Professional Documents
Culture Documents
24.0 PP 1565 1620 Inflectional Morphology and Related Matters
24.0 PP 1565 1620 Inflectional Morphology and Related Matters
Inflectional morphology
and related matters
Frank Palmer
Rodney Huddleston
Geoffrey K. Pullum
1 Preliminaries 1567
1.1 Lexical base, morphological operations, and alternation 1568
1.2 Overview of inflectional categories 1569
1.3 Speech and writing 1570
2 General phonological and spelling alternations 1572
2.1 Phonological alternations 1573
2.1.1 The sibilant suffixes: /iz/ ∼ /s/ ∼ /z/ 1573
2.1.2 The alveolar plosive suffix of the preterite and past participle: /id/ ∼ /t/ ∼ /d/ 1573
2.1.3 Bases ending in syllabic /l/ (/hmbl/ ∼ /hmbliŋ/) 1574
2.1.4 Bases ending in post-vocalic /r/: alternation in non-rhotic accents (/reər / ∼ /reərə/) 1574
2.2 Spelling alternations 1575
2.2.1 Consonant doubling (bat ∼ batt·ing) 1575
2.2.2 E-deletion (like ∼ lik·ing, subdue ∼ subdu·ing) 1576
2.2.3 Y-replacement (silly ∼ silli·er, try ∼ trie·s) 1578
2.2.4 Alternation between ·s and ·es in the plural and 3rd sg present tense 1579
3 Grade 1580
3.1 Inflectional comparative and superlative forms 1581
3.2 Inflectional vs analytic comparatives and superlatives 1582
4 Nouns 1585
4.1 Plural formation 1585
4.1.1 Regular ·s plurals (cats, dogs, horses) 1585
4.1.2 Irregular ·s plurals: modification of the base (wives, mouths, houses) 1587
4.1.3 Base plurals (cod, bison, series, Chinese, craft) 1588
4.1.4 The vowel change plurals (teeth, mice, men) 1589
4.1.5 The ·en plurals (oxen, children, brethren) 1590
4.1.6 Foreign plurals ( formulae, curricula, phenomena, crises) 1590
4.1.7 Compound nouns (grown-ups, commanders-in-chief ) 1594
4.1.8 Proper nouns (Joneses, Marys) 1595
4.2 The genitive 1595
5 Verbs 1596
5.1 Regular forms 1596
5.2 Irregular present tense forms 1599
5.3 Irregular preterite and past participle forms 1600
5.3.1 Class 1 verbs: secondary ·ed formation (burn, keep, hit, lose) 1600
5.3.2 Class 2 verbs: vowel alternations (drink, dig, find, come) 1603
5.3.3 Class 3 verbs: past participles formed with the ·en suffix (see, ride, take) 1604
5.3.4 Class 4 verbs: other formations ( flee, hear, stand, buy, can) 1607
5.3.5 Index to the classification 1608
5.4 Verbs with complex bases (underpin, become) 1609
5.5 Negative forms of auxiliaries 1610
1565
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1566
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1567
1 Preliminaries
This chapter and the next are concerned with morphology, that part of a grammar that
deals with the form of words. As explained in Ch. 1, §4.3, morphology is divided into
two subcomponents: we look first at inflectional morphology and then in Ch. 19 turn to
lexical word-formation.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1568 Chapter 18 Inflectional morphology and related matters
Morphological operations
The plural forms dogs, workers, and dregs are formed by adding ·s to the lexical bases.
We refer to this as a morphological operation – specifically, the operation here is that of
suffixation. This is in fact the major type of operation involved in English inflectional
morphology. However, it is not the only one. The plural teeth, for example, is formed
by changing the vowel of the base tooth, and the same applies to the preterite form
rang, formed from the base ring. In the case of knives, suffixation is accompanied by
modification of the base (with voicing of the final consonant in speech and corresponding
replacement of f by v in writing). Other, relatively minor, operations will be introduced
as they are needed.
Present-day English has a very simple system of inflection – much simpler than Latin,
for example, or indeed than earlier stages of English such as Old English. There are
few inflectional categories and relatively few types and combinations of operation are
involved in their formation. Most inflectional forms are either identical with the lexical
1
We use the notation ‘·’ when citing suffixes (like ·er) or prefixes (like un·) and, where relevant, for marking
morphological divisions within words (dog·s).
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 1.2 Overview of inflectional categories 1569
base or formed directly from it, as in the above examples. There are some, however,
where two steps are involved. One obvious example is the genitive plural, with children’s,
say, formed by suffixation from the plural children, not from the base child. A second
example is provided by a small subset of past participles like trodden: this is formed not
from the base tread, but from the preterite form trod, by suffixation of ·en (with doubling
of the final d in the spelling).
Alternations
We speak of alternation when a morphological unit has different realisations depending
on the context in which it appears. For example, the plural ending is realised in writing
as s in cats but as es in boxes: there is alternation between the suffixes ·s and ·es in the
formation of plural nouns. A good deal of this chapter will be devoted to the description
of such alternations.
Syncretism
When two or more inflectional forms of a lexeme are pronounced or spelled alike, we
say that there is syncretism between them, or that they are syncretised. To return to the
example used above, sheep has syncretism between the singular and plural forms.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1570 Chapter 18 Inflectional morphology and related matters
Pronouns
Most of the personal pronouns have nominative and accusative case forms, separate
dependent and independent genitive forms, and reflexive forms in which number func-
tions as an inflectional category in all three persons. The nominative–accusative contrast
is also found with relative/interrogative who. The forms have been listed in Ch. 5, §10.1.1.
The reflexive forms are morphologically compounds, formed with self (singular) or
selves (plural) combining with the dependent genitive in the 1st and 2nd persons, and
with the accusative in the 3rd person. There are no other significant morphological
generalisations to be made about pronouns, so they are not dealt with further in this
chapter.
(b) Grade
The system of grade applies to many adjectives and a few other lexemes:
[3] plain comparative superlative
weak weaker weakest [adjective]
soon sooner soonest [adverb]
The plain form is identical with the lexical base, and the inflectional comparative
and superlative are formed, for the bases that permit inflection, by simple rules of
suffixation, with very few irregularities in either speech or writing. Our discussion,
in §3, also deals with the distinction between, on the one hand, inflectional compara-
tives and superlatives, such as those in [3], used with a large but restricted class of bases,
and, on the other, analytic ones, such as more careful, most careful, used with all other
bases.
