Tamil Alphabet Puzzle
On Vacation, Trip Report
A new method is described and demonstrated here which may make it
possible to deliver literacy education to a smart and motivated
illiterate person at the cost of a sheet or two of paper, with little
or no investment of time or explanation from a teacher. Field
experiments are required to prove it out, but if this method is
successful even for only a small percentage of recipients, it would be
a practical and affordable way to bring literacy to millions.
The scale of the problem
Consider India, with an estimated 400 million illiterate people.
Effective, free, public education is not widely available in India,
and private teachers demand payments which most cannot afford. Given
the scale of this problem, well-meaning people, organizations and
governments are really helpless. It is impractical for
resource-limited institutions to pay for or organize literacy
development programs that can influence more than a few tens or,
optimistically, hundreds of thousands of people. So the great mass of
poor people in India, and elsewhere in the world, for that matter,
have no way out of the illiteracy trap. This work aims to a practical
solution to this dilemma.
Linguists and phoneticians have traditionally used sound charts to
relate letters or symbols to mouth positions and movements. Vowel and
consonant charts are tables labelled with names of phonetic
characteristics, such as "labial", "dental", and so forth, and they
are universally used by students in introductory phonetics and
phonology classes to learn the sound-symbol correspondence of the
International Phonetic Alphabet. Here we will use similar charts as
the basis for learning the sound-symbol correspondence, by making them
understandable to illiterate people so they can associate their native
language's sounds to the symbols of its alphabet.
How I thought of it
When I was visiting South
recently, since I'm a phonetician and a linguist, I made a
small personal study of the Tamil language, which is spoken in the
region I was visiting. I discovered what many already knew, that they
do not make distinctive use of the voicing dimension. That is, they
have no contrasts between /p/ and /b/, or between /s/ and /z/, or
between /k/ and /g/, etc. Instead, they have a sound that is
phonetically in-between our categories, and depending on the adjacent
sounds it can sound more like our voiced sound or more like our
unvoiced sound, but in Tamil this difference doesn't make a difference
between words as it does in English and many other languages.
That discovery got me to thinking, and I realized that the consonant
chart for Tamil only needs two dimensions, not three, since voicing
can be ignored. That means that a TWO dimensional table of sounds
will fully represent the sounds of this language.
The usual consonant chart is a three-dimensional table. The
- (1) the voicing state of the larynx (voiced or
vibrating, and unvoiced or non-vibrating),
- (2) the place of
constriction in the vocal tract (divided up into several places
between the lips and the glottis), and
- (3) the manner of articulation
(closure, grooved opening for whistling sounds like /s/ or /sh/,
closure with an opening on the side as for /l/, and so forth).
With Tamil, a two dimensional chart showing place and manner of
articulation can represent all the consonants. Such a display is so
pleasantly simple that I thought it ought to be understandable for
children and even adult illiterate people. Indeed there's no reason
that one can't figure out how a table works, just because one doesn't
know how to read.
Since I myself was learning to read Tamil during this whole trip, I
eventually had another insight, that would help completely illiterate
people make use of sound charts: simply change the labels of the
columns and rows of the sound charts from words to pictures!
How it works
The two insights, of having a two-dimensional chart for sounds and
using pictures as the labels of the columns and rows, are the basis of
this new method. No literacy skills are required for a reasonably
intelligent and illiterate person to interpret such a table. All one
needs to know is that the tables represent the sounds for each letter,
and then one can figure out the rest entirely alone.
The arbitrary sound-symbol correspondence is the biggest obstacle in
literacy. Without knowing in detail the elements of that
unpredictable, impossible-to-guess, correspondence, it is completely
impossible to read or write. But if you know that much, you can at
least sound out the letter sequences, and thereby figure out the words
that they represent (for words you happen to know).
And until now, without the commitment of a teacher or other expensive
methods to teach that correspondence, there is no way for an
illiterate person to learn it. THIS is the main problem; everything
else is a footnote.
In short, a completely illiterate person should be able to to
understand these charts, and learn the letter-to-sound
correspondences, without any previous knowledge of how to read. Since
that arbitrary correspondence is in itself the primary obstacle to
basic literacy, a single sheet of paper, along with a one-minute
explanation of the purpose of it and how to interpret the charts, may
be all the outside intervention that a motivated and intelligent but
illiterate person needs in order to figure out how to read.
Literacy also includes other skills:
- knowing how to hold, use, and draw letters with a pen or
- but one can figure this out alone.
