The "suborNo" Phonetic Transliteration scheme
The Phonetic Method used on supriyosen.net is called "suborNo". This Phonetic scheme was custom designed to be reflective of pronunciation and to be capable of two-way transliteration (so English to Bangla and Bangla to English both use the same character strings). This page describes how the transliteration tables were derived, and compares suborNo to the transliteration used by AVRO keyboard, a popular transliteration software.Why Do You Need a Transliteration Scheme
Most people use English keyboards to interact with computers. The standard keyboard has an assigned key for each English letter of the Alphabet.The sounds of letter of the Alphabet in Bangla are quite different from English. In many cases, more than one English letter may be required to correctly simulate the sound of a Bangla Alphabet. So how do you input Bangla letters into a computer with an English keyboard?
There are two approaches:
- Create a software "overlay" of the English Keyboard by assigning Bangla letters to different keys. For example, Apple ships a software keyboard with its Bangla fonts. The Key assignments are as follows.
Normal
With Shift pressed
As you can see, the user now has to memorize the positions of all the Bangla Letters. - Devise a Phonetic Transliteration table in which one or more English letters get transliterated to a single Bangla letter.
The "transliteration" is done through software and is transparent to the user. The user types in English letters, and they show up in the equivalent Bangla form. For example, say you set up a scheme in which you defined
- b='ব'
- a='া'
- ng='ং'
- l='ল'
- a='া'
The user does not have to memorize a new keyboard layout. The sound of the Bangla letters, entered in English, generate the Bangla letters on the computer.
The Phonetic scheme described here is based on the second alternative.
Evolution
We have to start somewhere. The most commonly used textbook for learning Bangla is Teach Yourself Bengali by William Radice, so let us start with the phonetic scheme used in this book. On page 53, a summary of the phonetic scheme is provided. This is reproduced below.
Radice’s book was typeset, so many symbols are used that do not appear on a modern computer keyboard. We want to make it easy to type Bangla using a standard keyboard found in Personal Computers (Windows, Apple and Linux) as well as on today’s mobile devices like smartphones and tablets. 
Assigning English Phonetic Letters
You will notice that Radice used some letters not found standard English keyboards. These are marked as '??' below in the table below.Vowels
অ ?? |
আ a |
ই i |
ঈ ?? |
উ u |
ঊ ?? |
ঋ ?? |
এ e |
ঐ ?? |
ও ?? |
ঔ ?? |
Consonants
ক k | খ kh | গ g | ঘ gh | ঙ ?? | চ ?was c? | ছ ?was ch? | জ j | ঝ jh | ঞ ?? |
ট ?? | ঠ ?? | ড ?? | ঢ ?? | ণ ?? | ত t | থ th | দ d | ধ dh | ন n |
প p | ফ ph | ব b | ভ bh | ম m | য ?? | র r | ল l | শ ?? | ষ ?? |
স s | হ h | য় ?? | ড় ?? | ঢ় ?? | ৎ ?? | ং ?? | ঃ ?? | ঁ ?? | ্ ?? |
Principles for Assigning Phonetic Representation
All assignments we have accepted from Radice's scheme are in lower case, and have 1 or 2 English letters representing the Bangla letters.-
Some letters in Bangla have similar pronunciation in modern usage but maintain distinct forms in spelling (e.g. the vowels ই and ঈ (hroshwo-i and dirgho-i), and the consonants ন and ণ (donten-naw and moddhen-naw))
Capital letters will be used to make the distinction, eg ন = 'n' and ণ = 'N' -
The last 5 consonants, sometimes called Consonant Signs, have an abrupt ending corresponding to the sound of other consonants (e.g. ৎ (khondo-taw) is an short, abrupt version of ত (letter-taw))
The character '^' will be used after the normal sound, eg chondro-bundu ' ঁ ' will be represented by 'n^' to maintain phonetic integrity
Assigning Phonetic representation
Vowel | Similar sound to | Assign |
---|---|---|
awe 'অ' | sounds like a short o | ‘o’ |
owe 'ও' | similar to ‘o’ | capitalized ‘O’ |
long i 'ঈ' | long ‘i’ sound | capitalized ‘I’ |
long u 'ঊ' | long ‘u’ sound | capitalized ‘U’ |
dipthong oi 'ঐ' | oi sound, but would conflict with 'o'+'i' as in বই (boi) | capitalized 'OI' |
dipthong ou 'ঔ' | ou sound, O already assigned to ‘ও' | capitalized ‘OU’ |
rri ‘ঋ' | rri sound | ‘rri’ |
Some authors have used 'ii', 'uu' and 'oo' to represent the long 'i', long 'u' and the vowel 'owe' respectively. I felt that words like দীর্ঘ (long), দূর (far) and দোলা (to swing) are better represented as 'dIrgho', 'dUr' and 'dOla' than 'diirgho', 'duur' and 'doola' when actually pronounced by an English speaker.
