2004 July 06 Tuesday
US Military Developing Computer Game To Learn Language, Culture
Regular ParaPundit readers are aware that I consider it a big mistake for the US military to deploy soldiers to Afghanistan and Iraq with so little training in local language and culture. Well, the US military appears to be aware of the seriousness of this deficiency. A New York Times report has brought to my attention a research program by the Office of Naval Research (ONR) and the Defense Advanced Project Research Agency (DARPA) to build a local language and culture training game for the Special Operations Command of the US military.
Sergeant Smith is not a real soldier, but the leading character in a video game being developed at the University of Southern California's School of Engineering as a tool for teaching soldiers to speak Arabic. Both the game's environment and the characters who populate it have a high degree of realism, in an effort to simulate the kinds of situations troops will face in the Middle East. Talle is modeled on an actual Lebanese village, while the game's characters are driven by artificial-intelligence software that enables them to behave autonomously and react realistically to Sergeant Smith.
The Tactical Language Project, as it is called, is being developed at U.S.C.'s Center for Research in Technology for Education, in cooperation with the Special Operations Command. From July 12 to 16, real Special Forces soldiers at Fort Bragg in Northern California will test the game and put Sergeant Smith through his paces.
The characters in the game respond to the game-players actions by increasing or decreasing their trust in him.
One of their most critical beliefs is their trust level, Ms. Si said. If Sergeant Smith behaves appropriately, he will gain the characters' trust and they will help him; if not, he is likely to cause suspicion.
A USC press releases provides a lot more detail. (worth reading in full if this sounds interesting)
Part of the system, the “Mission Skill Builder,” resembles an intensive version of the language laboratory programs that have been in use for generations. in these students imitate and practice words and phrases pronounced by native speakers.
“While our system is similar to drill-and-practice language programs that have been in use for some time, the Skill Builder incorporates some important innovations,” Johnson said.
- speech recognition technology that is able to evaluate learner speech and detect common errors;
- pedagogical agent technology that provides the learner with tailored feedback on his performance; and
- a learner model that dynamically keeps track of what aspects of the language the learner has mastered and in what areas the learner is deficient.
The game sounds like it is structured in ways very much like conventional adventure games with the added complexities that the game player must be able to speak Arabic into language recognition software and the simulated agents are written by experts in artificial intelligence to embody a lot of Arabic culture in their values and behavior
The examination or application part of the training system, the "Mission Practice Environment," is still more innovative. It is designed to give students an unscripted, unpredictable, and therefore challenging test of their mastery of these elements.
In this segment, students wearing earphones and microphones control a uniformed figure moving through a Lebanese village, complete with outdoor coffee bar. They meet animated Arabic speakers, who (thanks to artificial intelligence driven voice recognition programs) can carry on free-form conversations.
"These AI figures can understand what the students say, if it's said correctly - or won't, if it isn't. And they will respond appropriately," said Johnson.
In the exercise, after exchanging greetings the student learns the names of locals, the name of the place, the identity of the local headman and the location of his house, and must follow these directions through the game interface to get there.
"In typical videogame fashion, the idea is to get to the next level," said Johnson. "In this game, in order to get to the next level, the learner has to master the linguistic skills."
The program already has features to adapt it to each individual user, noting consistent errors or difficulties, which can be targeted for extensive or remedial practice.
So far, researchers have completed approximately seven hours of the program. The full program will have about 80 hours of instruction, and introduce perhaps 500 carefully chosen words of the "Levantine" Arabic spoken in Lebanon to learners. If all goes as planned, the system may be deployed next year.
Computer automation is the future of education in general. Computers are cheap and their patience unlimited. Computer games that correct your errors and automatically record and report on progress are needed across a large range of domains of knowledge unrelated to the US military. This is a sign of things to come.
Great post Randall. Maybe the US Military will become a major force behind changing education in America. They definitely have a vested interest in public education.
In terms of national security, it is important to recall that adults cannot learn a foreign language properly. How many anti-communists in Russia and elsewhere were heedlessly sacrificed by American officials who believed that their agents could learn foreign languages so easily? A lot is riding on this fiction of language-learning having no biological limitations. The immigration policy ,in particular, is shown to be unsound, if adults are up against natural limitations of this kind. It is not a matter of lack of technology or effort; the brain must literally grow into a language which is learned in an unhandicapped way, and this can only happen with young children. Recognizing this fact of life also discredits the over-optimistic assimilationists, who imagine a spontaneous process of assimilation will happen with adults.
