Wise to the words

Here's a bit of a different puzzle. It may not have a solution.

Produce an algorithm that, given a list of correctly spelled English words (such as the enable1 list that can be downloaded from the National Puzzlers League web page), will correctly determine which of the Ys are consonants and which are vowels.

Dan Tilque:

I assume that Y in words like 'way' is to be considered a vowel.

Here's a start:

If the word is a compound word, consider only the component word with the Y in it. Ignore silent E's in the word. Then Y is a consonant if it is NOT preceded by a consonant, and IS followed by a vowel. Otherwise Y is a vowel.

Determining whether an E is silent or not, and whether a word is compound or not, is left as an exercise for the reader.

The compound-word clause makes FARMYARD work right; Y is deemed not to be preceded by a consonant because you only consider YARD. If there are any non-compound words where a consonant - Y - vowel sequence is pronounced analogously to that in FARMYARD, then the algorithm will go wrong on these, but I can't think of any offhand.

Another class of exceptions is those where Y is not really either a vowel or a consonant, but just stands for its own name, as in YTH.

What else did I miss?

Mark Brader schrieb:

Quite some nifty rule! I'd been thinking about the same relations, buthadn't found such a nicely condensed rule.

'ing' needs to be considered a component so that words like 'relying' work.

The rule makes 'foyer' have a consonant Y, and 'boy' a vowel Y. My Webster shows the same'i'-sound for both.

Another word that breaks the rule is 'canyon' with a consonant Y.

I believe that to split words into components may not be enough; splitting into syllables seems necessary, but that's tough as well.

Any algorithm probably requires special cases for words such as 'cotyledon' and 'cotylosaur' where the Y is toneless.

'yth' works fine because by your rule Y is a vowel then.

Cheers

I'll respond to some of the points made by Michael Mendelsohn:

When I pronounce 'foyer', it's like 'foy-yay'. The Y contributes twice to the sound

That's a sly caveat all right, because we need rulings for 'payola' and 'kayo'. Here the Ys are vowels according to the 'way' assumption, but consonants according to the rule _unless_ the words are compound.

I think there's a reasonable case for calling 'kayo' a compound, but the case for 'payola' is dubious, because -ola is a suffix and not an independent word.

Speaking of '..yo..', are 'mayonnaise', 'coyote', and 'peyote' exceptions to the rule? I think so

Mark Brader schrieb:

Ok, upon closer examination my dictionary reveals that a) it is hyphenated foy-er and that b) there are several voicings ranging from foier, foir, foia to foiya.

I think this implies that there is no general solution to the problem as posed.

How do you know it's not a silent consonant?

So maybe the problem definition has to account for 'y' being both a consonant and a vowle sound, in both orders.

Cheers

Mark Brader:

Larry Tapper:

As far as I'm concerned, all 5 of these words

banyan bowyer bunyip halyard lanyard lawyer minyan pinyin pinyon sannyasi satyagraha sawyer

Robert Israel This e-mail address is being protected from spam bots, you need JavaScript enabled to view it Department of Mathematics http://www.math.ubc.ca/~israel University of British Columbia Vancouver, BC, Canada

So would you say there's an unwritten consonant in the word 'paella', then? For me, 'peyote' sounds more like a straight diphthong.

POLYNYA, an open area of water surrounded by ice, is listed in the National Scrabble Association's Tournament Word List.

In this case, the Y is half of the vowel 'ya', the word being spelt with in Cyrillic with a palatalized N, followed by the 'ya' vowel. (The palatalization is spelt using the 'soft sign: ïîëûíüÿ). Interestingly, my Russian-English dictionary translates the word as POLYNIA, which is *not* in the TWL.

It's a good start. But I can't see why silent Es need to be ignored. Also it doesn't handle the handful of words with YY (assuming that any Ys in the word are in an indeterminate state while the algorithm is in progress).

Unfortunately, that's really a job for the computer. Or lacking that, some additional rules (exceptions to the above rule) that don't depend on the computer knowing whether it's a compound or not. For instance, one such rule could be something like

If the Y is preceded by an N and followed by a vowel but the Y is not immediately followed by ING, ISM, IST, or ISH, then the Y is a consonant.

AFAICT, this takes care of all cases with -NYv- except for ANYONE and possibly some really obscure words.

Anyone want to estimate how many such exception rules would be