Consonant Sign | Similar sound to | Assign |
---|---|---|
khondo-taw 'ৎ' | sounds like a truncated taw as in হঠাৎ (suddenly) | assign ‘t^’, with ^ showing abrupt ending |
anuswar 'ং' | ng sound, commonly used in spelling বাংলা | ‘ng’ |
bisorgo 'ঃ' | abrupt h sound as in দুঃখ (sorrow) | ‘H^’ |
chondrobindu ' ঁ ' | nasal n sound as in চাঁদ (moon) | ‘n^’ |
hosonto ' ্' | stops pronunciation of inherent vowel after a consonant. In Unicode, hosonto is used to join two consonants eg 'ন' ্ 'ত'='ন্ত' |
backtick '`' |
Consonant | Similar sound to | Assign |
---|---|---|
ungaw 'ঙ' | Used to spell বাঙালি (Bengali), has a ng sound | ‘Ng’ |
ingaw 'ঞ' | ng sound, but ng and Ng already assigned | ‘NG’ |
chaw 'চ' | Assigned a 'c' by Radice. In most English words, c has a k sound as in cat. c has a "chaw" sound in a few cases like the Italian word "Ciao", and in the pronunciation of Radice which is pronounced as radichi. This problem also exists in the AVRO phonetic scheme. | ‘ch’ |
chhaw 'ছ' | an aspirated ch, was assigned 'ch' by Radice | 'chh' |
taw(hard) 'ট' | hard taw sound, must be consistent with taw(soft), already assigned 't' | ‘T’ |
ttaw 'ঠ' | aspirated version of previous letter, assihned 'T' | ‘Th’ |
daw(hard) 'ড' | must be consistent with daw(soft), already assigned 'd' | ‘D’ |
dhaw(hard) 'ঢ' | aspirated version of previos letter | ‘Dh’ |
moddenno-naw 'ণ' | same sound as donte-naw, already assigned 'n' | ‘N’ |
ontesto-jaw 'য' | same sound as borgiyo-jaw,already assigned 'j' | ‘J’ |
doye-sunno raw 'ড়' | same sound as raw, already assigned r. Associated with daw 'D' | ‘R’ |
ddoye-sunno raw 'ঢ়' | same sound as raw, already assigned r. Associated with dhaw 'Dh' | ‘Rh’ |
talibo-shaw 'শ' | same sound as donte-shaw, already assigned 's'. This letter is associated with the word শান্তি (peace) and is spelt in English translations as shanti. Taking common usage into account | ‘sh’ |
moddheno-shaw 'ষ' | similar sound to the donte-shaw, already assigned 's' and talibo-shaw, already assigned 'sh'. Used in many Sanskrit derived words like কৃষ্ণ (a Hindu God), and is usually translated as Krishna. Since, 'sh' is already assigned, we use a capitalized S | ‘Sh’ |
Additional, J in conjuncts | When used in consonant conjuncts, the sound of a trailing 'J' sounds like 'yaw', as in ব্যাথা (pain). 'y' is already assigned. | 'J,Y' |
Additional, baw in conjuncts | When used in consonant conjuncts, the sound of a trailing 'b' sounds like 'waw', as in God ইশ্বর (ishwor). | 'b,w' |
The resulting সুবর্ণ suborNo Transliteration scheme
The complete transliteration table looks like this.Vowels & Vowel Diacritics
Consonants and Consonant Signs
Using সুবর্ণ suborNo for Phonetic Transliteration
Each Bangla letter is represented by a sequence of 1 to 3 English letters, closely representing the phonetic sound of the Bangla letter. Note that upper and lower case English letters are used.Example: To write আমার নাম সুপ্রিয়, you would enter amar nam supriyo.
