Wish list of a language worker: Nikhil Mohan Pattnaik at FUEL GILT 2014: transcript

Key points

• His group of scientists began by making science learning reach beyond the classroom, with camps, workshops and a science magazine.
• In 2004 they set out to digitize a large early Odia dictionary; a commercial quote was far too high, so they built a cheap scanning setup and released a CD in 2006.
• They then digitized old Odia periodicals with a homemade camera setup and open source software, releasing seven DVDs with about 130,000 pages in 2008.
• Odia letters are rounded because of the writing materials used, and computers and typewriters were designed for one stroke per letter rather than Indian syllables.
• Commercial fonts were not interchangeable, few Unicode fonts exist, and there is a lot of old 8-bit data to convert.
• He asks the community for spell checking tools, working Indic support in Scribus and an OCR system for Odia, and hopes for a pan-Indian font.

Chapters

1. 0:00Overview and how the group began with science education
2. 2:13Digitizing a large Odia dictionary
3. 6:49Scanning old Odia periodicals
4. 8:55Odia script and technology problems
5. 12:24Fonts, Unicode and legacy data
6. 15:05Spell checking and publishing software requests
7. 16:53OCR gaps and a pan-Indian font

Full transcript

0:00And then, what are the problems we are facing, which are primarily from our experience with Buddha language and back to impartication work. What has been done or what we have done, we have gotten done with others' help to overcome some of these problems and then what remains that we would like to ask you, that's in part of our wish list.

0:31Very briefly, when I say we, we are a group of people with science background working in academic or research institutions and we decided to broaden our view as well as area of work. And one area that we wanted to touch and did touch was to make science, learning of science in your group and that should go beyond the classroom, that should go into the homes.

1:05So we devised a lot of packages, had no link with exams or classes, collected camps, groups, sit-ins on Sundays or holidays slowly led to many other things like teachers workshops. Printing of handbooks, a magazine, we ran a science magazine, we called it a magazine linked in education and science in society.

1:38And that was our entry point into developmental thoughts. It went on well, but slowly times cut up with us. exams, coaching, kept teacher fabulous doing very busy. The group, the first unit was found in ORIF, which unfortunately is not going anywhere. Hopefully we will try to revise that. Simultaneously, there was another interest of ours.

2:13When connected to the science articles, we saw how difficult it was to locate the whole And once we located it, gaining access to it and copying something out of it was really hard. So we decided instead of copying only the science articles, what would we try to copy the whole material? And that's how the whole thing about digitalization of the radicals started. And while writing the science articles for our magazine and books, we were using the dictionaries, various dictionaries, and bringing up our own word lists, everybody who used to write the articles, we insisted that they provide a list of words, either they have collected from other sources, glossaries, or coins themselves with a new note.

3:15And that these are the works slowly to cover our life to do, so to say. Another most important dictionary was produced in Odia in 1930s for almost 10,000 pages, 7 volumes. And it contains meanings in 12 languages, Odia in this extensive way. and meanings in Hindi and Dengali briefly.

3:47Kampaiver was very interesting. He was not a linguist or an employee. He was a lover. His life is very dangerous. He found them all murder himself by his lover. What this may be? Not only is that a book, it became very rare, because after his entire death, most of the copies were destroyed, because there were no writs, no dedication, lots of things.

4:19No, no story is really a mother. It contains nearly one lakh thirty thousand wrote words, who had been using. Right from the childhood, I have been seeing that set in my father's collection. Almost any literary book related to India in the 40s had a set. We were born and sort of in the 50s, whenever we were going to the school, we were always told this is one of the largest set of books in India, and that one had been over here.

4:54So whatever it is, it is almost sensible read. It contains no worse minutes. And he also had a reason. He says that I have given Hindi and English Bengali terms, so that we want to see how our words have affected the lettering Bengali and Bengali words. And how their words have affected us. And he regrets that I don't know Telugu, which is the other punctuation from his. But I would have been very interested to see the effect of Telugu on Uriah and vice versa. And so on. So we decided to work on it and wanted to digitize it in 2004.

