The Pashto Language as a Learning Tool for STEM

In a part of Afghanistan and Pakistan where the communities speak Pashto, the region is in general at an economic disadvantage to neighboring communities. Women and girls are often at an even greater disadvantage in terms of access to science, technology, engineering, and mathematics (STEM) courses. These communities speak Pashto, which can create a language barrier for educational resources.

The proposed solution is a program where the language Pashto is a learning resource. The mother of a child instructs the child on the proper use of their language. The skillset for ensuring the precision of the local language can be a resource for learning STEM topics. This is accomplished by means of making analogies to the Pashto grammar to STEM topics.

 There are two key parts of Pashto:

Circumposition and Split-Ergativity

Circumposition In a variety of languages, phrases such as "from ages ago" is often used for a variety of phrases. In English, it is more rarely used. In machine learning applications such Simulated Annealing and Quantum Annealing, this use of prepositions before and after a word help explain the way the systems work.

The Split-Ergativity

The glass breaks, the ship sinks...These are examples of ergativity in English. The objects have almost a character of their own in the examples given here. In Pashto, the language is rather unique in that is has what is called split-ergativity...some sentences based on theme or "theta role" are ergative and some are not. This can help with a course in Machine Learning:

DATA PREPARATION: Loading the Data

   In order to begin a machine learning exercise, data needs to be loaded. "Loading data" is an example of a sentence that will translate into Pashto as an ergative sentence. Pashto-speaking women in a machine learning course can have the confidence to now that their language has a unique grammar for the computer science exercise.

DATA FROM PAST EVENTS

   The data in machine learning is from events that have already taken place. Events in the past tense is also a requisite for the Pashto ergative.

VOLITION

Not all datasets are of events where people decided to take an action. Datasets could be of unexpected events such as weather. However, datasets on actions taken help to forecast what effect from actions against global warming may have for the future and can be learning tool for predictive analytics. Volition or actions willfully taken by people is also another condition where the Pashto ergative is used.

All in all, the use of the Pashto language can give confidence and assist in teaching machine learning concepts that might otherwise be overwhelming. At the same time, it encourages the use of the native language and may encourage communities to support a solution such as this one.

THE DEEP NEURAL NETWORK

The ice melts as the rice cooks on the oven. What is seen is that the subjects in the sentence carry on an action. In the deep neural network (DNN), a layer passes information from a previous layer and passes it onto the next layer. If someone uses a language that uses ergativity, this process can be explained in the language.

-The use of context or deixis in the description of movement. Going back to layers, the context of when the layer passes information from a previous layer to the next layer would be an example of this. 

The goal is for the solution to provide STEM educational resources to the women and girls of the Pashto speaking communities in Afghanistan and Pakistan.

From having researched this topic, many parts of this solution appear to be unique, and this may be the first time such ideas in applications of topics in Pashto linguistics have been published.

I have created a free online course for explaining how machine learning can be taught by means of analogies to Pashto grammar and so on. It is essentially an existing online course for "teaching the teacher".

This idea is likely to be an innovative way to assist those in need of STEM educational services.

The team lead is applying language resources for a community service. Outside of this, the Team Lead is not presently connected to the community in which the project is based.

I learned of Solve from an ad when I was taking an online course at MIT Professional Development.

It used a language in an economically disadvantaged community as a learning tool for teaching subjects such as machine learning.

The impact goal is for women to have success in STEM courses such as machine learning after they have attended this proposed online program.

The measurement of the impact goals would be based on the grades of Pashto-speaking women in STEM courses like machine learning after they had attended this proposed program.

The expectation is that by introducing the Pashto language as a resource for women to attend STEM courses, the Afghanistan communities will encourage more women to attend these types of courses because, in part, it brings value to their local languages by means of how it can be a learning tool. 

  Essentially, pride in the language of the local culture could help the community encourage women to take STEM courses because it can apply the knowledge of their language.

The solution applies the technology of machine learning and related technologies in terms of it being a subject that will be introduced to students in an innovative way.

One person

For about two years

The program itself encourages diversity in that its goal is to assist women in Afghanistan to have access to STEM courses.

The existing business model is a free online course that could be developed by means of grants.

The goal would be to sustain the program by means of grants.