Common Characteristics in Phonology, Grammar, And Lexicon

7/24/2019 Common Characteristics in Phonology, Grammar, And Lexicon

1/16

COMMON CHARACTERISTICS

IN

PHONOLOGY, GRAMMAR, AND

LEXICON

The closeness of the rela tions hip between the notion of m arked

and

unmarked

in

g rammar

and in

lexicon

is

evident

from th e

fact

that the same major criteria apply in what seems to

be

intuitively

the same manner

so

that

w e

have

the

assurance that

we are

dealing

with

essentially the same phenomenon in both cases. Only the

relatively

m ino r categories of defectivation, dom inance [ taghlib ]

which m i g h t in fact be considered lexical, and agreement a

potiori

are

missing

in the

lexical area. Ev en

of

these

the first,

defectivation,

ca n

possibly

be

exemplified, once more

in

k inship systems from

the absence of terms for certain kin types in certain systems, e.g.

more distant

affinals in

English such

as

spouse s cousin.

Indeed

in

certain cases,

we may

consider

th e

same evidence from

one point of view to

exemplify

a contrast of grammatical category

and

from

another

a

lexical contra st. Th us

in the

instance

of author

and authoress cited above, the addition of

ess

may be taken as

evidence

of the

non-zero expression

of the

marked member

of the

lexical

set

author

and

authoress .

On the other hand, given th e

recurrent nature of such pairs as author/authoress,

sculptor/

sculptress, etc.,

we

isolate

an

element

ess

labelled

as

derivational

so that a generalization of the relationship noted in the single

lexical pair leads to the over-all characterization of the derivational

category

as

marked

in

relation

to the

und erlyin g category with

zero

expression.

In

contrast

to

this obvious inn er relation

of the

lexical

and

gram-

matical uses of the

concept

of marked an d unmarked

categories,

its employement in

phonology

seems a quite different matter,

At

first glance it seems by no me ans implausible to see here perhaps

no more than a tenuous metaphor, or, at best, a partial

or

complete

Bereitgestellt von | Staatsbibliothek zu Berlin Preussischer Kulturbesitz

Angemeldet

Heruntergeladen am | 06 11 15 19:00


2/16

CHARACTERISTICS INPHONOLOGY GRAMMAR LEXICON 57

isomorphism. In fact such an isomorphism can be established

through a set of correspondences of which the fundamental one

isthat of the phoneme to theword or lexemic unit, when both are

considered to be constructed from features. For the phoneme the

features

are the

familiar ones

of

phonetics,

for the

lexemic unit

it

is the constituent mo rphemes. However, this analysis into m or-

phemesmust be of a particular type as we shall soon see. There is

further

the notion of environm ent which corresponds in both

cases,

preceding

and

following phonemes, preceding

and

following

lexemicunits.

Then in either case the concept ofmarked and unmarked is a

relation between feature s whichare m utua lly exclusive where they

are the

source

of

m inim al contrast between

tw o

phonemes

or two

lexemic

units. Thus

in

phonology

the

features

of glottalization

andnon-glottalization

are

m utua lly exclusive

and

susceptible

of the

relation

of

marked

to

unmarked w here theyformcorrelative pairs,

e.g. wherewehave such oppositions as

globalized

dental stop and

non-glottalized dental

stop. In

grammar

we

deal

chiefly

with

inflectional categories where, e.g. a form cannot be singular and

plural

at the

same time

and

where

we

compare formswith

the

same

bases or class of bases existing in these two inflectional categories,

e.g. no un stem singular vs. noun stem plural. Such contrasts can

be

expressed, of course, in terms of now traditional morphem ics as

morpheme

classofnoun bases singular morphem evs.morpheme

class of noun bases + plural morpheme, but we must note then

that 1) the base has a different status than the inflection since it

is the latter that corresponds to the feature susceptible to being

marked or

unmarked;

2) the base isoften a morphemicsequence;

3) wemus tinthis caseuse atypeofm orphemic analysisinwhich

inflections

involving

different

generic

categories

simultaneously

are treated as containin g as m any morphem es as there are categories

since

we must sometimes, for example, contrast a first person

singular

withitscorresponding plural and onother occasions with

the second or third person singular.