5:43We looked around, shopped around and came up with a bill for 20 lakh rupees for a digital setup which was just unthinkable for us. We played with it, finally, what we can't say our localization, we did it with a scammer for 20,000 rupees, it took us a lot of manual processing, the scammer was a beautiful scammer to put it upside down, but it didn't run in the market. It was a discounted nota within a year.

6:17But we got that done. And within a year, in 2006, December, we released an update on a CD. And because of the book itself was important, the whole idea got on. And the something digital but the pneumonia was under the cough. That's right. Hope these game hits gave our work much more importance than probably the user. Frankly, even today we will talk about it.

6:49But, that launched us entirely through the direction. After this was done, so what we do? We went back to the periodical. Kodiya's periodical publishing started in 20th, 1849, but the copies of the first one just don't exist. Second one a few copies are there, third one is again lost. But serious work started in 1866. It was a nationalistic movement because of other reasons. Kodiya was under severe growth and British were sympathetic, but the intermedia series never really informed the problem. It was started and very few copies like this. So once we did, again another localization, we built a homemade setup with an ordinary Canon camera, which at that time had given us the

7:54the capability of computer then broke And they are very, fairly inexpensive software, free systems, which gives us total control from a laptop. And the working again was only within about 25,000 rupees. And what we use for processing is a very, very small open source software called Scanteller, which is extremely popular. And compare that to the commercial software which was offered for two microbeas.

8:25There is a lot of manual processing, but within six years, in 2008, we released seven DVDs, 24 GB of data total, containing almost all available periodicals up to 1850. One lakh thirty thousand pages, and set up.

8:55What I am here to talk about, what I am running out of time, my talk is about, as you know I am old days, to various forms, depending on the material I have done. Strong writing, but also earliest forms in India, particularly in Odisha, we have. All kinds of writings from Rashopar, Tharapur and an editor in Hanstron. But as you know, it's a specialised job. You really have to hammer it everywhere. But things become simpler.

9:28Anyone who could draw, could write. And get to their own form on parties. This really not drawing, it's almost scripted. And the earliest alphabets on the left-hand column, This is fairly extraordinary, as shown on the right side, rounded letters are more recent forms. One thing interesting at that period of time, not only for India, but most languages, technology and material iterated the language development.

10:06As you notice, all our letters are rounded. There is not a single horizontal stroke as in Hindi. And this is not only for Odia, Odia and Downsouth. They are not known, they all have horizontal strokes. There is only a duration book. We use boundaries. And the moment you put a horizontal stroke on boundaries, it splits. Some naturally go to Namibia, just like the English letters.

10:40It had angles because of the stone's artery. So it's interesting and I'll come back to it, why it's more interesting enough. Slow. Slow with the writing evolved. But during the evolution, it was not localized anymore. Bad writers, computers were Russian brothers, designed for one stroke, one letter, not Indian. which are basically, we said, syllables and require more than the availability of course for the full representation.

11:22Now, we couldn't change the language anymore. Earlier we could change the script and the language added to the technology. Now the technology has been thrust upon us and the language has matured It's not possible to change it anymore. So what we need is to start adapting the technology. And it's interesting that Indian and West are grouped among complex languages. Now the Chinese and Goryambul Chinese should have been more complex.

11:53Yet, fortunately a lot of things got done. I don't know how much The military force was there in the 8th British era, it was still probably developing. But when we got a fair amount of work done, letterpress became obsolete. And totaling of type driving was replaced by basically computer phones.

12:24There were a large number of phones and input engines available. But every commercial vendor protected their turf. Nothing was interchangeable. Srinipi, Atruti and so on. They deliberately, probably, changed the port points a little bit. So, one thought doesn't match with another one. Whatever it is, pages, that would be printed out. Something was not possible. It was not good for data basically. Not much data in capability.

