Proceedings of the Xllth International
Congress of Phonetic Sciences
(ICPhS)
August
19-24, 1991, Aix-en-Provence, France
Volume
2/5, p. 142-145
===================================================
Sound
Distinction:
Universal
Inventories of Phonic Substance
or
Language-Specific Systems?
Vulf Y.
Plotkin
Siberian
Branch, Academy of Sciences
Novosibirsk,
Russia
ABSTRACT
Since
each language system is a unique segmentation of universal substance shaped
into its elements, ultimate phonological units are not items in universal
inventories of so-called ‘distinctive features’ to be incorporated into
language systems. Each phonological system segments the universal human arsenal
of sound-distinctive capacity in its own way, forming a specific set of
neuro-muscular impulses, of which each activates a fully automated aggregate of
articulations.
Phonological theory experiences today tremendous difficulties on account
of its growing isolation from the needs of modern phonetic technology, which,
finding little response to its requirements, has to rely on its own groping
solutions of phonological problems. This gap between theory and practice is the
inevitable result of the failure to work out an adequate answer to the
fundamental question of the ultimate phonological unit.
It is widely believed that phonology as a linguistic science started by
assigning that status to the phoneme – hence its other name ‘phonemics’.
However, as early as in 1936 Josef Vachek showed that the phoneme was not the
smallest indivisible phonological unit, because it could contain smaller
non-successive simultaneous units, e. g. sonority, palatality etc. [13]. The
idea was developed by Bohumil Trnka at the 3rd International Congress of
Phonetic Sciences in Ghent [12]. Then Roman Jakobson devoted four decades of
pioneering work to the search for the phonological quantum – the ultimate
language unit named ‘distinctive feature’ [7; 8; 9]. The best-known result of
the quest is the universal inventory of a dozen items, of which phonological
systems are built for all languages. The inventory was revised theoretically
and enlarged threefold by Noam Chomsky [4].
While fully recognizing the great scientific and practical value of R.
Jakobson’s achievements, we have to admit nevertheless that the entities he
discovered and catalogued are not what he thought they were, i.e. the ultimate
phonological units. It stands to reason that no items from a universal set can
be directly employed as units in a language system [5]. N. Chomsky was
therefore quite consistent in stressing the language-independent nature of his
inventory of ‘features’ [3]. What R. Jakobson and N. Chomsky really inventoried
is indeed universal, it is the common human arsenal of sound-distinctive
capacity. Naturally, all phonological systems are based on it as their
substance foundation. But no part of underlying substance can be directly
integrated into any system, and language systems are not exceptional in this
respect. The elements of the universal anthropophonic distinctive potential
listed in the above-mentioned inventories are certainly not ready-made units to
be selected by and included into a concrete language system. A language unit is
not a mere piece of substance, but substance shaped as an element to fit into
the unique structure of the given language system. Consequently, elements of
different systems cannot be identical with elements in other systems, however
close they might seem in substance. This has long been accepted for phonemes,
but not for ‘distinctive features’, which, according to R. Jakobson, coincide
with the same ‘feature’ in other languages [8]. Regrettably no theoretical
explanation was offered for this deviation from the general principle that
precludes the compilation of universal inventories for phonemes, morphemes,
words from all languages.
The phonological system of any language is a specific way of segmenting
the universal potential of phonic distinction and molding the segments obtained
into language units – ultimate phonological quanta. The segments are not
produced by selecting some ‘features’ as relevant and discarding the rest as
redundant; they are rather aggregates of several articulatory movements
together with their auditory correlates. In acquiring the sound pattern of a
language a child achieves automatic combination of the uniquely aggregated
movements, and the whole aggregate is then activated by a single neuro-muscular
impulse. The impulse is in fact the substance vehicle for the realization of
the corresponding ultimate phonological unit.
In many languages (e.g. German, French) vowel labiality and tongue
position are separate units, while in many others (e.g. Russian) they are parts
of the same aggregated unit; in the latter case there is no point in regarding
one of them as relevant and the other as redundant – they are jointly relevant
within the same unit in the given phonological system. As for the part which
each of these phonic actions plays within the aggregate, its automatic regulation
is performed at a lower sublinguistic level. In French and English the
consonantal subsystems contain ultimate units of post-centrality combining in
their aggregates the phonic features of velarity, palatality and
alveopalatality [7; 8; 11]. But the features are differently grouped and
realized in the two languages, and despite their similarity together with the
unavoidable common designation each unit is unique in being an element of a
specific language system.
Full recognition of the status of language units for the phonological
quanta calls for the creation of a suitable term, the customary designation as
‘distinctive features’ being vulnerable in its two components. To begin with,
the word ‘feature’ is incompatible with the status of a language unit in its
own right, as a feature is a mere attribute of a unit of higher rank. Indeed,
the term appeared when the phoneme was regarded as the basic phonological unit
possessing certain characteristic features. Now, when that notion has been
replaced by establishing the ultimate phonological unit as belonging to an
independent tier in the system, it must be given a designation that would
correspond to the new status and not be an adjunct to the phoneme.
Secondly, the new designation should avoid a reference to distinction as
the primary function of the unit in question. Units of every language level
fulfill that function, and all language units are equally distinctive. At the
same time they are all constitutive within higher units. Consequently, language
systems have no need for separate distinctive units, for all distinction is
achieved by the use of different constitutive elements. The ultimate
phonological unit is no exception: phonemes are distinguished by containing
different units of this level. Together with the customary designation we must
therefore decline the term ‘merism’ [2].
The best term for the unit in question was suggested by Jan Baudouin de
Courtenay at the beginning of the 20th century – the blend ‘kinakeme’ [1], containing the Greek roots for
‘movement’ and ‘hearing’ together with the suffix -eme.
Like all the other language units of every level in the macrosystem, the
kinakemes are elements in a subsystem of their own, which is naturally not a
mere inventory, but a well-structured body. Its structure displays two
principles. One is thorough binarism – all kinakemes are paired into
oppositions of positive vs. negative. Positive kinakemes are materialized as
neuro-muscular impulses to perform the respective movement or recognize the
respective auditory signal; their negative counterparts are realized in the
absence of the impulse.
The other structural principle provides for a hierarchy of tiers in the
kinakemic subsystem: it always contains two categories (modal and local) with
possible subcategories in them and with a further division into kinakemic
oppositions. The resulting variety of structural patterns is vast, so that each
language usually has a very individual organization of its kinakemic subsystem
[10; 11].
The purely negative step of discarding the obsolete notion of universal
inventories for ultimate phonological units is obviously insufficient. It must
be followed by constructive steps in two directions: first, the kinakemic
subsystems are to be described for as many languages as possible; second, a
typology of kinakemic subsystems is to be worked out to find their common
properties as well as possible diversity in them.