For

lexical items, e.g. kinship terms,

the

fea tures correspond

to

the components of contemporary comp onential analysis. Here


Angemeldet



3/16

58 CHARACTERISTICS IN

PHONOLOGY,

GRAM MA R, LEXICON

there is no

necessary hierarchy among

th e

components, such

as

that of base an d inflection. The essential

difference

is that

feature

here cannot be equated with morpheme but is rather a semantic

component or seme , if you

will.

Thus

brother

is gram matically

a single morpheme, but it can be analyzed semantically into such

components

as

male ,

zero generation ,

consanguineal which

do

not themselves have m orphe mic status.

Given

these equivalences, we can attempt to translate by sub-

stituting corresponding terms from the language of phonology

into that of gram ma r or lexicon and vice versa. Let us consider the

chief criteria for unmarked and marked categories of phonology

from

this point of

view .

It

will

be recalled

that

our first phonological

criterion wasthat of neutralization in wh ich the un m arked feature

appears. Our choice of terminology in the gram m atical discussion

suggests that it is possible to equate this with contextual neutrali-

zation. And indeed, one can be mapped into the other by the

appropriate

equivalences. The terms marked an d unmarked like

environment, are,

of

course, invariant under this transformation.

Hence

w e

have

the

follow ing: when

in a

particular class

of

environ-

ment

no

contrast occurs within

a set of {{ionemes

which differ from

each other only in a single feature, it is the un ma rked fe ature w hich

appears

in this environment.

The

second phonologic characteristic, greater frequency

of the

unmarked member, is likewise subject to straightforward transla-

tion from one mode of speech to the other. In both cases we are

dealing with relative text frequencies

of

members

of set

formed

by

phonemes/lexemes which differ

in a

correlative feature,

and we

predict

the

greater frequency

of the

unmarked member.

The greater allophonic variability of the unmarked member of

a

correlative set

w as

mentioned

as a

third indicator

of

unmarked

versus marked

status.

Translating allophone into allomorph, w e

have indeed one of the criteria of the unmarked category in

grammar.

We

have seen

that in

general, though exceptions

can

be found, the unmarked grammatical category shows greater

allomorphic variation , except of course w hen, as is ch aracteristically

the

case,

it is

expressed

by

zero.


Angemeldet



4/16

INPHONOLOGY

GRAMMAR LEXICON

59

As a fourth clue to unm arked status in phonology it was

mentioned that th e number ofphonemes withthe marked feature

is always less than or equal to the number with the unmarked

featurebut no t greater. Thus the num ber of nasal vowels is always

less than orequalto the numbero foral vowels. A resolute attempt

to translate this into the language of grammatical analysis will in

fact show that it is the analogue of syncretization provided we

keep

in

mind

that as

mentioned earlier

the

correspondent

of

feature

in phon ology is really the

inflective

or derivational morpheme

that is the semanteme of European terminology. Thus the smaller

number of nasal vowels in some languages means that certain

oppositions present in the unmarked category non-nasal are

syncretized in the marked category. Thu s comparing nasalization

with

plurality as ma rked features one may say that the opposition

between highand low vowels present am ong oralvowels is syncre-

tized among the nasalized in French justas the opposition between

masculine and

feminine

is sync retized in the plural of the article

the

demonstratives

and the

possessive adjectives

in the

same

language.

The fifth and last indicator m entioned for distinguishing m arked

from unmarked inphonology was that th ebasic allophone defined

in

terms of phonologic independence of its environm ent was the

one with the unm arked feature. The translation of this statement

into grammatical termin ology requires tha t w e find an equivalence

to independence in relation to env ironm ent. N ow itwillbe recalled

that by independence in this case was meant non-assimilation

phonetically

to

adjacent sounds.

A

sound

is

assimilated

to

another

it

it

shares more features

with it.

Similarly

a

lexeme

may be

said

to be

assim ilated

to

another lexeme

ifit

shares

an

additional feature

with it meaning in this connection as has been seen a semanteme.

Now the sharing of semantemes in grammar is concord. Hence

we

m ay

equate

the phonological character just m entioned

with

agreement potion Thus in Spanish the adjective agrees with the

noun it modifies in gender i.e.it shows a common semanteme.