- knowing that sequences of letters represent sequences of
sounds and that words are composed of sequences of sounds.
- One could also figure this out alone; on the other
hand, explaining it should take less than a minute.
- knowing the idiosyncrasies of a writing system
in which the spelling doesn't determine the pronunciation
or vice versa.
- Most languages are relatively phonetic, like Spanish
(where you write it like you say it), rather than inconsistently
phonetic, like English (where it's hard to know how to spell many
words), or non-phonetic, like Chinese (where symbols represent
words instead of sounds). This method will work best with
consistently phonetic written languages; not so well with
inconsistently-phonetic written languages; and not at all with
- learning written vocabulary that is not part of the
native spoken vernacular.
- Many languages do not have
a distinct literary form, though those with a long literate
history normally do. For the former this problem doesn't exist;
for the latter, it is usually a life-long learning process for
most speakers of that language anyway, and while we don't deliver
a full packaged curriculum solution here, we do get them started
with basic tools which will let them begin to learn unknown words
themselves, just as they learned their spoken vocabulary through
experience and context and without consulting a dictionary or formal
This method is therefore limited to languages with phonetic writing
systems, the more consistently so the better. It assumes learners can
figure out how to manipulate a pencil on their own. It doesn't try to
teach separate literate vocabulary; that's everyone's lifelong task.
In short, we do not attempt to provide full literacy, but only basic
literacy skills. Despite this limitation, the method solves the
central problem, which is enough to enable learners with absolutely no
other educational resources to learn enough to read and write for
themselves and to begin to understand and learn about more advanced
Implementation for Tamil
Since Tamil can be handled with a simpler consonant chart, using two
dimensions instead of three, our first work is with Tamil.
On a single sheet of paper can be printed the two tables displayed
below (along with some enhancements discussed after the tables).
These two tables relate the symbols of the language's alphabet to the
positions and movements of the lips and tongue.
- The first table is for vowels. For the rows, it shows the
jaw/tongue/lip movements associated with each of 5 vowel qualities.
For the columns it shows vowel length, indicated by a bar of shorter
or longer length.
- The second table is for consonants, and it shows "place of
articulation" in columns, and "manner of articulation" in rows.
The drawings that label each row and column are clear enough to
unambiguously pick out the sound associated with each symbol in the
Tamil speakers who do not know how to read should be able to figure
out the sound-symbol relationships for their script by carefully
studying these pictures and making the movements and sounds indicated.
Most people that I have shown these tables to, who are not Tamil
speakers or readers, have been able to guess the sound associated with
each symbol, on the first attempt, about half the time.
These pictures may provide all the information needed to figure out how to
read, by oneself, without a teacher or any other assistance or
support, simply by studying this puzzle-like document and figuring it
out. Not every illiterate person can be expected to do this
successfully, but the intelligent and motivated ones will succeed, and
they will be able to teach others.
Three enhancements will complete the puzzle of the Tamil alphabet.
Three additional enhancements will make it easier to figure it out.
The above items complete the coverage of the information content
of the Tamil script. Three additional items, below, will help to
ease the task of learning and memorization for students.
The two tables cover all the single phonemes ("simple sounds") of the
Tamil alphabet, except for /ai/, /au/, and /ksha/, which can be
represented as sequences of the simple sounds. Some formulas like
"a + i = ai" can provide the needed answers here.
The Tamil script also includes "syllable" symbols, composed of a
consonant symbol with some diacritic mark or additional symbol
representing the following vowel. These are essential to know in
reading Tamil. A third table, not shown here, can provide the needed
information. The table will have rows labelled with the consonant
symbols already learned, and columns labelled with the vowel symbols
already learned, and with each cell containing the consonant-vowel
syllabic symbols of the Tamil script.
- Vowel-less consonants.
Finally, the Tamil script includes a diacritic dot
above the consonant symbols, which indicates that the schwa vowel
implicit in each consonant symbol is absent, as in word-final
consonants or cluster-initial consonants. I'm tempted to say that
this dot's phonological meaning can be discovered without
explanation by readers who know the words that they are reading. But
some formula similar to "ka - a = k" is also possible to show the
- Cross-confirmation. A means of cross-confirmation is
needed so students know when they have guessed correctly. This can be
provided in the form of a list of object-name/picture pairs, one for
each letter in the alphabet. This "a for apple" idea can be applied
easily, by selecting, for each letter in the script, a short, easy,
unambiguous, and easily drawn object name starting with that letter,
and by including a drawing of that object in a corner of the
corresponding cell of the table. These pictures will provide both
cross-confirmation to the student that they have guessed the sound for
the symbol correctly, and indeed it provides an additional clue about
the sound itself.