(c) Verbs
For the great majority of verbs six inflectional forms must be distinguished, as argued
in Ch. 3, §1, and illustrated here for take:
[4] plain 3rd sg plain gerund- past
present present preterite form participle participle
take takes took take taking taken
The plain present tense and the plain form are identical with the lexical base: be is the
only verb without syncretism between the base and a present tense form. In addition to
the categories in [4], auxiliary verbs have negative forms.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 1.3 Speech and writing 1571
Primacy of speech
In some cases the spoken rules are clearly primary, the written ones derivative. One very
obvious example concerns the alternation between ·es and ·s in forming the plurals of
words ending in a consonant:
[6] i gases boxes buzzes bushes churches stomachs
ii /gæsiz/ /bɒksiz/ /bziz/ /bυʃiz/ /tʃ
r tʃiz/ /stməks/
Bases ending in s, x, z, sh, and (usually) ch take ·es rather than ·s, but this reflects the
fact that the spoken forms have /iz/. In speech the presence of the vowel /i/ in the suffix
depends on the phonological properties of the base. If the immediately preceding sound
is one of the subclass of consonants called sibilants, comprising /s/, /z/, /ʃ/, //, /tʃ/
and /d/, then the vowel /i/ is required in the suffix. The presence of e in the written
suffix, on the other hand, depends on how the suffix is pronounced in speech. Note, for
example, that while most bases ending in ch take ·es, as in the above churches, there are a
few that take ·s, as in stomachs, epochs, eunuchs. The choice depends not on the spelling
but on the pronunciation: in the former case ch corresponds to sibilant /tʃ/, so that the
suffix in speech is /iz/, whereas in the latter case it corresponds to non-sibilant /k/, which
takes /s/ as the suffix in speech. There is nothing about the letters ch (or indeed s, x, etc.)
that calls for ·es : it is simply a matter of matching the pronunciation.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1572 Chapter 18 Inflectional morphology and related matters
the preterite (and past participle) of pay is regular in speech, but irregular in writing,
where y is changed to i and the suffix is ·d, not regular ·ed (contrast the regular play ∼
played). Again, the plural of house is regular in the written form, but irregular in the
spoken, since the final consonant of the base is voiced in the plural, changing from /s/ to
/z/. And while money is regular in speech, in writing there is variation between regular
moneys and irregular monies (irregular because replacement of y by i normally applies
only if y is preceded by a letter representing a consonant).
In ordering the material in this chapter we have given priority to the spoken
forms: the sections on regular plurals and verb-forms deal with those that are regular in
speech, and cover the corresponding written forms whether regular or not.
Forms that are irregular in both writing and speech can usually be assigned to the
same subclasses. The verb read is an exception, however, as we see from these preterites
and past participles:
[8] i meet ∼ met hit ∼ hit read ∼ read
ii /mi
t/ ∼ /met/ /hit/ ∼ /hit/ /ri
d/ ∼ /red/
Meet belongs to a subclass where there is a change in the base from /i
/ to /e/ in speech and
ee to e in writing, while hit belongs to a subclass where the preterite and past participle
are identical with the base in both speech and writing. With read, however, we see that
it belongs with meet in speech but with hit in writing. Again, it proves easier and more
illuminating to base the classification on the spoken form. The divergent case of read
will therefore be handled primarily in the appropriate spoken class, i.e. with meet, but it
will also be given a secondary mention in the discussion of the class to which it belongs
in writing, i.e. with hit.
Before looking in turn at the various lexeme categories mentioned in §1.2, we intro-
duce a number of phonological and spelling alternations which apply independently of
particular categories.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 2.1 Phonological alternations 1573
2.1.2 The alveolar plosive suffix of the preterite and past participle:
/id/ ∼ /t/ ∼ /d/
This suffix, written ed, attaches to regular verb bases, and has the three alternants shown
in [3], with examples in [4]:
[3] i /id/ after alveolar plosives (/t/ and /d/)
ii /t/ after all other voiceless consonants (/p/, /k/, /f/, /s/, /ʃ/, /tʃ/)
iii /d/ after all other sounds
[4] i /heit·id/ hated /lænd·id/ landed
ii /lɑ
f·t/ laughed /his·t/ hissed
iii /lv·d/ loved /stei·d/ stayed
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1574 Chapter 18 Inflectional morphology and related matters
Again /i/ is present just where it prevents the juxtaposition of two similar sounds (this
time two alveolar plosives) and where it is absent there is voicing assimilation between
the consonant of the suffix and the final sound of the base. The alternation thus matches
that for the sibilant suffixes, and hence has been included in this section even though the
suffix attaches only to verbs.
2
See §3.1 for some exceptions. Compare also the past participle suffix ·en, which is pronounced as syllabic /n/
after certain consonants: see §5.3.3.
3
The /r/ that appears before the vowel-initial suffix is a special case of linking /r/. Some speakers also have an
intrusive /r/ in forms like drawing, sawing, and thawing : again, see Ch. 1, §3.1.1.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 2.2 Spelling alternations 1575
4
The verb combat may be stressed on either syllable and the suffixed forms are accordingly spelled with or
without doubling: combated/combatted; combating/combatting.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1576 Chapter 18 Inflectional morphology and related matters
In addition the plural or 3rd sg present suffix begins with a vowel when the base ends
in a sibilant (§2.1.1), and hence we have the following patterns of doubling:
[10] i quiz quizzes quizzed quizzing [verb]
ii fez fezzes [noun]
The letters h, w, y, and x are not doubled
It follows from the account of the rule given above that base-final h, w, y, and x will not
be doubled: they do not represent single consonants. Compare, then:
[11] verbs adjectives
i hurrah hurrahed hurrahing [no adjectives ending in h]
ii saw sawed sawing raw rawer rawest
iii stay stayed staying coy coyer coyest
iv box boxed boxing lax laxer laxest
On the traditional classification of all letters other than a, e, i, o, u as consonants, examples
like these have to be treated as exceptions. But they are not exceptions on the account given
here. The letters h, w, and y are never consonant symbols when they occur at the end of a
base with final stress: they are always parts of composite symbols (ah /ɑ
/, aw /ɔ
/, ay /ei/,
oy /ɔi/ in the examples of [11]). As for x, in base-final position this too is not a consonant
symbol; it is a single-letter symbol representing the two-consonant sequence /ks/, and hence
it does not fall within the scope of the rule either.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 2.2.2 E-deletion (like ∼ lik·ing, subdue ∼ subdu·ing) 1577
(a) Mute e
Application of e-deletion is illustrated in:
[14] i edge edg·ing edg·ed
ii hope hop·ing hop·ed
iii take tak·ing tak·en
iv simple simpl·er simpl·est
v square squar·ing squar·ed squar·er squar·est
As with consonant doubling, ·ed counts as beginning with a vowel even when it represents
/d/ or /t/, as in edged and hoped, and post-vocalic r counts as a consonant in non-rhotic
as well as rhotic accents, so that bases like square end in mute e and undergo the rule.5
E-deletion applies before the same suffixes as consonant doubling. The result is that
with such verb-base pairs as hope and hop or plane and plan, distinguished by the
presence or absence of mute e, the forms with vowel-initial suffixes are distinguished
instead by absence or presence of doubling: hope ∼ hoping ∼ hoped vs hop ∼ hopping ∼
hopped, or plane ∼ planing ∼ planed vs plan ∼ planning ∼ planned.6
With the suffixes other than ·ing we have assumed that the morphological division is before
the e, e.g. that hoped is analysed as hop·ed, not hope·d. On this account the alternation is in
the base, not the suffix. There are two arguments in favour of this analysis.
First, the omission of e before ·ing shows that there is unquestionably alternation in the
base: the proposed analysis is simply a generalisation of the rule of e-deletion needed for the
gerund-participle form. Note here that e also drops before non-inflectional suffixes beginning
with i: compare pure ∼ pur·ity, simple ∼ simpl·ify, etc.
Second, with ·er, ·est, and ·en the proposed division matches the pronunciation: nic·er,
nic·est, tak·en, for example, match /nais·ər /, /nais·ist/, /teik·ən/, and the same holds for ·ed
for bases ending in an alveolar plosive, as in hat·ed /heit·id/ and sid·ed /said·id/. The only
troublesome case is ·ed corresponding to /t/ or /d/, but we have already noted that this behaves
like a vowel-initial suffix with respect to consonant doubling, so the present analysis again
involves a generalisation of rules motivated elsewhere.