The unmarked masculine is however more independent of its

environment in

that

it may be used in a Spanish expression such


Angemeldet



5/16

6 CHARACTERISTICSIN PHONOLOGY GRAM MAR LEXICON

as cuello y camisa blancos

white

collar (masc.) and shirt (fern.)

where

blancos

which contains the masculine morpheme, appears

in th e environment of

camisa

shirt

which

is feminine while th e

feminine

morpheme

of an

adjective could never appear except

in

the

environment

of

a nother feminine.

The

possibility

of

translation for every

one of the five

character-

istics of the unmarked/marked dichotomy in phonology as enu-

merated earlier

into

grammatical terminology under fixed rules of

translation

and

with unmarked

and

m arked corresponding

to

each

other in each case is sufficient evidence that the analogy between

these concepts

in

phonology

and in

grammar

is not a

far-fetched

one. As

will

be developed in more detail later, what connects the

uses of the term unma rked and of marked in at least some of these

statemen ts with each other,and in corresponding ways in phonology

and grammar is the basic o r fundamental character of the unmarked

as against the marked. This can be shown more exactly, in the

following

way:

whenever

a

statement

of one of the

above

five

types

can be put in

terms

of a

universal implication,

it is the

unmarked

member

which

is the

implied

or

basic term

and the

marked which

is

the implying or secondary. Thus to the first type statement that

gendermay be syncretized in the marked category, i.e. in the plural

an d the phonologic statement that the opposition between high

and low vowels may be syncretized in the presence of nasality,

we have the

implicational universals:

(1) distinction of gender in

the

plural implies

its

distinction

in the

singular

but not

n ecessarily

the converse; (2) distinction of vow el height in nasal vowels implies

its presence in oral vowels but not necessarily the converse. In

both of these statements the

implicatum

is the unmarked category,

singularity

and

non-nasality respectively.

Viewed psychologically there is perhaps justification fo r seeing

a similarity between the implied, fundamental characteristic, that

is the unmarked member, whether in phonology, grammar, or

semantics, and the

Gestalt

notion of ground, the frequent, thetaken-

for-granted, whereas

th e

marked character would answer

to figure

in

the familiar dichotomy. It may be noted in passing that the

traditional arrangement

of

paradigms

in

gramm ars seems

to

display


Angemeldet



6/16

CHARACTERISTICS

IN

PHONOLOGY

GRAMMAR

LEXICON

61

an intuitive

recogniton

of these relationships. The singular is

always put above or in the

left

hand column, so

with

the active

versus the passive, etc.

It is

time, however,

to

turn

to

points where

th e

isomorphism

is

perhaps not complete. Specifically, we can ask w hether the a ddi-

tional characteristics of the unm arke d/m arke d in the gramm atical

and semantic spheres to which no correspondent has yet been

mentioned from phonology

do

have such

a partner

and,

if

they

do

not, to seek

for

an explanation of the impossibility of a m app ing

in these cases.

The

characteristics involved

are the following:

zero expression,

facultative

expression,

defectivation, and

dominance

taghlib).

This

last

can be

eliminated

as a

relatively minor phenomenon.

In

fact,

it only applies to the category of number since it refers to the

characteristics

of a

collection

and is

therefore irrelevant

to the

analyses of most gramm atical and sem antic categories. Defectiva-

tion also raises

no

real

difficulties. It was

seen that defectivation

is closely related to the concept of syncretism. In fact it might be

considered a variety of syncretism in which the representative of

the syncretized category can be definitely identified w ith a particular

member

o f that

category.

The

others

can

then

b e

said

to be

lacking

or

defective. Thus

if

gender

difference

is

syncretized

in the

plural

but the single gender present can, on some grounds, be

identified

as

masculine, then

the

feminine plural

is

missing. Similarly

we

can say

that

the oral

high vowels

in

French have

no

nasal partners

so thatthere is defectivation in the m arked category of nasal vowels.

This, however, leaves two conspicuous indicators of the marked/

unmarked in grammar an d semantics facultative expression of

the marke d and zero expression of the unm arke d. Itwill be recalled

that what has been called here facultative expression is given

definitional status

by

Jakobson

in his

discussion

of the

marked/

unmarked dichotomy in relation to grammar. The analogy to

phonology h as already been pointed out namely,thatth e unmarked

member acts

as a

surrogate

for the

entire category. However,

as

was

just pointed o ut,

the

more exact analogy

of

phonological

neutralization

is

contextual neutralization.

