  Now instead of stopping the fight, some of them were actually joining the fight. At some point you see the main the main the main complainant or maybe the person trying to recover his thing, he’s even standing behind and it’s another person beating the boy, then all of a sudden again he takes over and people are shouting, but you also don’t see like maybe leaders or people stopping, nobody’s trying to find out really what happened. Yeah.
Subsequent posts will look at narrative construction of doubt/certainty and how it combines with agency to conceptualize culpability. 

References:
Barlas, O., 2013. Freedom, Choice, and the Sense of Agency. Frontiers in Human Neuroscience, 7:514. (online) http://www.frontiersin.org/Journal/10.3389/fnhum.2013.00514/full (Accessed 19 March 2014).
Boroditsky, L. 2010. Does Language Influence Culture? The Wall Street Journal, 23 July. (online) http://online.wsj.com/articles/SB10001424052748703467304575383131592767868  (Accessed 20 September 2014).  


Friday, 3 April 2015

Quant Peace Research Part 1: Whose truth/justice/reconciliation?


This is the first in a series of posts on quantitative peace research.  Relatively new and often relegated to the 'huh?' panel stream at conferences, quant research in peace and conflict studies has been gaining ground in the past decade for several reasons.  Besides imbuing the word peace with less fuzzy/idealistic qualities thereby making it easier to build policies, raise money, frame political action platforms, and fuel solutions to conflict (Although it is the source of conflict with qualitative purists.  I won't hide that I am pro mixed methods... a topic for another post), the most promising reason quant research is having an impact in the field is that studies are being conducted by researchers with some surprising backgrounds beyond the usual polisci/sociology/social science disciplines.  These individuals bring with them alien concepts from neuroscience and even field experience from actual military combat that make for a truly multidimensional approach.

Whose Truth/Justice/Reconciliation? is a short version of an upcoming talk at The 8th Human Welfare in Conflict Conference. 

While presenting a seminar on the findings from my doctoral research (looking at narrative distortions produced via mobile ICT applications by bilingual participants), someone made an interesting comment about the applicability of my research, and methodology in particular, to transitional justice contexts.  This paper is an opportunity to develop that idea further.

In very generalized terms, post-conflict peace processes known often as ‘truth and reconciliation’ programs have been criticized for imposing a framework of justice that is culturally mismatched to the participating population’s concepts of justice. (see, for example, Avruch, 2010)  In this paper, I focused on concepts of culpability and agency by providing a nuanced quantitative measure to distinguish culturally rooted concepts of justice, responsibility, and agency. (Boroditsky, 2010; Costa et al., 2014)  The novel methodology adapted from cognitive linguistics combined quantitative measures of conceptual frames surrounding doubt, agency, and event structure to describe the concept of culpability.  Results have the potential to enhance dimensionality for articulating complex processes such as justice and  reconciliation as well as discussing the efficacy of such post-conflict programs.

I began by looking at the influence of ICT in conflict contexts because it is inescapable.  Due to the increasing use of information and communication technology (ICT) applications in the fields of peacebuilding and conflict resolution for gathering human rights abuse reports, election monitoring, polling, violence reporting, and other conflict management data collection activities that inform policy-making and participatory governance, this research performed a bilingual experiment with methodology from cognitive linguistics in order to describe the problematic nature of the ICT used in the conflict management context. This study was the first to incorporate cognitive-level communication variations and preferences as design considerations in the context of conflict management. (a.k.a radical alien interdisciplinary research) 

I have written extensively in previous posts about what cognitive-level means, but a brief recap. Pulling in research from both cognitive psychology and linguistics that examines memory, thought patterns such as categorization, problem-solving, cause-effect relationships, concepts of time and space, and use of language, this methodology focused, in particular, on conceptualization.  Conceptualization often requires complex relational understandings of objects, persons, time, space, and events.  The type of concepts that interest me concern events such as those that might be reported during conflict such as violence at a polling station or other incidents recalled as narratives and collected with mobile ICT applications.  (These narratives, once aggregated, become data for policy makers indicating hotspots of violence, political unrest, economic need, or even health crises.)  In order to observe concepts (because I can't see thoughts), I observed 'conceptual frames.'  A frame is something that computer scientists refer to, cartoonists refer to, cognitive psychologists refer to.  It is a fragment like a subject or a predicate, a basic unit of cognitive capacity that describes a perception such as he vs. they or something falling vs. something rising.    

Building from earlier work I had done which examined the consequences of a language barrier for ICT in crisis contexts (post-earthquake Haiti 2010, Libya and Egypt 2011, and Somalia 2011/12) which asserted that:
This flawed application prevented the original contributors from interacting with the information directly related to their own life-threatening situation, and the information it amassed formed an unsound basis for decision-making by international actors…. (Sutherlin, 2013, p.1)
my doctoral research (as well as this new paper) pursued the idea that the conceptual structure underlying language—the ‘organizational logic’ that occurs at the cognitive or thought-level—remained problematic for participation with ICT tools and the power they can leverage for policy-making for use by local actors.  In order to investigate conceptual structures, this research adapted experiments from cognitive linguistics that provided a quantitative means to assess the communication of concepts.  