5
BrE has mute e in bases like centre, whereas AmE has the spelling center, with non-final e ; the e therefore drops
in the inflected forms of BrE (centr·ing ∼ centr·ed), but not in AmE (center·ing ∼ center·ed).
6
Note, however, that in the much rarer type of pair seen in bathe vs bath the distinction is lost in the inflected
forms bathing and bathed: e-deletion applies to bathe but consonant doubling cannot apply to bath because it
doesn’t end in a single consonant letter.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1578 Chapter 18 Inflectional morphology and related matters
7
The loss of e in catalogue ∼ catalogu·ing falls under the mute e case, with gue a composite consonant symbol,
while the retention of e in segue ∼ segueing is due to its being a single letter vowel symbol, representing /ei/.
8 Notice that spelling facts from lexical word-formation reinforce this: while freelance and freewheeling are
spelled as unhyphenated words, in free-enterprise system a hyphen is called for to prevent the impossible
∗freeenterprise.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 2.2.4 Alternation between ·s and ·es: plural and 3rd sg present 1579
etc.). Note that y-replacement does not apply where y is part of a composite vowel symbol.
Compare, then, the verbs in [19i], the adjectives in [ii], and the nouns in [iii]:9
[19] single vowel symbol composite vowel symbol
i a. try tries tried b. stay stays stayed
ii a. silly sillier silliest b. coy coyer coyest
iii a. city cities b. guy guys
Again it must be emphasised that the rule cannot be stated simply in terms of letters: we
need to consider what sounds they represent. Both guy and soliloquy, for example, end
in uy, but whereas y-replacement does not apply to the former because uy is a composite
vowel symbol (representing /ai/) it does apply to soliloquy since u is here a consonant
symbol (representing /w/) and y a simple vowel symbol: the plural form is therefore
soliloquie·s.
A handful of verbs have final ie rather than y in the base: die, lie, tie, vie. The other
forms, however, are the same as for try:
[20] i try trying tried tries
ii die dying died dies
For die the ie is the default spelling, so that the replacement works in the opposite
direction: ie is replaced by y before the ·ing suffix.
2.2.4 Alternation between ·s and ·es in the plural and 3rd sg present tense
This alternation can be most economically described by stating the conditions under
which ·es is used, and then saying that ·s appears everywhere else: it is in this sense that
·s can be regarded as the default alternant. This suffix is very different from the default
preterite and past participle suffix ·ed. The difference is particularly clear in pairs like
sip·s and sipp·ed, where ·ed triggers doubling of the base-final consonant letter p.
9
For some exceptions among the verbs, see §5.1.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1580 Chapter 18 Inflectional morphology and related matters
The bases here end in /s/, /z/, or /ʃ/ (x represents /ks/, the second component of which is
sibilant /s/). We have seen that a single z doubles and a single s may do so, but the suffix
is still ·es: fezz·es, gas(s)·es.
Bases spelled with final ch
These take ·es when the base ends in sibilant /tʃ/ but ·s in the less common case where
ch represents non-sibilant /k/:10
[22] i bench·es branch·es catch·es coach·es lunch·es [(t)ch = /tʃ/]
ii epoch·s eunuch·s monarch·s stomach·s triptych·s [ch = /k/]
Bases ending in mute e
Bases ending in the sibilant /d/ have mute e in the spelling: edge, judge, age, change.11
The same applies to the relatively small number ending in // (mirage, barrage, rouge),
and to some of those in /s/ (dose, niece), or /z/ (gaze, nose), and a small number in /ʃ/
(douche, niche). This e drops before a suffix beginning with a vowel by the e-deletion
rule (§2.2.2), giving edg·es, mirag·es, dos·es, etc.
An alternative analysis is edge·s, which does not involve loss of the base-final e. We adopt the
analysis edg·es, however, since this both matches the pronunciation (/ed·iz/) and allows a
more general statement of the alternation, namely that ·es occurs with all bases ending in a
sibilant. Note, moreover, that these bases do lose the e when they are followed by the ·ing
suffix: edg·ing, chang·ing, gaz·ing, etc. (see §5.1 for a few exceptions).
3 Grade
The inflectional system of grade applies primarily to adjectives, but also to a few ad-
verbs that do not end in the ·ly suffix and a handful of determinatives and prepositions
(see Ch. 6, §2.2). We look first at the inflectional forms, and then at the distinction
between inflectional comparatives and superlatives (e.g. taller, tallest) and analytic ones
(more distinct, most distinct).
10
Bases ending in nch can be pronounced with /nʃ/ instead of /ntʃ/ (except that in nudibranch and elasmo-
branch – types of mollusc and fish – it represents /ŋ k/, so these take ·s). The base loch may be pronounced
with a velar fricative rather than /k/, but in either case the base does not end in a sibilant and hence
takes ·s.
11
Foreign words like hadj and raj are exceptions, but the plurals of these words rarely occur and have somewhat
questionable status.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 3.1 Inflectional comparative and superlative forms 1581
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1582 Chapter 18 Inflectional morphology and related matters
Irregular inflection
The following have irregular forms:
[4] i good, well better best
ii bad, badly worse worst
iii much, many more most
iv little less least
v far farther/further farthest/furthest
As indicated in [i–iii], the distinction between good and well, bad and badly, much and
many is lost in the comparative forms. Well and badly, moreover, can be adjectives or
adverbs. Compare, for example:
[5] i a. This one is good. b. That one is better.
ii a. I’m feeling well. b. I’m feeling better. [adjective]
iii a. They played well. b. They played better than ever. [adverb]
Better in [iib], moreover, is ambiguous between the ordinary comparative sense “better
than before”, and the sense “recovered, well again”.
In addition, old has the regular forms older and oldest, but also irregular elder and
eldest, as used in:
[6] i my elder brother her eldest daughter
ii the elder (of the two) the eldest (of them)
These forms are highly restricted both semantically and syntactically. Semantically,
they indicate relative order of birth within a family – contrast ∗the elder of the two
editions. Syntactically, they modify a following noun, as in [i], or appear in fused
modifier-head function, as in [ii]. Elder can’t be used predicatively (∗Which one is
elder?) or with than (∗an elder brother than Max). The regular forms can be used as
variants of the irregular ones in [6]. Elder is also used in the idioms elder statesman/
stateswoman and is the source for the converted noun elder.
13 The asterisks here apply to the use of these expressions as ordinary comparatives/superlatives. More good is
possible in metalinguistic comparison, where inflectional comparatives are excluded: I’d say it was more good
than excellent (“more properly classified as good than as excellent”). Most good has most as an intensifier, not
a strict superlative marker: It was most good of you to invite us.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 3.2 Inflectional vs analytic comparatives and superlatives 1583
There are some generalisations that can be made, however. One is that participial
adjectives, as illustrated in [8], take only analytic forms:
[8] i amazing amusing boring frightening pleasing wearing worrying
ii amazed amused bored frightened pleased worn worried
(A marginal exception is tired, though more tired is much more usual than tireder.)
The remaining generalisations are best dealt with by separating adjectives out into
sets according to the syllabic composition of the base.
(a) Monosyllables
Adjectives with monosyllabic bases almost always have inflected forms, but there are some
that do not. First, the generalisation mentioned just above overrides monosyllabicity:
participial adjectives do not have inflected forms even when they are monosyllabic.