The

comparison

of


Angemeldet



7/16

6

CHARACTERISTICS

IN

P HONOLOGY GRAM MAR LEXICON

phonological neutralization howe ver to facultative expression does

serveto point out the important ambiguity of the unmarked terms

in

grammar

and

semantics

as a

simultaneous bearer

of the

generic

category meaning and the specific unmarked subcategory and the

similar

am biguous role of the archiphonem e. However it rem ains

to be

pointed

out

that,

a s

important

as the

phenomenon

of

faculta-

tive expression is it does not in fact apply for a number of the

grammatical categories mentioned.

These include

the

important categories

of

positive/negative

and

declarative/interrogative. Thus it is not true that a statement in the

positive form that is with out an overt indicator of negation can

be taken as either positive ornegative and is merelyp r excellence

positive. Likewise in a language w ith a question particle it is not

the case that

the

absence

of

this particle indicates tha t

the

sentence

can be

taken

a s

either de clarative

or

interrogative.

The

same holds

for

the lexical case of adjectival opposites. Ag ain it is not tru e th at

the unmarked member

wide can

also mean

narrow

saving

indication

to the

contrary.

A

further instance

is

lower versus higher

numerals.

This of course leaves us with the alternative of excluding cases

such

as

those just cited

from

consideration

as

instances

of

marked

and unm arked categories and setting up perhaps a

still wider

notion

under which these cases

can be

subsumed along

with a

different

subset including true instances

of the

marked/unmarked dichotomy.

However

the

instancesjust considered

are so

like

the

other cases

that this seems inadvisable.

It

thus turns

out

that

one of the two

remaining types which does

not

seem

to

have

a n

exact respondent

in

the

area

of phono logy is not itself

universally

present in gramm ar

and

semantics. Th is still leaves

us

with

the

very important gram-

matical indication

of the

unmarked category

by

zero expression.

Here,

however

a

literal transla tion

at

least usin g

the

equivalences

which

showed themselves to be

efficacious

in instances considered

above is not

possible.

For zero expression involves the relation

between content the grammatical or semantic category involved

and

expression

in

this case

the

lack

of overt

sound

seq uences.

A t

this point

the

fundamental

difference

between

the phonological


Angemeldet



8/16

CHARACTERISTICS IN PHONOLOGY GRAMMAR LEXICON 63

an d

grammatical level

asserts

itself,

the

sound-meaning

relationship

which

is absent in the former and present in the

latter.

Up to now the criteria o f the m arked and unm arked, whether in

phonology or gram m ar/sem antic has been treated as an empirically

given

bundle of concurrent phenomena; that is, such questions as

the following have not beenasked. Why, for example, should the

less

frequ ent category be the one w hich is subject to syncretizations?

The following

remarks

are to be

taken merely

as

exploratory

soundings.

Consider first the situation in phonology. Here the funda m ental

factor

is quite possibly consitituted by certain dynamic

diachronic

factors. Of these the chief would be th e tendency for a more

complex (marked) item to lose its mark whenever it no longer

contrasts with the corresponding unmarked item.

Thus

in the

presumed course of events embodied in Grim m s first law, once

unvoicedstops

had

become fricatives,

w e

would

be

left with such

sets as

b

b

b

t

f. The b with its marked feature of voicing having

no

partner

p w as free to

lose

its

mark

and

become

p. Now

given

ft

h

,p f in

similar fashion

th e b

b

having

no

partner

b

could lose

its marked feature of aspiration, although it became a voiced

fricative rather

than

a

voiced

stop in

most environments.

In

this

schematic statement various complications are not considered,

notably those concerning V erner s law .

The

Grimm s

law

changes

do

not,

in

general, involve merger.

In

other

cases of

complete

or

conditioned merger under conditions where, typically but not

always,

functional yield

is low, it seems to be the general rule

that

the merger is produced by the marked feature losing its mark.