- Minimal text.
A minimal text which maximally exercises the reading abilities of
the student will help as a memory aid. In English, the famous example
is the sentence, "A quick brown fox jumped over the lazy dog", which
contains every letter in the alphabet. A similarly short and
exemplary text should be concocted to add to the page.
- "Pass it on." The last thing on the delivered document
should be a request written in the students' language: "You received
this as a gift from the many people who helped to make it and bring it
to you, who all wish you well. When you can read these words, know
that you have made us very happy. We now ask you in return to
give the same gift to others by teaching them what you have learned.
For this we thank you."
- This method does not teach the consonant/vowel distinction.
Students just have to infer each sound from the pictures. Fortunately
the pictures implicitly contain the information that some are
consonants and others are vowels, so this may be unnecessary.
- This method does not explicitly teach the fundamental alphabetic
principle, namely that the symbols represent sounds (as opposed to
words or other imaginable entities), and that words which are spoken
as sound sequences are themselves represented in writing as sequences
of the corresponding symbols. Either this principle must be taught
when passing out the papers, or it must be guessed or understood
implicitly by the student -- which could lead to failure of the
method. So it may be necessary to give an explanation along with the
document; fortunately the explanation can be very brief, as in: "These
are symbols (point at the contents of the table) for the sounds you
make with your mouth doing these movements (point at the drawings
which label the rows and columns)." That one sentence may be enough to
eliminate this problem.
- Test it on 20-30 Tamil illiterates (a "market survey"!)
- Refine it as needed.
- Publicize it and get it into the hands of every illiterate
Tamil speaker in India and Sri Lanka.
- Send it to every newspaper (even though illiterates
don't read papers, friends of illiterates do).
- Send it to every charitable organization and ask their help.
- Perhaps organize a door-to-door distribution campaign, if funding
can be arranged.
- Get some linguists to do the same thing
for all kinds of other written languages, and get those out to
their respective populations.
By following this plan, it is not unreasonable to hope that a
significant percentage of the world's illiteracy problem can cheaply
and quickly be overcome. Not every illiterate person will be able to
puzzle it out, but a significant percentage will be able to, and each
one who does can teach others, using the same document.
How You Can Help
Please help us to make this vision a reality. See our funding proposal
for the validation experiments. One way you can contribute is to help
validate the method by visiting the Tamil
Alphabet Puzzle and giving your guess of what English sound is
closest to each of the various symbols. Your guesses will help us
evaluate how well the system works. (In early data, >90% of submitted
answers have been correct, so guessing accuracy is indeed pretty
This project needs help with these additional items:
Of course if I go to Chennai I can probably solve these pretty
quickly, but here in the US, your help is the only way. Please contact
me! Praise and dialog is what gets me working on this. Your
suggestions and moral support are so precious!
- The artwork is poor, I would say it is only passable at best. I
drew them as well as I could but I am no artist. How to draw a "flap"
sound (fourth row of the consonant chart) is a difficult question; do
you have an answer?
- We do not have a list of "a for apple" words for Tamil.
- We do not have pictures for the "a for apple" words
(we need PostScript format, black-and-white, outline or filled-outline
drawings which are clear and unambiguous when printed
at a very small size (30 pixels by 30 pixels)
- We do not have a minimal text as described above.
- We do not have a translation into Tamil of the "Pass it on"
- We do not have any of these things for any language other than
This project also needs suggestions and contacts of people and
institutions who can help in any part of it. Right now volunteer work
doesn't seem to be getting things done too quickly. So I am seeking
pilot-project funding to pay for art-work, linguists, testing, and
distribution, basically to send myself to India to do some experiments
and come back and write it up, before scaling it up to be a
large-scale delivery project.
- Initial concept and implementation are due to Tom Veatch.
- The font system, itrans, was built by Avinash Chopde.
- The Tamil fonts used in itrans were made by Thomas Ridgeway.
- The web interface to the Tamil font generator was built by Hari Adiseshu.
- The "a for apple" idea was contributed by Mary Veatch.
- The Tamil alphabet puzzle idea for validation of the concept via web forms
was contributed by Michelle Effros.