Showing posts with label Google. Show all posts
Showing posts with label Google. Show all posts

Saturday, 4 February 2012

Learning Indian scripts based on shape similarity

Originally published at http://indopersica.blogspot.com
Visit http://fon.gs/indopersica for more

An Introduction To Tamil Script – Reading & Writing” was the book that helped me learn to read Tamil. I first picked it up more than 8 years ago, and made good strides thanks to its rather novel approach, effectively being able to read Tamil fairly well within a month.

This book, authored by two of the most respected names in Indian linguistics – Debi Prasanna Pattanayak and MS Thirumalai – and originally published by the Central Institute of Indian Languages (CIIL) in 1980, uses a method we could call the ‘Shape Similarity Method’. Unlike the traditional method of teaching Indic alphabets – which involves teaching vowel letters followed by consonant letters, and then followed by the special forms of vowels when attached to consonants (vowel signs or mātrā) – this book starts off with letters that have the simplest shape, and slowly moves on to more complicated ones.

Also deviating from tradition, vowel signs / mātrā are taught before the individual vowel letters themselves. Here again, the books begins with the vowel signs that have the simplest shape and then goes on to the more complicated ones (such as signs that come before the consonant).

Renu Gupta in her paper “Initial literacy in Devanagari: What Matters to Learners” mentions that for members of the South Asian diaspora wishing to learn their heritage language, learning the script presents an additional hurdle. This is also true of Indians in India who would want (or need) to learn a script, for instance, the script used in the state or region they currently live in. On a related note, I constantly encounter a number of young people – almost all from urban areas, I should add – who are able to speak their mother tongue to a reasonable extent but are unable to read it well or at all.

In such a context, this shape-based learning methodology seems to be an extremely useful tool to help adult learners learn new scripts, for whom time is limited and similarities in shape can prove confusing if the script is taught the traditional way.

Though Gupta in her paper mentions that “[t]he effectiveness of the shape similarity method is not clear because the results have not been documented”, she goes on to say that “there may be some advantages in using this approach… [f]or literate adults”. I agree wholeheartedly with the latter statement, as I myself have been through this experience and can vouch for its effectiveness.

Unfortunately, it seems that CIIL hasn’t given Google Books permission to publish (even a part of) their wonderful publications on script learning, which is a pity, because the method used in these books is a highly practical and innovative one and has potential to help adults in the subcontinent become multi-scriptal instead of just multilingual – a huge practical benefit.

Here is a picture of the first page of the Tamil script book (I hope this isn’t a violation of copyright; it’s just one page!) –

IMG_20120204_195743

Thankfully, CIIL has taken at least some of their script-teaching material online. Their Kannada script page, which also uses the Script Similarity Method, is available at http://www.ciil-learnkannada.net/ccck/webpages/contents/script.htm (you might have to create a user account and log in, and also download special fonts – it’s really unfortunate that a body like the CIIL has still not moved to Unicode).

Saturday, 28 November 2009

Dogs Bastard


Originally published at http://indopersica.blogspot.com

Google has unveiled some new features for its Translate feature, including some for Hindi. Considering that they’ve introduced some amazing functionality over the last few months, including the Translator Toolkit, I decided to see how much their machine translation has improved.

Two of the new features include a) typing non-Roman-script languages in Roman script, which is then automatically transliterated into the correct script and b) translate-as-you-type.

So in went Dharmender Paaji’s trademark line. With translate-as-you-type, I got the following result – 

kutte 

Quite accurate, considering that it’s not easy for a machine to distinguish that कुत्ते /kʊt̪ːeː/ is not just the plural of कुत्ता /kʊt̪ːaː/ ‘dog’, but also the vocative case for this word.

I continued typing, and translate-as-you-type slowly came into its own –


kutte kamine

kutte kamine, main

kutte kamine, main tera khun

kutte kamine, main tera khun pi

kutte kamine, main tera khun pi jaoonga 

While probably an appropriate representation of how Dharmender Paaji himself would have translated his famous dialogue into English – he once called Bobby Deol ‘a piece of his heart’ (cf. जिगर का टुकड़ा /ʥiɡər kaː ʈʊkɽaː/) – what intrigued me was why the same word was translated ‘fucking’, then ‘motherfucker’, and eventually ‘bastard’.

Maybe it was Google Translate getting rid of some deep-rooted frustration. I felt sorry for it and made its task easier by improving the punctuation –

kutte, kamine, main tera khun pi jaoonga 

Comma after कुत्ते /kʊt̪ːeː/ gave improved results. But it was still lacking the required effect. In went the exclamation marks – 

kutte! kamine! main tera khun pi jaoonga! 
 
Why the exclamation marks cause ‘dog’ to become ‘dogs’, and ‘you bastard’ to ‘bastard’, I’m not too sure.

Playing around a little more, I got rid of an exclamation mark, which gave me –

kutte! kamine main tera khun pi jaoonga! 

I still haven’t been able to understand when कमीना /kəmiːnaː/ means ‘fucking’, and when ‘bastard’.


PS: For those who can’t read Hindi, here’s that immortal dialogue in the IPA -
kʊt̪ːeː ! kəmiːneː ! mɛ̃ː t̪eːraː xuːn piː ʥaːũːɡaː !
roughly meaning -
‘Dog! Wretch! I’m going to drink your blood!’

 
UA-12744356-2