Conditional mergers

will

evidently produce neutralization. Thu s

in German

an d

other languages voiced

and

unvoiced obstruents

have m erged in w ord o r sentence final by the loss of voicing in this

position.

Of

course not all sound changes

operate

in this direction. For

example, by assimilative changes a complex m ay acquire a marked

feature of an adjacent so und as in assimilative voicing. There are

further sources of phonemes with marked features. An important

one is surely the deve lopment of complex articulations from previous


Angemeldet



9/16

6 CHARACTERISTICS IN PHONOLOGY GRA MM AR LEXICON

sequences.

A

typical instance

is

nasalized vowels, conjectured

by

Ferguson to arise in all cases

from

sequences of oral vowel and

nasal consonant. In such instances it would presumably be the

case that first the oral vowel is nasalized non-distinctively before

the nasal consonant, and the consonant is subsequently lost.

In their relation to marked and unmarked features then, two

major classes

of

regular sound changes

m ay be distinguished. The

first includes unconditioned changes, particularly mergers, and

those conditioned changes in which the specific class of environing

sounds is irrelevan t, e.g. changes in word final. In these which m ay

be assigned to the paradigmatic aspect of language the overall

tendency

is for the marked or phonetically complex series to give

way

to the

unmarked

or

simpler. Thus

i t may be

asserted

as a

diachronic universal that a glottalized series may merge

with

the

corresponding unglottalized series in an unconditional merger

but

n ot vice versa. If the opposite occurred it w ould produce a

phonological system

in

which glottalized consonants occurred

without

an

unglottalized series

an d

such

is not

known

to

occur.

The o ther class o f changes which m ay be considered syntagm atic

consists

of the mass of assim ilatory conditioned changes which

often

give

rise

to m arked features. Thu s the answer to the objection

that

ease*

of articulation, an expression which is avoided here,

but which can be given objective content should produce constantly

simpler phonologic systems in the evolution of language is that

there are two kind s of

ease ,

paradigm atic wh ich favors sim plifica-

tion by loss of additional articulatory f eatures regardless of context

and syntagmatic which favors the genesis of new assimilatory

modifications conditioned

by the

phonetic environment

and so

gives

rise

to

articulations wh ich taken

in

isolation

are

m ore com plex.

The greater frequency of the unmarked set can be largely ex-

plained

as a

resultant

of the two

processes just described.

In

positions

of

neutralization only

the

unmarked member appears.

Where

a set of

marked phonemes arises

from a

sequence,

the

original frequency

of the

undifferentiated protophoneme would

presumably be smaller before the limited set which furnished the

second members of the sequence, and this lesser frequency will


Angemeldet



10/16

CHARACTERISTICS

INPHONOLOGY

GRAMMAR

LEXICON 6

be

reflected

at a

later stage

by the

correspondingly smallerfrequency

of the

marked set. Thus

in

Latin

th e

frequency

of any

vowel before

al l the non-nasal phonemes was presumably greater than before

the nasals alone. This same hypothesis

will

also explain another

characteristic of themarked category in phonology;namely that

the

number

of

marked phonemes

of a set of

correlative pairs

is

usually

less than or equalto thenumberofunmarked. When they

arise

in

thisfashion they will

in the

beginning

be

equal

in

number.

They may then decrease by mergers as withthe French nasal

vo-

wels Given their initially smaller frequency their functional yield

with

each other

is

necessarily small.

A further

psychological

factor

is the

probably greater acoustic similarity

of

sets

which

share

a

marked feature

as

against

an

unmarked feature.

In a

psycholinguistic experiment of Greenberg and Jenkins subjects

judged each pair distinguished

by

voice

as

closer together than

correlative pairs distinguished by

voicelessness.

1

Thus bid w as

closer than p:t; b g than p:k etc. It is remarkable to cite the

exampleofnasality as themarked feature thatachangem > nis

not uncommon butb > dorp > t is practically unheard of.

The

greater frequency

of the

unmarked then would

be a

resultant

of

certain common diachronic factors. Where other diachronic

factors

are at

work however discrepancies

may

arise. Thus

as

w as pointed out some languages have

a

larger number

of

long

vowelphonemes than short vowels becauseof thecommon mono-

phthongization of the

diphthongs

ajand ay. Of

course

S and

having

no

short partner

may be

expected

to

become shorter,

but

various morphological

or

canonical form factors

may

serve

to

maintain length. For these reasons while there is a far better

chance tendency not onlyfor thetotal text frequency of an un-

marked set to begreater than thatof the corresponding marked

but evenfor each individualpair,thereareoccasional exceptions.

