Thus compositionality may work only partially with some stratum 1 affixes: we may be able to work out
their morphological and phonological properties but their semantic contribution to a word may be
idiosyncratic. Where that is the case (cf.
Protestant)
, listing in the dictionary will be needed, despite the
residual element of semantic predictability.
The third point concerns PRODUCTIVITY. The relative productivity of morphemes is reflected in the
hierarchical organisation of strata. It is significant that in our example
-er, which is a phonologically
neutral, stratum 2 suffix, is more productive than
-ant, the non-neutral stratum 1 suffix. This is typical.
Stratum 2 affixes are typically more productive than their stratum 1 counter-parts. Thus, the stratum on
which a morpheme appears in the lexicon reflects its productivity (cf. Aronoff 1976:35–45).
Turning to inflectional morphology we can make the same points. The hierarchical organisation of strata
can be used to account for the fact that in INFLECTIONAL ALLOMORPHY a
particular stem does not co-
occur with all the available affixes. We shall illustrate this with plural formation. In addition to the regular
plural with the phonologically conditioned allomorphs [-z, -Iz, -s] (as in
birds, classes and
bricks: cf.
section (
5.2
)), English has several other
ways of forming the plural, including the following:
[6.10]
a.
Singular
Plural
b.
Singular
Plural
foot
~
feet
deer
~
deer
tooth
~
teeth
sheep
~
sheep
goose
~
geese
salmon
~
salmon
[6.11]
a.
Singular
Plural
b.
Singular
Plural
locus
~
loci
addendum
~
addenda
stimulus
~
stimuli
curriculum
~
curricula
fungus
~
fungi
stratum
~
strata
The use of internal root vowel change to signal plural as in [6.10a] is assumed to take place at stratum 1
since this is obviously a phonologically non-neutral, stem-changing process. This particular kind of plural
formation, which is rare, cannot be successfully treated using normal suffixation rules. But it can be dealt
with better if we assume an analysis like the ‘word-and-paradigm’ one for Latin in section (
3.6.3
). In this
account there would be a rule which simply says, for example, that the lexical item
foot, when associated
with
the grammatical property plural, is
realised by the word-form feet.
By contrast, the zero suffix causes no problems. There are no phonological upheavals in the stem; we
place it at stratum 1 in order to reflect the fact that it is a very restricted and unproductive method of
pluralising nouns. Unless a noun is expressly flagged in the lexicon as taking a zero plural suffix, we
assume that it does not (cf.
section 5.3
). The formation of the plural of the nouns in [6.10] at stratum 1
blocks regular plural formation by attaching the regular
-s suffix at stratum 2. That way, the formation of
disallowed
plurals like * foots and
* gooses or
*deers and
*sheeps is blocked.
Similarly, the plural suffixes
-i and
-a in the words in [6.11] which come from Latin also have to be at
stratum 1. The suffixation of
-i blocks the suffixation of
-s. So, we get only
loci not
*locuses. Thus, the
notion of blocking enables us to explain why the formation of plural at stratum 1 hinders the formation of
the plural using the regular [-z, Iz, -s] allomorphs. By the time they reach stratum 2, the words in [6.11] are
already marked as plural and so are unavailable for the suffixation of regular [-z, Iz, -s].
76 A
LEXICON WITH LAYERS
However, because many speakers of English are unaware of the Latin origin of words with the
-i plural,
blocking is not any more successful in inflectional morphology than it is in derivational morphology.
Commonly used words of Latin origin such as
virus and
bonus have been ‘regularised’ so that they only
take the
-s plural (i.e.
viruses, bonuses). In other cases, such as
syllabuses ~ syllabi and
aquariums ~
aquaria,
usage fluctuates between the Latin
-i or
-a and native
-s. Some speakers use the stratum 2 suffix
-s,
bypassing the rule that attaches the stratum 1 Latinate one, while others use the rule that attaches the
Latinate stratum 1 suffix and thereby preempt the
suffixation of the regular -s suffix at stratum 2.
As mentioned with respect to the zero plural suffix, the placement of inflectional affixes on different
strata reflects their relative productivity. The borrowed stratum 1 plural allomorphs
-i and
-a are not
productive. If a new noun ending in
-us or
-um entered the language, its plural would be formed with the
regular
-s ending rather than
-i or
-a.
Do'stlaringiz bilan baham: