`Rich Phonology’:
Reasons
for Rejecting Phones and Phonemes.
Some of the Evidence: Read it for
yourself
Indiana University
Introduction.
For the past century scientists of
language have believed that
words
are composed from an inventory of phonemes
(or phonological segments) which are selected from a larger set of
universal phonetic units called phones.
Many linguists further assume that all phones and phonemes can be fully
represented as vectors of phonetic
features selected from a small, universal set.
Unfortunately, despite
the strong intuitive appeal of this description, there is essentially no
behavioral
evidence
supporting this idea. The purpose of this webpage is to
provide a
selection from the experimental evidence supporting my
seemingly radical claim,
and to make the material easily accessible.
I am working out a theory called `Rich
Phonology'. It is an approach to linguistics that assumes words
and
phrases, etc. are stored in memory using a much more detailed code than the
traditional view allowed -- representations specified in continuous
time rather than in discrete time. Alphabet-like representations are
useful and practical for those of us trained in their use, but
they apparently play a very small role in everyday, real-time use of
language by
our brains. This means that the bit-rate of linguistic memory is
larger by at least an order of magnitude than previously imagined based
on a
speaker-independent, abstract segmental representation. One
implication is that exemplar theories
of
memory and episodic memory
models are
much better models of our linguistic memory than we thought. They
are certainly better models than any abstract
alphabetical representation. The strong appeal of
letter-based abstract representations, and the sense that this is all
the information speakers and hearers really need are intuitions that
reflect primarily our lifelong experience translating letters into
speech and speech into letters.
Some Papers
Port.
R.
F. How are words
stored in
memory? Beyond phones and phonemes. (2007). New Ideas in Psychology 25, 143-
170 (Elsevier)
This paper
presents the central story of
this webpage and reviews all the
arguments
made here. The rest of this site can be viewed as an elaboration
on the argument in this paper.
Port, R. F. (2008) All is prosody: Phones and
phonemes are the ghosts of letters. Keynote address for Prosody2008, Campinas, Brazil, May,
2008. To appear in the conference Proceedings.
This is a much shorter version (10 pages)
of the same
central story - but the telling gets a little clearer each time I
write it. So one might like to read this short version before the
longer one above. This version, as suggested by the title, is
oriented toward participants in the prosody conference.
Port, R. F. (2007) Phonology is not psychological and speech
processing is not linguistic. Mspt submitted to the Society for
Philosophy and Psychology annual meeting in Toronto. (About
5-pages)
This article shows how my distinction
between Phonology and Speech-Language Processing differs
from Chomsky's distantly related notion of Competence vs. Performance.
Every language requires a very
high-dimensional
space of phonetic
properties (relative to the size of any
phonetics previously proposed). This space includes indexical
speaker- and
voice-dependent
properties,
speaking rate information, etc. along with all the usual phonetic and
phonological parameters. Since the acoustic features employed in
these representations are acquired independently by each speaker from
their auditory and linguistic experience, the features employed will be
different from speaker to speaker (rather than fixed and
universal). Furthermore, any given word or phrase will have many
independent representations in
memory. The evidence for these ideas comes from
surprisingly many directions, while actual experimental data in
support of our intuitive
description of speech (based on letter-like segments) is
essentially nonexistent, as far as I can tell.
Port, R. F., & Leary, A.
(2005). Against
formal phonology. Language
85, 927-964.
This paper argues that an alphabet cannot capture most of the important structure in speech, thus providing the foundation on which the later papers rest. An alphabet provides a very poor model for speech perception although it is a very useful technology for reading a language one knows well. Languages differ in far more ways than any learnable alphabet could represent. For example, speech sounds differ in timing details for which symbol tokens are particularly poor descriptive devices. Furthermore, there is no evidence that speech sounds across languages differ by discrete jumps, as required by the formalist views. Much of the evidence about the inadequacy of traditional transcription has been available for a long time, but most of us have ignored it or made excuses for it. Since all formal models of language are built on a foundation of discrete, apriori phonetic symbols, the very idea of a formal model of language is rendered impossible, unless someone figures out a way to provide a genuinely discrete universal phonetics. But there is little hope for that.
Port,
Robert (2006) The graphical
basis of
phones and phonemes. In Murray Munro and Ocke-Schwen
Bohn (eds.) Second
Language Speech Learning: The Role of Language Experience in Speech
Perception
and Production. Benjamins,
Argues that our lifelong experience with the alphabet biases our
perceptual
experience of speech. Segmental descriptions of speech
present powerful and compelling intuitions to us. But these intuitions
are largely an illusion. Chomsky's insistence that we need only trust
our intuitions to
find
reliable linguistic descriptive tools has been a costly error.
We CANNOT trust our intuitions to report what the real
structure of language is. The nature of our linguistic
intuitions, along with their origin, should be viewed as just another
empirical problem requiring careful
investigation.
If you are
interested in reading some of the background material which supports
these surprising claims, please follow the
link.
Robert F. Port, February 24, 2008
First put up in
September, 2007