While

frequency

is

thus merely

a

resultant,

though

a

very

im-

portant one

of

overall diachronic tendencies

in

phonology

it is

tempting

to

adjudge

its

role

in

grammar-semantics

as

primary.

J. H.

Greenberg

and J. J.

Jenkins, Studies

in the

PsychologicalCorrelates

of

theSound SystemofAmericanEnglish . Word

20.157-177

[esp. 177](1964).


Angemeldet



11/16

CHARACTERISTICS IN PHONOLOGY GRA MM AR LEXICON

There is a real

difference

between frequency phenomena in phono-

logy

and in the grammatical-semantic sphere. For the former,

we

do n ot choose our expression in terms of sounds, except perhaps

marginally in

poetry

so

that phonologic frequenc y

is an

incidental

characteristic which bears the marks of

past

diachronic changes.

But we make grammatical and semantic choices based on the

momentary

situation. It is therefore plausible, inso far as there are

constants in the human situation, that, for example, everywhere

the

singular should

be

more frequent than

the

plural

and

that this

remains

quite

constant over

time

in

spite

of

changes

in the

means

of

expression. Hence also generalizations regarding

relative

phoneme frequencies are more

precarious

and exceptions are to

be

expected. De Saussure here, perhaps anac hronistically inter-

preted, had a real insight where he has sometimes been judged to

be

obviously wron g; nam ely, in his identification of the diachronic

with the phonological and the synchronic with the grammatical.

The

important phenomena

of

zero

and

f acu ltative expression

c an

be understood in terms of frequency phenomena based on the

situation in the world w ith which the users of language must deal.

In fact there is here no real

difference

between semantic and gram-

matical phenom ena. For example, it is not so muc h in English

that

m ale is in general the u n ma rked category in relation to

female,

but the

frequency

of

association

of

things

in the

real world.

Author

means

facu ltatively a

writer

of

either sex, butpar excellence

male,

because

in

fact most authors

are

male.

W e see

this

if we

compare the term

nurse*.

Since nurses are usually female, nurse

takes

on the meaning of nurse in general,

or

non -male nurse. To

express the

maleness

of the

nurse, when

relevant, we use the

marked

expression male nurse . Just so we may compare the ordin ary

semantic interpretation

of

words with

or

without syntactic mod ifiers

with

the

morphological expression

of

corresponding categories.

In a

language without

a

grammatical category

of

d iminutives

an d

augmentatives,

where size

is

indicated

by

modifying adjectives,

if

we

use house in a

sentence w ithout mod ifiers,

the

size

is unspecified

but the

house

may in

fact

be

unusually large

or

unusually small.

We will usually assume that

it is of

normal size because most


Angemeldet



12/16

CHARACTERISTICS IN PHONOL OGY GRA M M AR L E XICON 67

houses are of normal size. On the other hand, small house or

large

house exclude explicitly

from

interpretation as normal size.

The frequ ent assimilates the ambigu ous, save contrary indications.

There

are

other advantages

to a

frequency interpretation

of

marked

and unmarked in grammar and sem antics by

w hich

marked

simply means definitionally less frequent an d unmarked means

more frequent. To begin w ith there is the obvious methodological

advantage that frequency phenomena can be explored fo r every

language w hereas the other criteria are mo re limited in this

respect,

e.g. neutralization of certain subcategories m ay not exist in a given

language. Frequency data will allow of degrees of marked and

unmarked by

which

the

associated phonemes

will

be

expected

to

be m ost co mm on and least subject to exception w here the frequency

disparity

is the greatest. T his indeed seems to be the case inso far

as, for example, the hierarchy of persons is both less certain and

overwhelming in regard to frequency and also less clear in other

matters, whereas the hierarchy of numbers shows

almost

no

exception

in

non-frequency phenomena

and

great constancy

together with large frequency disparity

fo r

singular, plural,

and

dual.