Second, there are also a few morphologically simple exceptions:
[9] cross, fake, ill, like, loath, prime, real, right, worth, wrong
These do not inflect – or at least, their inflected forms are in practice virtually never
encountered. This is not because these adjectives do not express gradable proper-
ties: there can certainly be degrees to which one can be cross with someone, loath
to do something, or in error; yet ∗crosser, ∗loather, and ∗wronger appear never to
occur.
Most monosyllables allow analytic forms, either as an alternative to inflection or as
the only way to express comparative or superlative degree, but the irregular inflectional
forms better, worse, further pre-empt use of ∗more good, ∗more bad, ∗more far, and the
inflectional forms are very much more usual with such common adjectives as big, large,
small, high, low, fat, thick, thin, long, tall, short, fast, slow, hot, cold, cool, old, young,
clean, great, wide.
(b) Disyllables
With disyllables the analytic forms are always possible, while the inflectional ones are
sometimes possible and sometimes not. Many of the conditions making inflection im-
possible relate to the ending of the lexical base. With initially stressed bases, the endings
in [10i] (only the first two of which have the status of suffixes) generally permit inflection,
while those in [10ii] reliably exclude it:
[10] i ·y angry, dirty, early, easy, funny, happy, hungry, noisy, pretty, silly
·ly beastly, costly, deadly, friendly, ghastly, ghostly, likely, lovely, manly
le able, ample, feeble, gentle, humble, little, noble, purple, simple, subtle
ow hollow, mellow, narrow, sallow, shallow, yellow
ii ·ful bashful, careful, cheerful, faithful, graceful, harmful, skilful, useful
·ish boorish, boyish, brutish, fiendish, foolish, priggish, sheepish, ticklish
·al focal, global, legal, lethal, local, moral, primal, rural, venal, vital, vocal
·ic caustic, chronic, comic, cyclic, epic, magic, manic, public, septic, tragic
·ous anxious, bumptious, callous, cautious, conscious, famous, jealous, porous
There can be no doubt, however, that the matter is very much lexically determined,
and certainly not a matter of phonology; note, for example, the following contrasts
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1584 Chapter 18 Inflectional morphology and related matters
Adverbs
Most gradable adverbs take analytic forms: softly ∼ more softly ∼ most softly. Only a
handful of adverbs have regular comparative and superlative inflection:
[12] early fast hard late long often soon
The majority of gradable adverbs have bases formed with ·ly, and the inflectional endings
are never attached to bases of this kind: early of course is not an exception, since it is not
formed from an adjective base ∗ear.
All the adverbs in [12] except often and soon are homonymous with adjectives.
Often is a somewhat marginal member of this class: the analytic forms are much more
frequent and for some speakers are the only possibility. With the others, however, only the
inflectional forms are normally possible: The meeting lasted longer / ∗more long than usual.
Earlier and later have a wider range of meaning than the corresponding plain forms. In
such examples as Earlier he had adopted a rather aggressive position and I later realised he
had been joking, where a than complement could not be added, the meanings are approx-
imately “previously” and “subsequently”; the plain forms have no corresponding use.
Adverb pairs of the type loud ∼ loudly
There are a number of pairs of adverbs where one is formed from the adjective by
·ly suffixation and the other by conversion: loud ∼ loudly, easy ∼ easily, slow ∼ slowly,
quick ∼ quickly. The ones with simple bases have regular inflection (loud, louder, loudest),
while those with the ·ly suffix take analytic more/most (loudly, more loudly, most loudly).
The former are of more limited distribution than the latter, and are commonly subject to
14
Historically, there has been a trend to move increasingly towards the analytic, though with fluctuations in the
treatment of disyllabic adjectives during the twentieth century. Early Modern English used the inflectional
type more freely (apter, privatest) and sometimes allowed both types to be combined (the most unkindest cut
of all ) – a doubling damned out of existence by prescriptivists – and even in the nineteenth century there were
occasional examples like properer, playfullest, scornfullest, sociablest.
15
Lewis Carroll’s ∗curiouser and curiouser, involving a trisyllabic base in ·ous, is ungrammatical, and was intended
jocularly, or as indicating that his young heroine Alice had not quite grasped the limitations of the inflectional
system yet.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 4 Nouns 1585
prescriptive criticism. The inflected forms, however, tend to be somewhat less restricted
and more acceptable than the plain form without ·ly :
[13] i a. They complained loudly/∗loud about the service
b. They complained louder than anyone else
ii a. He was walking quite slowly/?slow because of his injury
b. He was walking slower than usual
4 Nouns
Prototypical nouns inflect for number and case. The singular non-genitive is identical
with the lexical base, and genitive marking is added after the plural marking: we will
therefore look first at plural formation and then at genitives.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1586 Chapter 18 Inflectional morphology and related matters
16
For do in the colloquial sense of “social event”, which is converted from the verb and pronounced /du
/, the
plural is sometimes spelled do’s, with the apostrophe separating the suffix from the base, so that it is not
misconstrued as affecting the pronunciation of o (cf. also [8]).
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 4.1.2 Irregular ·s plurals 1587
This practice is less common than it used to be; with dates and abbreviations ending
with an upper case letter, the form without the apostrophe is now more usual: in the
1960s, two candidates with Ph.D.s.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1588 Chapter 18 Inflectional morphology and related matters
17
Including spouse, since here there is alternation between /s/ and /z/ in both singular and plural.
18
Swine belongs in [ii] when used as a term of abuse, usually applied to humans.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 4.1.4 The vowel change plurals 1589
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1590 Chapter 18 Inflectional morphology and related matters
[ia–iiia] /u
/ or /υ/ alternating with /i
/, and in [iva–va] /aυ/ alternating with /ai/ (where
it is just the second component of the diphthong that alternates). For the [b] items man
and woman, the changes in the written form are alike (a to e), but they are quite different
in the spoken form (/men/ and /wimin/). In complex bases in ·man the vowel difference
between /æ/ and /e/ is normally lost, with both reduced to /ə/: the result is that these
therefore have base plurals in the spoken language, as noted above.19
19
Pluralisation by vowel change is now effectively dead, and new uses and adaptations of the above words are
beginning to show regular plurals. Thus when louse is used to mean “despicable person” it has the plural
louses, and mouses is becoming increasingly common as the plural of mouse in the sense of a computer
cursor-movement peripheral.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 4.1.6 Foreign plurals 1591
Latin plurals
From Latin there are four common patterns:
[20] latin plural regular plural
i formula formulae formulas
ii radius radii radiuses
iii curriculum curricula curriculums
iv index indices indexes
The Latin plural involves changing the ending of the base, while the regular plural
adds the plural suffix to the base. For patterns [i–iii] there are some words which allow
only the Latin plural, some that allow either, and others that allow only the regular
plural. Thus for [i] larva takes only the Latin plural, replacing a by ae (larvae/∗larvas);
formula takes both (formulae/formulas); and arena takes only the regular English plural
(arenas/∗arenae). For pattern [iv] we have only the second and third of these possibilities.
(a) Bases ending in a
Bases of the three types just distinguished are shown in [21], those in [i] taking only
the Latin plural, those in [ii] allowing either, and those in [iii] taking only the regular
English plural:
[21] i ae only alga, alumna, larva
ii ae or s amoeba, antenna, fibula, formula, lacuna, nebula, persona,
retina, tibia, vertebra
iii s only algebra, area, arena, dilemma, encyclopedia, guerrilla, phobia,
quota, replica, rumba
The nouns in [iii] have a variety of origins – Arabic (algebra), Greek (phobia), Spanish
(rumba): only area and arena belong etymologically with those in [i–ii]. The normal
pronunciation of ae is /i
/, but /ai/ is found as a variant in algae, formulae, lacunae, and
/ei/ in vertebrae. Singular alga is uncommon and algae is often reanalysed as non-count
singular.
(b) Bases ending in us
The Latin plural of bases ending in us replaces this ending by i ; the default pronun-
ciation of this is /ai/, but the nouns marked † in [22] have a variant pronunciation
with /i
/.
[22] i i only alumnus, bacillus†, homunculus, locus, rectus, stimulus†
ii i or es abacus, cactus†, focus†, fungus†, hippopotamus, narcissus,
nucleus†, radius, stylus, syllabus, terminus, thesaurus, uterus
iii es only apparatus, census, excursus, foetus, hiatus, impetus, prospectus,
status, virus
Foci with the /ai/ suffix usually has /s/ rather than /k/ in the base. None of the nouns
in [iii], apart from virus, belongs etymologically with those of [i–ii]. Similarly, cor-
pus, genus, and opus do not belong in this group etymologically; they have either the
Latin plurals corpora, genera, opera, or the regular ones in es. Octopus does not be-
long here etymologically either (it derives, indirectly, from Greek); it behaves like the
nouns in [ii], though octopuses is more common than octopi (which is often criticised
by prescriptivists).
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1592 Chapter 18 Inflectional morphology and related matters
Greek plurals
There are two types to be considered.
[25] i basis bases / beisis/ / beisi
z/
ii phenomenon phenomena /fənɒminən/ /fə nɒminə/
For the spoken versions of type [ii] the plural simply drops the final /n/ of the base.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 4.1.6 Foreign plurals 1593
Bases ending in is
Most nouns ending in is are from Greek and follow the pattern of basis, with es replacing
is ; there are a few with a regular plural, es being added instead of replacing is, but there
are none with alternation between the two types:
[26] i change is to es: analysis, antithesis, arsis, axis, crisis, diagnosis, ellipsis,
emphasis, genesis, hypothesis, metamorphosis,
neurosis, oasis, paralysis, parenthesis, psychosis,
synopsis, synthesis, testis, thesis, thrombosis
ii add es after is : iris, metropolis, pelvis, penis
The last two of the nouns in [ii] derive from Latin rather than Greek (as indeed does
testis in [i]). Note that while the plural of basis, bases, is the same in writing as the regular
plural of base, the two plurals are pronounced quite differently, /beisi
sz/ vs /beisiz/;
the same applies to axes (from axis or axe) and ellipses (from ellipsis or ellipse).
Bases ending in on
The foreign plural replaces on by a. Again there are three classes:
[27] i a only criterion, phenomenon, prolegomenon
ii a or s automaton, ganglion
iii s only electron, neutron, positron, prion, proton, skeleton
Horizon, pentagon, polygon, etc., do not belong to the same etymological class as
the above, but are like [iii] in having only regular plurals. With criterion, examples
of the regular plural criterions are in fact attested, but they are very rare; much more
common is the reanalysis of criteria as a count singular (?No criteria exists), but it is
not widely regarded as acceptable. Nor is the (less common) use of phenomena as a
singular.
French plurals
French words ending in s have base plurals in writing, whereas in speech the singular has
no final consonant and the plural a regular /z/ ending. There are others, ending in eau
or ieu, that are again regular in speech but in writing have a French plural in x as well as
a regular one in s :
[28] i /ʃæsi/ /ʃæsiz/ chassis chassis
ii /plætoυ/ /plætoυz/ plateau plateaux/plateaus
Other lexemes following these patterns are:
[29] i Like chassis: chamois, corps, faux pas, patois, rendezvous
ii Like plateau: adieu, bureau, chateau, milieu, tableau
Other foreign plurals
Two further patterns, from Hebrew and Italian, are the following:
[30] i kibbutz kibbutzim [Hebrew]
ii paparazzo paparazzi [Italian]
The Hebrew plural in im is found in religious language with cherub and seraph, and in
borrowings via Yiddish like goy. It coexists, however, with regular English forms: plurals
like cherubs, goys, and kibbutzes will be found and are quite acceptable.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1594 Chapter 18 Inflectional morphology and related matters
The contrast between Italian singular o and i plural seen in [30ii] is extremely
marginal in English. Words such as paparazzi and graffiti were borrowed into English as
plurals; the singulars followed later and are not well established. Thus one of the paparazzi
is more usual than a paparazzo. Pasta terms like cannelloni, capellini, macaroni, ravioli,
spaghetti, tagliatelli, and tortellini, and some similar words such as confetti, are likewise
plurals in Italian. In English, however, they are non-count singulars; and paparazzi and
graffiti already show signs of following them in this.20
In contrast, a number of Italian borrowings in the sphere of classical music are
known primarily through their singulars: concerto, contralto, libretto, soprano, tempo,
virtuoso, etc. For these, however, the regular plurals are much more common than the
foreign ones in i, which are generally restricted to very specialised contexts such as concert
programme notes and likely to be perceived as affected elsewhere. There are thus almost
no signs of the o ∼ i pattern being active in English.
Dictionaries sometimes contain various other plurals from certain other languages
(e.g. erg ∼ arag “sand dunes” from Arabic; as ∼ aesir “gods” from Norwegian), but
none of them are in common use.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 4.1.8 Proper nouns 1595
The ’s genitive
This is the default alternant: it occurs except in the special circumstances described
below where the non-genitive ends in /s/ or /z/. In speech the ·’s suffix has the same three
alternants as we have in regular plurals and 3rd person sg present tense verbs – /iz/ after
sibilants, /s/ after voiceless non-sibilants, and /z/ after voiced non-sibilants:
[33] /hɔ
r s·iz/ horse’s /kæt·s/ cat’s /dɒg·z/ dog’s
In writing it is invariantly ’s. The apostrophe separates the suffix from the base, which
does not undergo any of the regular or irregular modifications that apply in plural
formation:
[34] i wife’s lady’s potato’s quiz’s [genitive singulars]
ii wives ladies potatoes quizzes [non-genitive plurals]
The bare genitive
In writing this has the form of an apostrophe at the end of the word: dogs’. In speech
it has no realisation at all, such genitives being identical with the non-genitive: /dɒgz/.
Notice then that, as spoken, /dɒgz/ is ambiguous between genitive singular dog’s, non-
genitive plural dogs, and genitive plural dogs’.
The bare genitive is normally restricted to nouns ending in s. It is either obligatory or
else optional, with the ’s genitive as a variant:
[35] i cats’ dogs’ horses’ wives’ indices’ theses’ species’ [obligatory]
ii Socrates’ Xerxes’ Moses’ Jesus’ Burns’ Jones’ James’
iii Socrates’s Xerxes’s Moses’s Jesus’s Burns’s Jones’s James’s [optional]
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1596 Chapter 18 Inflectional morphology and related matters
The bare genitive is obligatory with plural nouns ending in s, regular or irregular,
including foreign plurals like indices. Nouns like species which have identical singular
and plural forms with final s take a bare genitive in the singular as well as the plural, and
in writing this will apply to nouns like chassis too (§4.1.6). The bare genitive is likewise
the only possibility in more or less fixed phrases with sake : for goodness’/convenience’
sake (the latter having spoken /s/ but not written s).
