2

Dictionaries contain near 80,000 entries (less or more than that) and most of those entries have phonetic pronunciations written beside them.

However, English might have more than a million words, if we count every word which can be distinguished from others by looking at its characters.

For example, while teach, teacher, and teaching are words with the same root, they actually have different characters which usually result in having different pronunciations.

This way we might say that more than 400,000 English words don't have a formal phonetics representation written down in any dictionary for them. However, some words do need to have phonetics. Names are good examples. How should one pronounce Archimedes for example?

I just wonder if it's possible to create phonetics based on morphological analysis of a given word or not? Does such a tool exist online? Does it mean that dictionaries can create phonetics for almost any given word?

3 Answers 3

2

To some extent, there are rules you can apply to pronounce, say, teaching based on the known pronunciation of teach. But there will always be countless exceptions.

Consider a read-only file: is it /rɛd/ or /riːd/? You need to know the etymology of the word to be able to tell, or have a separate sub-entry for it in the dictionary (it is /riːd/).

Then there are many, many names that are irregular and require individual treatment, like Thomas (/toməs/, not /θoməs/) etc. etc. The list goes on.

1
  • The list goes on and on:)
    – Noah
    Aug 23, 2012 at 11:22
1

If the entries are sufficiently rich, it is certainly much possible. You will just have an algorithm for the regular cases, then annotations for different exception classes. The entry for Thomas would have a feature in it noting that the th is pronounced atypically, etc. Text-to-speech systems do something like this.

3
  • 1
    I read Saeed's question as asking whether it might be possible to create an algorithm that can predict how names and all other words that are not in a dictionary are pronounced. If you are allowed to annotate exceptions in this algorithm (so outside the dictionary), it can't really be said to predict pronunciation based on orthography and what information this dictionary contains. You can pronounce anything if you can annotate anything.
    – Cerberus
    Aug 24, 2012 at 1:09
  • @Cerberus - not quite. Annotating everything won't help you as the language develops, and new words come along. If you have a good algorithm that doesn't need very many annotations, then that algorithm should be able to be applied to any additions to the language.
    – user780
    Aug 26, 2012 at 0:59
  • @DavidWallace: The idea is that would add new annotations as new words are coined or old words changed. But then, if you had such an algorithm, which I think is not possible, you would still need at least some annotations, as you say.
    – Cerberus
    Aug 26, 2012 at 4:23
0

Sounds like you have a great idea for developing linguistics software! A mutational dictionary that could change on the fly would be possible, it just has to be made. Putting it all on paper would be another thing though! Explaining the phonological rules of English phonotactics in a paper dictionary might be more feasible.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.