In addition to g radualizing and quantifying th e scale, it

also

allows the construction of a much more subtle and manifold

hierarchy,

fo r example, for the cardinal and ordinal numbers.

In addition the frequency definition will cover at least one

case

in which none of the other criteria is present but which has been

considered as an example of the marked/unmarked distinction by

Jakobson; nam ely, norm al (unm arked ) versus emph atic (marked)

word order. The so-called normal order, it would seem, is neces-

sarily

th e

most frequent.

We may

refer

here

to the

wel l -know n

story of the boy who cried wolf.

Finally it may help to overcome the problem of lack of inter-

linguistic com parability of categories. Thus, for gender categories,

we

may at least conjecture that the associated phenomena such as

zero expression

and

neutralization

will be

present

to the

degree

that frequency

differences

exist amo ng the genders. Since these

are largely or completely conventional

semantically

and

differ

in

size

of mem bership, it is entirely plausible that the gender labelled


Angemeldet



13/16

8

CHARACTERISTICS

IN

PHONOLOGY GRA MM AR LEXICON

masculine in one language

w ill

be of much greater text frequency

than th e

feminine

in that language,

while

in another language,

the

relationship

is

reversed.

W e may

hypothesize that

in the first

language the

masculine will display

the

other characteristics

of the

unmarked

category, w hile

in the

second

it

w ill rather

be the

feminine.

Where the categories a re not con ven tiona l , e.g. for cases, the

w ay lies open to explain the frequencies of specific cases as a sum-

mation of anumber of discrete uses, each substantially similar in

frequency among languages but

differently

combined in different

languages.

For

example, traditional grammar describes

the

uses

of

the ab lative in L atin un der such rubrics as the ablative of personal

agent, separation, in strum ent, etc. If w e had the frequencies of each

of these, w e could then, fo r example, compare it

with

the Russian

cases

by

equating

a

component

of

separation w ith

the

genitive

w ith

prepositions o

and

zw hile agent

and instrument

w ould

be

equated

with the Russian instrumental.

There is at the mom ent a great practical

difficulty

here, of course,

as

well

as the theoretic problem s of sampling. It is rare to have

frequency studies of grammatical categories, and even these do not

specify

the separate uses of the categories But th is can in principle,

of

course,

be

overcome

in

order

to

test

the

hypotheses presented

here.

The connection betw een freque ncy and the phenomena of gram-

ma tical or sem antic neutralizatio ns and morphological irregularities

has not yet been discussed. It has

often

been noted that the most

frequent

form s are the most irregular. These are indeed now by

our

definition

th e

unmarked forms.

Where there

is a

complex

set of

intersecting categories,

the

frequency

differences betw een comb inations

of

unm ark ed categories

and of marked categories are very

great

For example, in

Avery s

study of the Rigvedic verb, the fo rm w hich involves all of the most

unmarked

categories, singular, third person, present, active,

indicative

has

1404 occurrences, w hile

the

dual, second person,

medio-passive perfect optative has zero frequency . Such enorm ous

disparities

must surely have an effect in thet such a highly infrequen t

formation must follow analogically other partsof the system, w hile


Angemeldet



14/16

CHARACTERISTICS

IN

PHONOLOGY G R A M M A R LEXICON

69

only a

fairly

f requent

fo rm

can preserve irr egulari ties. Hence also

syncretismsproduced

by the

accidents

of

sound change

wil l

i n

such

cases not lead im m ediately or inevitably to new f or m ation s to

reintroduce

th e

lost distinctions. Thus

th e

general course

of the

reduction of the

case

system in Indo-European languages leads

to the coalescence of the marked oblique cases and whe re th e

whole structure finally collapses it seems to be one of the direct

cases no m inative or accusative w hich is the historical source of

th e

nouns

now undifferentiated for case.

Thus

in

phonology

diachron ic process explains frequency whi lein g r am m ar

frequency

explains diachro nic process. Frequency

not

included

in l

l ngue

definitionally

is in

fact

an

ever present

and

poweful

factor

in the

evolution of

grammatical categories

and

thus helps

in

explaining

the types of

synchronic

states

actually f oun d.

That

such things happen

is not to be

w ondered

a t.