An optional bare genitive is found in certain types of proper names, where it is more
likely in writing than in speech, in formal style than in informal. There is a good deal
of variation here and it is not possible to give hard and fast rules. The bare genitive
is most widely used with classical, religious, and literary names like the first five in
[35ii]. Elsewhere it is normally restricted to names pronounced with voiced /z/ rather
than voiceless /s/ (as in all the names in [35ii] except Jesus). Examples like Ross’ are
sometimes attested but they are of questionable acceptability: Ross’s is the normal form,
and in speech the /iz/ is required. Even with final /z/, a bare genitive is hardly possible if
/z/ is preceded by a vowel, as in Les or Ros.
5 Verbs
All non-defective verbs other than be have syncretism between the plain form and the
plain present tense, both identical with the lexical base. Many verbs, including all regular
ones, also have syncretism between the past participle and the preterite. And a few have
syncretism between these last two and the plain form. Lexical verbs thus have five, four,
or three overtly distinct forms:
[1] plain plain past 3rd sg gerund-
form present preterite participle present participle
take took taken takes taking
love loved loves loving
cut cuts cutting
The gerund-participle is regular in speech for all verbs, and the 3rd sg present for all
but four: our major focus will therefore be on the preterite and the past participle.
We begin in §5.1 with the forms of regular verbs, and then deal briefly in §5.2
with the irregular present tense forms, before turning to the preterite and past par-
ticiple forms in §5.3. In these first three sections we consider only verbs with simple
bases, but the analysis extends very straightforwardly to verbs with complex bases,
as shown in §5.4. Finally, in §5.5 we examine the formation of negative auxiliary
verbs.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 5.1 Regular forms 1597
The gerund-participle
This is regular in speech for all verbs, even be ; it is formed by means of the suffix ·ing,
phonologically /iŋ/:21
[2] /laik·iŋ/ lik·ing /plei·iŋ/ play·ing /bi
·iŋ/ be·ing
The 3rd sg present tense
The suffix here is identical with that of regular plural nouns and has been discussed in
§2.1. It is written as s or es and in speech has the alternants /iz/, /s/, and /z/, depending
on the phonological features of the base:
[3] /wiʃ·iz/ /sip·s/ /rɒb·z/ wish·es sip·s rob·s
The preterite and past participle
Regular forms of these are always the same, and have the suffix written as ed and pro-
nounced /id/, /t/, or /d/, depending again on the phonological features of the base:
[4] /heit·id/ /laik·t/ /plei·d/ hat·ed lik·ed play·ed
Spelling alternations
Four sets of spelling alternations were discussed in §2.2, but there is a little more to be
said about each of them.
Doubling of base-final consonant letter before suffixes beginning with a vowel
The general rules given in §2.2.1 account for such forms as those in [5], where doubling
applies to the final consonant letter of a base preceded by a vowel symbol consisting of
a single letter and stressed on the final syllable.
[5] i bat batted batting
ii occur occurred occurring
iii gas gassed gassing gasses
Doubling extends to some bases meeting the above conditions except that they have
non-final stress:
[6] i level levelled/leveled levelling/leveling
ii focus focussed/focused focussing/focusing focusses/focuses
iii worship worshipped/worshiped worshipping/worshiping
iv handicap handicapped handicapping
We noted in §2.2.1 that bases of this kind ending in l, as in [i], take doubling in BrE but
not AmE. This applies quite systematically, except that BrE doubling is optional with
parallel (no doubt because the base already contains one instance of double ll ). The
base in [ii] ends in s : bias is the one other verb of this kind.
Worship and handicap are representative of a number of other verbs taking optional
or obligatory doubling respectively; further examples are as follows:
[7] i Optional: bayonet, benefit, diagram, kidnap, program
ii Obligatory: format, hobnob, humbug, leapfrog, sandbag, waterlog
21
In non-standard dialects in both the BrE and AmE families, and also in some now largely extinct upper-class
dialects in Britain, the ·ing suffix is pronounced /in/ in the gerund-participle use (but much less so where it is
part of the lexical base, as in belongings, planking, railings, etc.).
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1598 Chapter 18 Inflectional morphology and related matters
Where both forms are used, the one with doubling tends to be characteristic of BrE,
the other of AmE.22 For the most part these verbs have, or look as though they have,
complex bases – and the doubling can then be seen as following the pattern of complex
bases with a verb as final element, such as wiretap. In the latter, the inflection matches
that of the simple verb, irrespective of the stress, so that we have obligatory doubling,
as with tap: wiretap ∼ wiretapped ∼ wiretapping (see §5.4). Note the contrast with such
items as the following (likewise with non-final stress) where doubling does not occur:
develop, dollop, gallop, gossip, hiccup, scallop.
Comparable to consonant doubling is the (obligatory) addition of k after final c,
as in:
[8] picnic picnicked picnicking
Other verbs of this kind include: bivouac, frolic, magic (away/up), panic, tarmac,
traffic.23
22
In BrE the base program is used only in the sense relating to computers; for other senses the base is programme,
which itself contains double mm.
23
Arc, where the c follows r, optionally takes k: arced/arcked, arcing/arcking.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 5.2 Irregular present tense forms 1599
Taxying is clearly formed by analogy with lying, while the use of the apostrophe with
ski’d, to establish that the base is ski, may be motivated by a recognition that skied might
be taken to be the preterite or past participle of the verb sky.24
Where y forms part of a composite vowel symbol, it normally remains unchanged, but
there are three verbs in ay which have aid in the preterite and past participle – compare:
[13] i regular pray prayed praying Also: other regular verbs
ii irregular pay paid paying Also: lay, say
Say does not strictly belong here, for it is an irregular verb in speech by virtue of a vowel
change not matched in the spelling (/sei/ ∼ /sed/): see [54].
Modal auxiliaries
The modals are highly irregular in that they show no person–number distinctions at
all, and may be treated as having neither the plain present nor the 3rd sg present, but
24
An apostrophe is also optionally found in mascara’d, where the motivation seems to be merely to avoid the
unusual sequence aed.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1600 Chapter 18 Inflectional morphology and related matters
just a single undifferentiated present tense form. Must and ought derive historically from
preterite forms.
5.3.1 Class 1 verbs: secondary ·ed formation (burn, keep, hit, lose)
A number of common verbs are clearly somewhat irregular, but can nevertheless be
treated in terms of the ·ed formation, supplemented by four other operations: devoic-
ing of the suffix, vowel shortening, consonant reduction, and devoicing in the base.
Seven subclasses of these verbs may be recognised, each involving one or more of these
operations. In all cases the preterites and past participles are the same.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 5.3.1 Class 1 verbs 1601
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1602 Chapter 18 Inflectional morphology and related matters
past participle. This can be accounted for by the combination of consonant reduction
(∗ /bendd/ and ∗ /bildd/ giving ∗ /bend/ and ∗ /bild/) and devoicing of the suffix (giving
/bent/ and /bilt/). The archaic gird (with girt /g
r t/ as its preterite and past participle)
is similar but would need to be explained in terms of another sonorant, /r/; this would
be correct for rhotic accents but, as we have noted, the /r/ has been lost in modern BrE
and is reflected by vowel quality alone.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 5.3.2 Class 2 verbs 1603
occur, but they are more often found as participial adjectives than as verbs (and see 3f
for cloven).
