Why don't text-to-'speech' programs process IPA or other such phonemic, phonetic or phonological scripts?

Why don't text-to-'speech' programs process IPA or other such phonemic, phonetic or phonological scripts?
Charles Jannuzi
University of Fukui, Japan

What language would such a program put out? Text-to-speech/spoken language
programs basically work on the principle of recognizing whole words in a
language (based on their relatively unique spellings and the spaces in between
each word) and matching them up with the recordings of pronounced whole words.

The misconception here is that, when we speak,  we somehow 'generate' phonemes or allophones and these are strung together into something that adds up to 'real speech'. There is no phonemic or allophonic model of a  language that can do that--which is why speech recognition programs work only if  you train them and bark simple words and phrases at them. Even the most algorithmically powerful ones depend on you speaking to them not in sense or  breath groups of fluent speech but in a greatly slowed down form of 'clear speech', with more pauses added.

Phonemes and their allophones are really only written descriptive models of  language and as such simplified idealizations. They can't be found in  articulation, they can't be found in the acoustic stream, and no one has shown  convincing evidence that they form some sort of phonologically bottom-up unit in  language comprehension (for one thing, such models slow the process down way too  much to account for comprehension such as how humans actually do it).

I suppose an IPA program could be created by assigning an IPA-based spelling to  each word in a given language's lexicon. Since EFL learners dictionaries use such notation to show some sort of 'phonemic', canonical pronunciation of the word, such whole words written in IPA characters could be used just as English spelling  conventions are.  I guess no one so far has seen such a step as useful or necessary. Do learners of EFL need to learn the English lexicon as spelled in IPA too? Most would balk, as would a lot of EFL teachers.

I did see a pronunciation-training program back in the Win 98/NT era of  computing that seemed to be based, not on whole words, but syllable types and  morphemes. You typed a word, and the program basically produced pronunciations  for the word you typed. The animation seemed to be a sequencing of photo stills  of a woman's face showing 'visemes'--the visual equivalent of a phoneme (e.g.,  an open round mouth for the English sound [ou]). By playing around with the input, I tried to 'back engineer' this program in my head to see how they  analyzed the language. This seemed to be:

1. Audibly prominent phonemes
2. Their visual equivalents in oral gestures--visemes
3. Pronunciations of syllables (syllable types)
4. Pronunciation of whole words

It seemed to be quite a clever bit of programming to make the sequence of stills  match up and produce connected speech that wasn't just whole word sequences.  However, the visual oral gestures were much simpler than the ones in real speech  (think of the tricks animators know to give the visual illusion of a face  speaking, which our own brains fills in and makes match the soundtrack we are  listening to). Also, it didn't sound like very natural speech for phrases.

I think audio-visual files of whole words and phrases should be compiled into an  audio-visual lexicon--for example, Ogden's Basic English plus the 3800 most  frequent words and lexical phrases of spoken English. Today's technology makes  it possible. The state of ELT and its publishing make it a mostly undesired niche.

Truespel - the English Based Phonetic Notation – for ESL

Truespel - the English Based Phonetic Notation – for ESL

Tom Zurinskas

Truespel phonetic notation was created in 1986 to answer the following question: What would be the best way to design a phonetic notation based on the sounds and most typical spellings of English?  After all, if English is the lingua franca of the world, shouldn't an accessible approach to phonetics (for ESL and EFL learning ) be based on it as well?

As a human factors specialist at the time, I found that pronunciation guides were not standardized nor computer-friendly, not even for copy/paste. Moreover, academic notations such as the International Phonetic Association (IPA) system (often called 'the IPA') were not English or computer- friendly, using a mix of languages and codes (such as exotic symbols) to spell sounds.   Because of this, phonetics is not taught to children in the US, which impacts, for example, the learning of phonemic and phonological awareness, key correlates to reading development and success.  But truespel makes this possible.

To research truespel, I found only one place in the world interested in English phonetic notation, the Simplified Spelling Society, now The Spelling Society.  Various notations were depicted, some very similar to truespel,  the objective being that if English, like Finnish, were spelled consistently, it could be learned more easily, like Finnish, (such as for literacy purposes), in months rather than years.  Yet major reform of English spelling standards is extremely hard to do.

The truespel goal instead was to supply a pronunciation guide to assist readers with pronunciation and phonetic look-ups.  The design rule was to use the most frequent sound spellings, or those that are least conflicting with traditional spelling.  For certain letters or sounds,  this was not easy to do, especially vowels, because there are five vowel letters to spell 17 (or more) vowel sounds in English.  For example, the letter is found in the following words: on, do, of, for, word, motor, and pilot.  In each of these words, however, each instance of the letter stands for a different sound.

Using a scientific approach, I analyzed English, as set down in truespel books 1 and 2, to determine spelling norms for English sounds.  Book 1 is a dictionary list where words are spelled once.  Book 2 is a frequency list of words found in typical text (which counts phonemes/categorical sounds for repeated words).  The analysis verified the reasonableness of the sound spellings of truespel.

The reason these studies were possible was that English was respelled in truespel.  But to do that, a guide for proper pronunciation was needed.  The answer was 'talking' dictionaries.        The ones used largely represent USA English, one of the most widely taught 'standards' because of the US's large population (300 million people) and its economic and political power in the world.  The resulting phonetic spelling is more accurate than any other transcription, because (1) it's recent and (2) all schwa symbols are removed and clarified (in written English, the schwa sound is represented by many different spellings, and not all reduced vowel sounds in English are actually schwa).  Truespel book 3 shows the Voice of American simplified English dictionary with a truespel guide.  This guide includes many typical USA alternative pronunciations that all other dictionaries should include but do not.

In statistical terms, English's writing is overall mostly consistent, especially in the consonants that form the framework of written syllables.  My research shows that for English, 90% of consonants are spelled in the top most frequent way, while only 50% of the vowels are.  Thus, many words need to be sight memorized, and a phonetic guide would be most helpful as to the pronunciation of words.  Truespel provides such a guide, and it's free.  Truespel.com has a free phonetic converter provided by Joe Davison, where text can be pasted and converted to phonetics.  Teachers can create pronunciation lessons that way.  The site also has a web page converter.  Thus, the internet can be converted to truespel using it. Because of the power of computers and internet, truespel is everywhere.

Truespel is simple.  University grad students took only 15 minutes to learn it.  This is because 30 out of 40 phonemes are self-evident.  English-speaking 3rd graders can learn truespel in about an hour.  The beauty of it is that truespel is also capitalization-friendly, punctuation-friendly, copy/paste-friendly, filename-friendly, and spreadsheet-friendly.  No other notation does this.  Usefulness is the key.

Teachers can use truespel to:

1. Show beginning readers the sounds of English to teach phonemic and phonological awareness
2. Show English pronunciation to ESL students.
3. Show pronunciation of words from other languages in truespel  – translation guides.
4. Integrate English reading instruction, pronunciation materials, dictionary pronunciation guides, translation guides, phonemic aptitude tools, and analysis tools.  The future goal is complete integration, which does not exist now.

For a quick tutorial on truespel regarding the 40 sounds of USA English in a simple story see http://tinyurl.com/yh46rgc  and see http://tinyurl.com/yls55da for indicating stress in truespel.

For further information about truespel, contact me at tzurinskas@yahoo.com
Tom Zurinskas, creator of truespel.

Tom Zurinskas, USA - CT20, TN3, NJ33, FL7+
see truespel.com phonetic spelling