Though

w e

m ay

justifiab ly

define

our

subject

in a

coherent

and

con sistent w ay

the world is under no obligation to

respect

these boundaries and

it is a com m onplace thatw em u stoften b r ing in external explanatory

factors.

A particular type of connection between marked categories in

phonology and

g rammar

m ay be

pointed out

and its

explanation

will

now be clear on the basis of the above

considerations.

Some-

times

the

marked category

in

phonology

is the

expression

of a

m arked category in gram m ar. Thus certain Am erind languages

use

the

marked feature

of glottalization to

express

the

marked

grammatical category of the diminutive. In German umlauted

vowels

may be considered a marked phonetic category as against

their non-um lautedpartners. Rounded fr on t vowels alw ays im ply

rounded back vowels in a

particular

language; their number is

never

greater

and their text frequency is generally

less.

Umlaut

is used in German as a gramm atical

process

to expressthe marked

categories

of plurality in the noun comparative and superlative

in the adjective and

past

subjunctive in the verb. These phenom ena

result f rom zero expression of the unmarked where a phoneme

involved in the expression of the marked disappears after having

modified th esim ple preceding soun d to producea m arked complex


Angemeldet



15/16

7 CHARACTERISTICS IN PHONOLOGY GRAMMAR LEXICON

sound, e.g. umlauting produced by a former

or

glottalization

from

a former

glottal stop

Another example of ph onological-gramm atical connection is the

Widespreaduse of the marked category of final rising pitch for the

expression of interrogation. Here the problem is somew hat

different in

that since

the

intonational pattern

has

this meaning

directly we may seem to be tautologous in asserting that the less

usual intonation

expresses the less

frequent

category

However,

there

is

further independent evidence

for the normality* of tonal

descent in that phonemes of pitch often have progressively lower

allophones the later they occur in the sentence, but the ph enone non

of allophonic raising never seems

to

occur

If it

turns

out

that

in

fact frequency

is an

adequate unifying

principle for the

domain

of the

marked

and

unmarked

in

semantics

and gramm ar, a

great

over-all sim plification will have been achieved.

But frequency is itself but a sym ptom and the consistent relative

frequency relations which appear to hold for lexical items and

grammatical categories are themselves in need of explanation.

Such

explanations will not, in all probability, arise from a single

principle. Thus it may b e noted

that

in adjectival opposites where

a theoretical scale with an implied zero point is unmarked, e.g.

heavy, large, w ide, deep, etc., there is obviously a

unifying

principle

but itwill not even apply to all adjectival opposites, e.g. good/bad,

and

is irrelevant in a host of other exam ples. Again the center

of

a normal frequency distribution is unmarked in relation

to

the

extremes, e.g. normal size as against diminutive or augmentative.

This topic is

left

fo r future exploration.

In phonology, a third level principle which, while requiring

further refinement,

is

evidently

sufficient to predictfor a

w ide range

which

features will

be

marked

an d

which unmarked

is articulatory

complexity which is correlated with acoustic complexity.

This

can

be

defined

in an

objective man ner independently

of the

distribu-

tional and frequency phenomena employed here to distinguish

marked and un ma rked categories. A particular articulation is to

be

considered more complex than some other if it includes an

additional articulation defined in terms of departure of an organ


Angemeldet



16/16

CHARACTERISTICS

IN

PHONOLOGY,

GRAMMAR ,

LEXICON

71

from

the position it normally has in the absence o f speech. This

notion

can be

extended

to

include successive add itional articula tions

in thecaseoflengthand diphongization.

An

appa rent exception is nasality. Aco ustically the nasal is

more complex in that it involves additional nasal resonances but

from the

articulatory

view it seems to be superficially the oral

articulation that is complex since it requires a raising of the velum .

Note however

the remarks of Heffner

regarding nasal vowels.

The contraction of the

pillar

of the fauces is a feature of the pro-

duction of nasal vowels and ... nasal vowelsare produced by

adding the vigorous low ering of the velum , accom panied by some

constriction of the palatopharyngeal arch, to theusual movem ents

ofarticulation peculiar to the analogous oral vowel.

2

R-M.

S.

Heffner,

eneral

phonetics Madison, 1964),

31,

113.

Common Characteristics in Phonology, Grammar, And Lexicon

Documents