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1604 Chapter 18 Inflectional morphology and related matters
Class 2e: vowel change in preterite, past participle identical with base
Just two verbs show this pattern:
[37] come /km/ /keim/ /km/ come came come
run /rn/ /ræn/ /rn/ run ran run
5.3.3 Class 3 verbs: past participles formed with the ·en suffix (see, ride, take)
There are verbs in which the past participle has the suffix we refer to as ·en. This suffix is
never used for the preterite in standard English, though there are some forms, notably
done and seen, that are used as preterites as well as past participles in certain non-standard
dialects.
The suffix is most often added to the base, as in eat·en, sometimes with modification
of the base, as in swoll·en, but there are also verbs where it is added to the preterite form,
as in brok·en.
In speech it is pronounced /ən/, /n / (syllabic /n/), or /n/, normally under the following
conditions:
[38] i /ən/ after /l/ or /k/ /fɔ
lən/ /teikən/ fallen taken
ii /n / after /t/, /d/, /v/, /z/ /bi
tn / /tʃoυzn / beaten chosen
iii /n/ after vowels /si
n/ /groυn/ seen grown
There is variation between /ən/ and /n / after consonants other than /l/, so that /n / can
occur instead of /ən/ in taken, etc., and vice versa in beaten, chosen, etc.
The spelling alternations for bases taking ·en are the same as with other vowel-initial
suffixes with respect to deletion of final e, as in tak·en (formed from take), and doubling
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 5.3.3 Class 3 verbs 1605
of final consonant, as in trodd·en (formed from trod). We also have consonant doubling
in such forms as ridden and written: the bases ride and write end in an e that forms part
of a discontinuous symbol representing the diphthong /ai/, but the change to the short
vowel /i/ leads to the dropping of the e, so that the d and t are now in final position and
hence subject to doubling.
In addition ·en is reduced to ·n after a composite vowel symbol, and i replaces y at
the end of such a symbol; with bases ending in re, both the e of the base and that of the
suffix are dropped, with /r/ therefore remaining post-vocalic and pronounced only in
rhotic accents:
[39] i after composite vowel symbol see seen sew sewn
ii after composite vowel in y slay slain lay lain
iii after re tore torn wore worn
The e of ·en is retained after the single-letter vowel symbol of be to give been. With
the three verbs do, go, bear the suffix has the exceptional spelling ne : done, gone,
borne.
Some of the verbs have similar vowel changes to those that have already been consid-
ered in §§5.3.1–2. We distinguish eight subclasses.
25
Shaven is used only as a participial adjective and hence is not included in our list.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1606 Chapter 18 Inflectional morphology and related matters
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 5.3.4 Class 4 verbs 1607
AmE get belongs here, with the forms /get/ ∼ /gɒt/ ∼ /gɒtn /, get ∼ got ∼ gotten (the
same vowel pattern as tread), except that there is a variant of the past participle that is
identical to the preterite, as it always is in BrE. Lie1 has the meaning of position: lie2
“tell a lie” is regular. The preterite of intransitive lie1 is homophonous with the base of
transitive lay (listed in [13ii]): compare It lay on the floor (lie1 ) and He asked me to lay
it on the floor (lay). Lay, however, is often confused with lie1 and used intransitively,
but this usage is regarded as non-standard: ! He asked me to lay on the floor. In BrE, the
past participle of bear is spelled born only in the passive, in the childbirth sense (He was
born in 1900) or metaphorical extensions thereof (His anti-social behaviour was born of
frustration at being constantly ignored); elsewhere it is borne (It has borne fruit). Cleave
has cleft as a variant for the preterite and past participle (Class 1g).
5.3.4 Class 4 verbs: other formations ( flee, hear, stand, buy, can)
There are some verbs that have features that are not covered in the previous discussion.
Most of them appear to have an ·ed type suffix, but with vowel changes that are not
typical of the secondary ·ed formation.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1608 Chapter 18 Inflectional morphology and related matters
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 5.4 Verbs with complex bases 1609
Irregular verbs
The same applies here: the complex verb has irregular forms matching those of the simple
verb in final position. For example, arise has the forms arise ∼ arose ∼ arisen, matching
those for rise, the preterite and past participle of retell is retold, and so on. This pattern
26
Compare the two verbs relay. In the sense “lay again” it is a derivative of lay and shares its irregular preterite
and past participle: relaid. In the sense “send by relay” it has no morphological connection with lay, and is
regular: relayed.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1610 Chapter 18 Inflectional morphology and related matters
applies even where the meaning of the complex verb is relatively opaque. For example,
mistake is like take and understand like stand, in spite of the fact that there is no evident
semantic relation between the simple and complex verbs:
[60] i take took taken mistake mistook mistaken
ii stand stood stood understand understood understood
The inflectional-morphological relationship is thus maintained long after the semantic
connection has been lost. A sample of complex verbs of this kind is given in:
[61] become befall behold beset betake forbear
forbid forecast forgive forgo forswear hamstring
overbear overcome oversee overtake partake undergo
undertake uphold upset withdraw withhold withstand
Forbid derives from bid2 “order” of Class 3e: forbid ∼ forbade ∼ forbidden. Forget has
the forms forget ∼ forgot ∼ forgotten, following the model of get in AmE, except that
the past participle has only the form with ·en. Beget normally follows the same model
(beget ∼ begot ∼ begotten), but it also has a distinct archaic preterite begat.
Exceptions
There are a very small number of exceptions, where a complex verb has regular forms even
though the simple one is irregular. One example is gainsay, which differs from irregular
say in that it retains the vowel /ei/ of the base in the 3rd sg present tense (/geinseiz/
vs /sez/) and optionally does so in the preterite and past participle too (/geinsed/ or
/geinseid/ vs /sed/ only); in writing gainsay follows the pattern of say (with irregular
change of y to i).
A second exception is broadcast, which in addition to the irregular preterite and past
participle broadcast, matching that of cast, has the regular form broadcasted; likewise the
more recent telecast. Similar is deepfreeze, but here the irregular deepfroze and deepfrozen
are the normal forms and the regular form deepfreezed is of only marginal acceptability.
Regular forms
Most of the negative auxiliary forms are regular, with the suffix simply added to the
preterite or present tense form. This applies with the following:
[62] aren’t /ɑ
r nt/ couldn’t /kυdnt/ daren’t /deər nt/ didn’t /didn t/
doesn’t /dzn t/ hadn’t /hædn t/ hasn’t /hæzn t/ haven’t /hævn t/
isn’t /izn t/ mightn’t /maitn t/ needn’t /ni
dn t/ oughtn’t /ɔ
tn t/
r
shouldn’t /ʃυdn t/ wasn’t /wɒzn t/ weren’t /w
nt/ wouldn’t /wυdn t/
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 5.5 Negative forms of auxiliaries 1611
We omit ?mayn’t (/meint/, /meiənt/); though current in the earlier part of the twentieth
century, it has now virtually disappeared from the language.
Irregular forms
Half a dozen negative auxiliary forms are irregular, either in speech alone or in both
speech and writing.
Forms that are regular in writing but irregular in speech
[63] i do don’t /du
/ /dəυnt/
ii must mustn’t /mst/ /msnt/
% %
iii used usedn’t /ju
st/ /ju
snt/
In [i] there is a change of vowel, and in [ii–iii] the final /t/ is lost before the negative
suffix. Usedn’t is the only form where the suffix is added to a preterite with the ·ed suffix,
and it has a variant irregular spelling usen’t in which the final consonant is dropped in
writing too (see Ch. 3, §2.5.9).
