Irregularity is certainly a matter of degree. In this case, I would measure it
as the number of rules necessary to generate the correct pronunciation from the
letters. For example, one rule could be that a single vowel followed by a
consonant and final 'e' is pronounced long. In a large majority of cases, this
rule gives the correct pronunciation, but there are a number of exceptions such
as "have". One can either have additional, more complex rules that catch the
exceptions, or simply list them. When they are listed, each would count as a
separate rule. Note: when you look at words weighted by frequency some
surprising rules pay off, such as "ould" => "ʊd"
Given that, Dutch and Spanish are quite regular, German slightly less so, French
or Danish significantly less, English quite irregular, and Japanese even more
"Reynolds, Gregg" wrote:
> > -----Original Message-----
> > From: Mark E. Davis [mailto:firstname.lastname@example.org]
> > Sent: Wednesday, December 08, 1999 11:12 PM
> > It's actually much worse than that; I have done some analysis
> > in the past, and on a
> > average given line of English text you will have quite a few irregular
> > pronunciations -- that is, cases where the pronunciation is
> > not predictable from the
> > letters without actually knowing the words involved.
> Not surprising; that alone does not imply irregularity. Unless you define
> "regular" to mean phonologically transparent; but I would challenge such a
> definition. Rather than ask if the orthography allows non-speakers to
> puzzle out the sound, I would ask if native speakers of normal intelligence
> can learn to read with reasonable effort. A cultural rather than
> "scientific" norm. Needing to know the language in order to be able to read
> it is not a bad thing.
> Philosophical question: if somebody with absolutely no knowledge of English
> managed to produce sounds approximating an English utterance - say, "I've
> fallen in the woods and I can't get up" - would they be, ontologically,
> English sounds?
This archive was generated by hypermail 2.1.2 : Tue Jul 10 2001 - 17:20:56 EDT