Adults can, however, learn the basic aspects of a language and culture, they can even learn them thoroughly. The aim of the game is to train *as quickly as possible* a number of soldiers *as large as possible* to deal with the new situations, and that is an achievable goal, and the technology is outstanding, even if the ordinary infantryman is never going to become an accomplished arabist.
If the hope is that these types of technologies will spill over into education generally, that prospect has factors working as much against it, as for it. The nature of methods which work by entraining auto-didacticism, is such that a small percentage on the high end get great benefit, while those in the lower half of the distribution of ability and motivation have to be prodded and dragged, just to keep from falling further behind. Yet the government looks at the relative performance between targeted populations, which have been chosen in order to maximize the possibilities of ethnic and racial hostility in the society. Therefore,as soon as they determine that a method will impart its benefits disproportionately to the high-performers, officials are inclined to militate against that method. An absolute improvement is worthless to them, if it causes the intercommunal gaps to broaden.
Can't they just whip up some pamphlets with "Prepare to Die" in all foreign lingos?
John, I see computer software, recorded lectures, and other techniques for automating education as ways to allow bright kids to escape the mediocrity (or worse) of so many public schools. The brights could learn much more rapidly but are kept advancing at slow rates by school systems that are focused more on the dumber kids.
On such priorities, there is a great deal to support high hopes for automation of education. Does today's technology allow for the proctor to be dispensed with in standardized testing, and with perfect assurance? If it does, this might allow accelerated certification of learning. Perhaps a foundation could develop programs which could be gone through seriatim, much more rapidly. The strongest suit of such automated systems might be their potential for thoroughly testing the learning which occurs. Regarding national security and knowledge of unusual languages; suppose that the government were to offer prizes for pre-teenagers to learn languages which are not spoken by more than 1% of the people here, and test them for their proficiency at several levels. It could be required that they be at high English-language knowledge levels, so as not to just throw money at those who already know an odd language from their family circumstances. Wouldn't there be a good response, if instructional programs were available which could be more or less self-taught, and even if the prizes were not rich ones?
Can't they just whip up some pamphlets with "Prepare to Die" in all foreign lingos?
you must mean this:
Automated instruction will only be better than live pedagogy if the essential concepts can somehow be caught, distilled and served up in a way which is grasped better by the students. Trying to cover all of the possible ways something can fail to be understood properly and address them to put the student on track - without any human interaction - is a much bigger job, requiring more and different skills.
This is the issue, I think. Those skills are bound to be in short supply for the simple reason that they are both more specialized and more varied, and the promise of automated instruction will have to wait on the time of the people who are capable of producing programs which work. In the mean time, programs produced by people merely trying to cash in will poison the well.
“Trying to cover all of the possible ways something can fail to be understood properly and address them to put the student on track - without any human interaction”
That is an excellent goal for automated learning. There are reading programs that diagnosis and correct common reading mistakes. I expect it wouldn’t be too hard to build similar programs for elementary arithmetic and mathematics. (Perhaps they already exist?)
However it sets the bar far too high for most automated learning. Even without modeling the student, what’s online today for free is far better than what I experienced in public schools and a private university.
I’m teaching myself Spanish using a free online site. The online tools are far better than the grade school Spanish, high school Latin, and college German teachers I had.
The online course materials in biology, microbiology, and genetics are excellent.
(With the exception of a physics teacher and a chemistry teacher, I felt my teachers limited what I learned. I don’t remember teachers individualizing instruction based on a student’s interests, learning rate, or misunderstandings. I do remember many rote lectures on material directly from the textbook.)
E-P, Learning software can be very valuable even if it is not be capable of detecting and reacting properly to every mode of failure to learn. Not every student will fail to learn a given lesson. Of those who do some will fail to comprehend in ways that are easy to compensate for with an additional automated explanation. If learning software works for some people some of the time it is quite valuable.
Keep in mind that some material is too difficult for less capable minds to grasp. No software can compensate for limits in cognitive ability.
Learning software combined with recorded lectures by the best teachers would give most students higher quality instruction and at the same time costs would be lowered and convenience would be greatly increased.
I had understood that children who were bilingual learned languages more readily as adults. My Romanian sons, who spoke Romanian and Hungarian in daily life and were taught English and French in school, understood Spanish very quickly, though neither developed any facility for speaking it. As understanding and speaking occur in somewhat different locations in the brain, perhaps this method of instruction will prove more useful with one.
Why do they use the term "learners" when they mean "students?" Can't those people speak English?