12:56We have very, very large dictionaries like the one we talked about. It was in the earliest Indian dictionary, they were written for poets, but the ending letter matched, ending sounds matched, not alphabetically the first letter, but then much letter, the first one being 18 and 11 is in Korea, a large variety of fonts.

13:28We use GIST, so these are GIST forms mostly in the office or ISM, but down below are the only two ODA fonts widely available, there are a few others available in UNICOL. So when we tell people to choose over to UNICOL, because of other capabilities, particularly database, the first objection is, first thing another thing I have come to, first objection is there is no variety. And this is what is going on in the other room, I hope will solve part of the problem.

14:02This will be for technology, eye-catching parts. Still, more urgently, input methods are now more or less standardized and things are going on okay. Developing computers have been a problem. We have huge amounts of 8-bit data. We can't just dump it. So what we need is to download it to the recourse.

14:34And fortunately, it's a lot of technology worked into that. Relatively. Some of them are being recorded and we hope in the next coming months we can have a full set, at least for the major. Great of selectively, approved each and each. Dictionaries are being actively produced. We ourselves have prepared a strabic package, but strabic is sick. It hasn't been seen many development for four years.

15:05Besides, people are not very confident of installing another application and using it. So, now we are requesting the community to help us go to move our data list into something like no-spell software checking and prepare stand-alone externalities for those who want to use it. Next comes base level packages. Everybody knows Bacemakers. Omocor Express can only be a Bacemaker telephone.

15:37It's brand beautiful with AP funds. Most of the publishing, even today, is done with PagesMaker and Adob. But PagesMaker doesn't run even on Windows 7 and doesn't recognize the universe at all. Fortunately, Adob's InDesign MCS-6 is fully Indie compliant. And that you have been producing hard boostfully in design, that is, you have to index a capability on the foreign languages you have to manipulate it in several software.

16:17Scribus is there. In 2012, the ATPs in Kerala announced that they have prepared an ending patch. It's available, but we have not been successful in installing it and running it well. There are some inbox still. And now the problem is, it's not going to run on anything other than Ubuntu 12.04. And now we have to travel to Ubuntu 14.04. Fedora, they don't have a, hasn't been tried or Manoj had some problems getting into them.

16:53So this is an intermediate problem, we would really seek help from the community if you want to convince others to switch over to your record. And what we call the last frontier in the OCR, we have a huge amount of belief documents, all image documents, All our own magazines, several books, hundreds of books we have already reached out, it has not been made public because of annotations and other things. But it will be.

17:24Buddha Pramush here we cannot really... Appreciate what is in there, until we produce a printed form of selected things. But as things are now, it's a very, very, very distant thing. In Oregon practically there has been no development, although there have been several claims. So much so, the PDF language city, Cororia, has a folder collusion, but it's blank, empty folder.

17:55They will put something in, they put the folder, but nothing was there. What do we do? The community. Because established vendors are not interested, because the moment they develop unicode forms and release it, sell it to one person, it's lost. We can just like English in anywhere. So they are not interested in religion, unicode forms. Even Siddhartha is not interested, please don't.

18:26Sakal Bharati we had great expectations from, no, we have used Sakal Bharati, and it will be great if Sakal Bharati is really improved, because that is beautiful that it has all forms to write Indian languages, and we are suggesting that to Lohit, make a pan-Indian book, rather than Lohit would realize Lohit Hindi and all that. So that way, at least we use Hindi and Oregon. Somebody may use the Gali. You don't have to go through multiple installations and all that. And as for us, this is our plan.

Source and licence

Recording: File:Nikhil Mohan Pattnaik speaking during FUEL GILT Conference 2014, Pune.webm by Subhashish Panigrahi (2014-12-09), via Wikimedia Commons. Licence: CC BY-SA 4.0. The transcript, summary and chapters on this page are shared under the same licence (CC BY-SA 4.0).

The transcript is machine speech-to-text by the TidyTools Audio & video transcriber; the summary, key points and chapters were written by Claude from that transcript. Both may contain errors: check the recording before quoting it.