ARL and Howard University join forces to examine social meanings central to Soldiers' situational awareness

September 09, 2014

By Jenna Brady, ARL Public Affairs

Story Highlights

  • ARL researchers recently teamed up with Howard University sociolinguists and computer scientists to probe how sociolinguistic cues in ordinary conversation can reveal social meanings potentially important to Soldiers' situational awareness.
  • These researchers came to ARL through an ARO-funded Howard University project to study code-switching and other linguistic cues for the extraction of social content from African Languages and English.

U.S. Army Research Laboratory social, computer and language scientists from the Computational and Information Sciences Directorate recently teamed up with Howard University sociolinguists and computer scientists at Aberdeen Proving Ground in Maryland to probe how sociolinguistic cues in ordinary conversation can reveal social meanings potentially important to Soldiers' situational awareness.

Sociolinguistic lead for the Howard University team was Dr. Nkonko Kamwangamalu, a professor of linguistics and an expert on code-switching. Code-switching is when a speaker changes dialect and languages in the middle of a segment of discourse.

Code-switching, also referred to as CS, can be a subtle but rich indicator of social intent, whether to signal group identity, to seek solidarity, or to bid for power, often without the speaker's awareness.

Completing the Howard University team were computer scientists Drs. Mohamed Chouikha and Robert Rwebangira Mugizi and sociolinguist Dr. Alla Tovares.

These researchers came to ARL through an Army Research Office-funded Howard University project to study CS and other linguistic cues for the extraction of social content from African Languages and English, or E-SCALES.

The E-SCALES project was initiated under the Partnerships in Research Transition program by Dr. John Lavery at ARO, working with Dr. Melissa Holland in ARL's Multilingual Computing and Analytics Branch, to address the need to integrate linguistics, social science and computation for analysis of less studied languages spoken in Africa.

Army-relevance of the E-SCALES project was made clear during a recent visit to ARL by United States Army Africa, or USARAF, which, since 2008, has been strengthening African land forces to create secure environments.

Soldiers who were present for the USARAF visit confirmed the significance of CS in remarks to the effect that Soldiers' jobs are rendered more difficult when local speakers slip out of a standard native language and into a regional dialect or patois.

The Soldiers noted that more important than the concrete meaning of the switched words was the motivation for the switch, especially when situations entailed issues of trust.

Dr. Elizabeth Bowman, who hosted the week-long exchange at ARL, introduced the idea that the social meaning of linguistic cues such as discourse markers and CS could be illuminated by computational social network analysis.

Working with Bowman, Dr. Michelle Vanni, Dr. Sue Kase, and John Dumer showed the Howard University team how text analytic tools being tested at ARL extract communication networks from social media and other online text and permit those networks to be visualized.

The Howard University researchers were quick to see the utility of this analysis, in that it provides extra-linguistic contextual information, unavailable in text alone yet contributing to its meaning.

Kamwangamalu explained that CS is a dynamic phenomenon whose social interpretation is often context-dependent. For example, he noted that one of his students, a 2013 ARL intern, "studied CS in postings to Facebook pages of African leaders, but soon found that the context available on Facebook was insufficient for assessing social meaning."

The Howard University team observed that social media activity visualized through networks can potentially fill some of the gaps in social context, such as who associates with a Person-of-Interest, as well as how central and active a PoI is in communicating.

Relational links in networks may help sociolinguists tie linguistic phenomena to social meanings. To probe this possibility, the group took first steps toward applying text analytics to data rich with CS and other sociolinguistic cues.

For a region in Kenya, ARL College Qualified Leaders interns collected Twitter data for a qualitative analysis by Howard University sociolinguists, and a concurrent computational analysis by tools that automatically extract social networks and relationships.

Bowman remarked that for ensuring accurate Soldier awareness in dynamic situations, "sociolinguistic analysis, combined with network visualization, holds promise."

At the conclusion of the exchange week, both ARL and visiting researchers were thrilled to have been a part of this first study.

"I am happy to report that the study was a resounding success, and our technical report is complete just four and a half weeks after we concluded the week-long study," stated Bowman.

"This experience not only allowed Howard University to claim a real collaboration from their key effort, but it demonstrates the extension of social network analysis and software to a sociolinguistic study and extends a research partnership to APG from Howard that previously only existed between Howard and ARL's Adelphi Laboratory Center."

 

Last Update / Reviewed: September 9, 2014