Evolution of Pashto dialects of PF Members

[Toramana]

Thank you so much Toramana for the information.You are right about the bioinformatics softwares approach.Basically the alignments were treated as protein sequences which are more diverse in terms of letters than the DNA sequnces (which are only A, T, G, C).

@gulalai

That (i.e., limited alphabet)might limit the efficiency of the software somewhat...Still, it is interesting how accurately did it categorize the words... As for introducing gaps, that is obvious looking at the limited alphabet employed to represent protien sequencing...One way to tackle this limitation would be to break words into syllables (smaller lexical units) and feed the software syllables rather (but then similarities accross syllable boundaries probably wouldn't be counted for)...

Anyhow, interesting...

@Omar Usman, these software are just symbol processing systems and are blind to the meaning of the data they process ---what they do is just to look for patterns/(sub) sequences in the character-sequences fed to them...

[Admin Khan]

Omar Usmana,
Last time we spoke, you said you were getting your Ph.D in Physics right?

[Nokia_Apridy]

Da Nar Bachay...........Mata nawa chay zmung Gulali 2.0 ba hem dumra takra

wi wakh wakh dar kha

Sallooooooooot khorakay..

[PFgulalai]

Thanks Jasmine, PashtunQueen, Sher Khan, Qrratu, and GoPashtunGo.

Jasminey, it took almost two hours-the class was quite boring so I thought of having some fun

GoPashtunGo,

Thanks for the interest but the problem is that I need to first make it scientifically plausible before I can even suggest/allow to use it.I hope you would understand =)

[Omar Usman]

Okay sorry for being out of sight a few days. I will come back to some of the posts later during the weekend.

I did a little home-work and came across this Masters Thesis for dialect comparison of Dutch. It is very relevant to what we want. Other works available are not that much refined.

http://www.martijnwieling.nl/files/dialects.pdf

They have explored different techniques, of which the Levenshtein distance is quicker and easy to implement. Most of the brief literature I looked at uses this (or a modified form) of this technique. For comparing results ... other simpler distance measures such as Euclidean or Manhattan can also be given.

The only downside is that comparison is made based on written representations of dialects. In the case of Pashto, our written script has little or no representation for the various dialects. At most a few forms of the Hard and Soft Pashto is represented in our script. To go about this, we would require usage of a transliteration scheme that is not based on the existing pashto script. Another option is speech processing ... but that is a different area and of course requires more time. By the way this problem (and a workaround) is discussed in the link.

Gulalai can you comment anything on the technique used by your software? Is it open-source?

[Amir al Ghaznavi]

gulalai khoray thank u for this post, what a great contribution

i just wanted to make one point thought

you have me listed under "Kakar" but I should be listed under "Ghazni" because unlike Afridis, etc Kakars are not limited to one area, and thus in relative isolation, have not developed their own dialect

a Kakar from Kandahar speaks Kandari pashto and a Kakar from Ghazni speaks differently

[Zmarak Khan]

Graane Gulalai,

You should change your name to Genius Gulalai or Intelligent Gulalai.
I am so overwhelmed after reading this.All of a sudden I am in love with your brain and diagrams.
I have worked with a number of bioinformatics softwares but I never thought of using them in a similar way.lol.Your thinking process here is really novel and interesting.
The issue of scientific validity is an important one but not at this point.You see, languages exist in patterns and elements, which makes it possible to understand, reconstruct, and recognize the properties of language via a computational process.This is just a layman's perspective.I am not an expert in Language and Linguistics though.
Wow!A Pakhtun female treating Pashto dialects as protein sequences is really shocking to me (in a good way though).By the way, do you read Noam Chomsky?
I do follow your posts as guest user and I have also forwarded them to my friends at times.I won't be surprised the day you will bring Nobel Prize in Medicine to Pakhtunistan
I just joined to say Thanks for this contribution. You are such a great source of inspiration.
Please keep posting and take good care.

P.S I will post my feedback later.

[pir_Rokhan]

Gulalai

This is pretty creative. I think that once we translate any science into mathematical forms using software technology it will eliminate a lot of ambiguities and give us an accurate picture. I had done some experimentation on this idea in my field using matlab and had designed a C.A.I for television system.

I wonder if you could transform this idea to an A.I based XS software which could give us accurate info about which our geneological tree entering some input data like the one for the diagnosis of diseases.