In the northern region of Uganda, Gulu district, Gulu town, 29 bilingual Acholi-English participants completed a three-stage experiment. (I know it doesn't sound like a lot but that's an average number of participants for this type of bilingual study.)  Participants viewed a YouTube video depicting a chaotic street brawl, and were then asked to describe what they had seen in three distinct narrative forms: oral Acholi, written Acholi on a mobile device, and oral English.  By comparing narrative construction and identifying concepts unique to certain narratives, the experiment looked at the level of thought before language, the cognitive level, and thus followed in the footsteps of earlier research in the field of cognitive linguistics that examined how concepts from one language can be observed to transfer into another.  

**A quick note about language/culture/cognition: because this was a bilingual experiment and the data was in the form of 'language' but the variables under investigation were cultural and cognitive variation, the two comparison languages used in the experiment should be considered exemplars of cultures with certain characteristics that have a high cognitive impact such as orality or how categories are used. The characteristics which differ form a really long list, but part of the reason these two languages are compared/contrasted is their linguistic and cultural distance to one another which brings the issues under investigation into relief.  (Linguistic distance was  proposed by Greenberg in 1956 and extended by Lieberson in 1964 and even has a Wikipedia entry so it has got to be pretty well established. It quantifies how different dialects and languages such as German and Dutch vary from one another.  Cultural distance is adapted from this idea.  So English represents the culture that produced the ICT application and Acholi represents the culture using the application for data collection/aggregation/policymaking in conflict contexts.)  To summarize, it's not an experiment about English vs. every other single language or any specific language at all; it's about variations in underlying thinking (preceding or accompanying language).  Because we can 'see' thinking, the experiment observes language production and makes inferences about cognition and the culture that influenced it.

During the analysis, I looked for evidence of English to transfer into Acholi due to the presence of ICT.  For example, in the ICT recall stage, although participants were reading in Acholi and writing in Acholi, the logic of the ICT application which had been designed (as nearly all software has been) with the logic of English in its core would trigger English concepts in participants bilingual brains.  Concepts from English would transfer into their Acholi narratives that would not normally appear in an Acholi narrative.  My hypothesis was that, in essence, the ICT format would prescribe the participants' narratives in a way that was not natural to Acholi; there would be distortions or dissonance.  In 3/4 of the cases this was true.  There was a narrative shift and not simply one attributable to speaking vs. writing because I was looking at specific schemata and narrative structure. (From speaking to writing you might change how you describe something, but you don't change the story.)   

Before conducting the experiment, I spent three months in the field.  I did intensive language immersion.  I had discussions with local university professors, hunted for literature to review (anthropology, literature, poetry, linguistics, narrative studies, psychology).  All so I could identify specific cultural schema and narrative patterns. Schemata (sing. schema) are cognitive shortcuts that our brains use to make sense of the immense amount of sensory information we take in.  They are made up of conceptual frames.  For example, you can recognize a dog in a fraction of a second out of the corner of your eye because it fits the model/shortcut/set of conceptual frames for that animal.  We rely on schemata in order to be more efficient with our mental energy as well as to make sense of unusual or new situations by slotting what we see/hear/etc., onto the scaffolding of existing a schema and proceeding with a 'best fit' guess.  By focusing on schemata, this connected the experimental results to culturally formed concepts and the level of thought rather than a discourse analysis on language.  Schemata are culturally informed in this way-- you are probably familiar with the adage, 'When you hear hoof beats think horses, not zebras.'  Does everyone everywhere think horses?  It may depend on place/culture.   The video prompt for the experiment was chaotic and shared some familiar characteristics (because it was a street scene in Nigeria and the market stalls and taxi stand looked similar to Uganda as well as YouTube having made Nigerian videos popular viewing across the continent); however, the unfamiliar language in the video and, again, the chaotic scene, made it likely that it would trigger in participants the reliance on their culturally learned schemata.  That was the idea anyway.

Among the key findings, the concepts of culpability (who was guilty) and agency (who was involved) emerged as unique between what was described via ICT and orally in Acholi.  Crucially, several participants claimed that one specific individual was to blame for the incident in their ICT recall while they had only described a group having possibly been involved in something during their initial Acholi oral recall.  In addition, several participants changed the very nature of the event between these two recalls.  If we imagine these reports as part of a police investigation, the initial set of oral reports seems to indicate no action is needed while the ICT reports point the finger at one man.  Troubling to say the least.

In conclusion, if cultural constructs such as justice, culpability, and agency are both consciously and unconsciously programmed into technology, then the ICT application is putting limitations on the narrative, perhaps even prescribing conceptual elements of narrative for something as vital and nuanced as justice.  If we imagine a field poll being taken about what form transitional justice should take, if technology is involved, even in the aggregation of narratives later, this could radically alter the results by altering authorship/intentionality/voice/participation.  In addition to this practical impact, the methodology I used (with or without the mediating factor of technology) could offer a deeper understanding of the core conceptualization of justice within a society by being able to break the concept down at a cognitive level.  Subsequent posts will continue to look at each of these concepts in more depth (culpability and agency) as well as build on comments/reactions to the paper.