Usage Notes
- Automatic selection of Vowels or Vowel Signs (-kar) based on context
e.g. amar will be shown as আমার, with the correct form of the letter আ chosen - Automatic generation of Juktakkhor
e.g. pr produces প+র=প্র. - Virtual Vowel "o"
All Bangla Consonants end with the "awe" sound, as in "kawe" or "ko" for the Bangla consonant ক. The letter "o" can be thought of as a virtual letter, and can be used at the end of a word to make the phonetic word more aesthetic.
e.g. to produce কর, you can write "kor" or "koro". - Overriding Automatic Juktakkhor
To prevent automatic juktakkhor, insert an o between the Consonants.
e.g. pr would have produced the compound letter প্র. The letter o does not create a character but is used to break an automatic juktakkhor. If you want to write পর, you will want to break the automatic juktakkhor. So you would enter poro to get the desired result. This is also the correct phonetic representation of পর. - Forced Hosonto
The special sign Hosonto is sometimes necessary especially when writing foreign words. To put a Hosonto under a Bangla consonant put in a ` (reverse quote or grave accent) after the letter.
e.g. Entering in`puT will show as ইন্পুট.
Differences from Avro Keyboard Phonetic scheme
Many Windows users are familiar with the Phonetic scheme used in the Avro Keyboard software from Omicron Labs. Here are the main differences you will see when you use suborNo.Bangla Letter | Avro Keyboard | suborNO | Comments |
---|---|---|---|
'চ' | c | ch | Pronunciation matched ch more closely than c |
'ছ' | ch | chh | Aspiration matches chh, takes into account assignment of ch above |
'য' | z | J,Y | Jaw is more correctly pronounced with letter j |
'শ' | sh,S | sh | secondary S dropped because there are 3 shaw sounds |
'ৎ' | t`` | t^ | ^ adopted for abrupt sound endings |
'ঃ' | : | H^ | Wanted to avoid punctuation signs in phonetic representation |
' ঁ ' | ^ | n^ | The nasal sound is better represented by including the n |
J-phola [cons]+J | y,Z | Y | Consistent with J as single letter |
reph [vow]+r+[cons] | [v]rr[c] | Not Reqd | Normal rules take care of reph |
hosonto '্' | ',,' | backtick '`' | Avoid use of punctuation |
colon : | ':' | Not Reqd | Necessary because of assignment of : to bisorgo 'ঃ' |
Software used with suborNo on the website
The transliteration schemes currently in use, are primarily employed at the point of writing, where the English Phonetic characters are transformed inline to Bangla. This is true of Google Transliterate and Avro Keyboard.For a learner of Bangla, It is important to simultaneously see both the source English and the generated Bangla. This way, all errors can be corrected in the Phonetic English version and the correct way to render a Bangla word will be made more memorable. At the expense of some speed in entry, the website implements a 2 panel input area, with the user entering Phonetic English and the "system" rendering it into Bangla.
The core English-to-Bangla transliteration code is currently written in PHP. The user interface uses Javascript and AJAX. A number of software modules are used within the site.
- English-to-Bangla Encoding - This function is written to handle large amounts of text. It is useful when you want to save the English Phonetic input from one session and get it re-rendered to Bangla at a different time.
- Bangla-to-English Decoding - This function is usually not available when English is rendered into Bangla as you type (like in Google Transliterate and Avro Phonetic). As more and more text appears in Unicode Bangla format, it will bocome necessary to convert the Bangla back to Phonetic English so that editing and text manipulation can be done using the Phonetic English.
- Trailing "o" during B2E Decoding - The English letter 'o' is the normal inherent vowel at the end of Bangla consonants. So, for E2B transliteration 'k' and 'ko' are equivalent and will be rendered as 'ক'. For example, take two words ending in the letter 'ত'. The word 'আহত' injured would be written ahoto with a trailing o sound, while the word 'উচিত' should would be written as uchit without the 'o'. By analysing 16,000+ words in the site dictionary, the algorith for deciding on the trailing 'o' has been coded in software.
- Syllable identification in Bangla words - Syllables are normally formed by a consonan-vowel combination. Unfortunately, compound consonants and consonant signs make this a formidable task. Software to aid the identification of syllables has been written. The importance of the syllable breakdown is that Bangla words have the stress on the first syllable, and then the odd-numbered syllables after that.