Forms that are irregular in both speech and writing
[64] i can can’t /kæn/ /kɑ
nt/ (BrE), /kæ
nt/ (AmE)
ii shall shan’t /ʃæl/ /ʃɑ
nt/ (BrE), /ʃæ
nt/ (AmE)
iii will won’t /wil/ /woυnt/
All the negative forms in this group lose the base-final consonant, and with will there is
also a change of vowel, in both speech and writing. In BrE, can’t and shan’t have a change
of vowel in speech.
The form cannot
Can has an additional variant form cannot, unique in that not (the etymological source
of the /nt/ suffix), complete with its vowel, is attached to the lexical base. This form is
more common in the written language than in speech, though the distinction in writing
between the single word cannot and the word sequence can not is matched in speech
by that between /kænɒt/ (one /n/ and stress on the first syllable) and /kən nɒt/ (two
/n/’s and stress on the second syllable). In cannot, as in can’t, the negative is invariably
external, having scope over the modality, whereas this is not so with the analytic negative
can not : see Ch. 3, §9.10.
Cannot is more formal in style than can’t. It is hardly possible in pre-subject position:
Can’t /? Cannot we stay a little longer?; Not only can’t /∗cannot he find the key, he’s not even
sure the papers are in the office anyway!
27
Until the nineteenth century or even later, however, ain’t was common in colloquial upper-class BrE speech.
Its effective proscription has been one of the greatest successes of prescriptivists.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1612 Chapter 18 Inflectional morphology and related matters
The negative of am
The regular negative form % amn’t is restricted to certain dialects, notably of Scotland
and Ireland. In interrogatives with subject–auxiliary inversion /ɑ
r nt/ is widely used:
Aren’t I going to be invited?; I’m right, aren’t I? The earlier spellings a’n’t and an’t
have given way to aren’t, reflecting its homophony with the negative of are. Aren’t I
is fully established in BrE or AmE, though in informal AmE ain’t I? would often be
substituted.
However, in uninverted constructions ∗ I aren’t is not admissible even in informal
style. Those who have ain’t, use that here, while others use analytic negation: I am not
or I’m not.
Inflectional morphology is concerned with the form of grammatically distinct words that
belong to a single lexeme, but we turn now to certain cases of phonological variation in
the form of a single grammatical word. Compare, for example:
[1] a. I think Pat has seen it. /əz/ b. I haven’t seen it, but Pat has. /hæz/
We are dealing here with the same word has (the 3rd sg present tense form of have),
but a natural pronunciation in the context of [a] is /əz/, whereas in [b] it is pronounced
/hæz/. These are called weak and strong forms respectively.
The weak form can be regarded as phonologically reduced relative to the strong
form, but this kind of reduction is found on a much larger scale in rapid casual
speech. Compare, for example, the three pronunciations of What do you want?
shown in:
[2] i /wɒt du
ju
wɒnt/
ii /wɒt dυ jυ wɒnt/
∼
iii /wɒdə wɒ ʔ/
Version [i] contains strong forms of do and you, and represents a somewhat unnatural
pronunciation, involving unusually careful or emphatic enunciation. Version [ii] con-
tains weak forms of do and you, and represents a natural pronunciation in ordinary
∼
connected speech. Version [iii] involves a much greater degree of reduction (with /ɒ /
indicating a nasalised vowel in place of vowel + nasal sequence, and /ʔ/ indicating a
glottal stop in place of /t/) and is restricted to casual style.
There are of course more than three possible pronunciations of the sentence, showing
varying degrees of phonological reduction. The study of pronunciation in rapid casual
style belongs to the field of phonology, and is outside the scope of this book. Weak forms
do merit some consideration, however, for two reasons. In the first place, they are used,
as we have said, in ordinary connected speech and in many cases represent the default
or stylistically neutral pronunciation. The natural pronunciation of Look at the cat, for
example, is /lυk ət ðə kæt/; the version with strong forms of at and the, namely /lυk
æt ði
kæt/, would generally be unnatural to the point of unacceptability, sounding like
a sequence of disconnected words. Secondly, the use of weak forms interacts with the
grammar in that it is subject to grammatical restrictions: it is not possible, for example,
to have a weak form of has in [1b].
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 6.1 Weak forms 1613
We will also consider in §6.2 clitic forms, which represent a somewhat greater degree
of reduction. A clitic is a form which merges phonologically with an adjacent word, as
when the first two words of Pat has seen it are pronounced /pæts/ (the written language
indicates this with the spelling Pat’s seen it).
28
Forms containing /ə/ followed by /l, m, n/ have an alternate possible pronunciation with no vowel and syllabic
/l, /m, /n; these phonologically predictable alternants are not shown in the table.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1614 Chapter 18 Inflectional morphology and related matters
29
This statement may need to be qualified to cater for constructions where the gap is preceded by a sequence
of two (or more) auxiliaries or prepositions, as in I didn’t tell her, though I realise I probably should have
or What did she make it out of ? While strong forms are required for the first element (should and out),
both strong and weak forms are found for the second (have and of ). The strong version is the more usual,
however, and it may be that the weak forms are confined to the rapid casual style that falls beyond the scope
of this book.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 6.2 Clitic versions of auxiliary verbs 1615
(a) Most restricted: clitic forms of am, are, have, and will
These attach only to a preceding subject pronoun, as in:
[6] /aim/ I’m /wiər / we’re /ðeiv/ they’ve /hi
l/ he’ll
The clitic forms of am, have, and will consist of a single consonant: /m, v, l/. In the case
of are it is not possible to give a satisfactory representation for the clitic itself, as the
host + clitic combination may not be phonologically divisible into two corresponding
parts. For example, they’re in BrE is usually homophonous with locative there. Matters
are complicated by the rhotic vs non-rhotic contrast and by the fact that the host + clitic
combination may itself have strong and weak forms. Thus you’re has a number of strong
forms, including /jɔ
r / (rhyming with sore) and /jυər / (rhyming with pure), but also a
weak form /jə/.
The distributional restrictions on these clitics can be seen by comparing the examples
in [6] with those in [7], where cliticisation is not normally possible:
[7] i Jo and you are in for a shock.
ii Both of you have been pretty inconsiderate.
iii The Smiths will be there, and so will I.
In [i–ii] you is not itself the subject, only part of the subject; and in [iii] so is a connective
adjunct. We are concerned here with the phonology: so will, for example, can’t be reduced
to /soυl/ (homophonous with soul ). In writing, the use of contraction marked by an
apostrophe is somewhat more extensive, at least for ’ll and ’ve. In Pat’ll do it, for example,
’ll corresponds to the weak form /əl/, and in You could’ve been hurt the ’ve corresponds
to weak /əv/.
The contrast between [6] and [7] is very clear. The difference between clitics and
weak forms can, however, be quite slight, and there would seem to be some variation
among speakers concerning the use of these clitics. Some, for example, accept non-subject
interrogative words as host, as in Where have you put the keys? (% /weər v/).
30
In examples like What’s it matter? the /s/ is a clitic version of does, but this use belongs to a more casual style
of speech than we are concerned with here.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1616 Chapter 18 Inflectional morphology and related matters
%
ix What salad’s that man over there eating? [multi-word interrogative]
%
x Don’t use more force than’s absolutely necessary. [comparative than]
∗
xi She often’s right about things. [central adjunct]
∗
xii Never’s it going to be easier. [other preposed constituent]
The clitic may attach to the last word of the subject, which does not have to be an NP. In
[iii–iv] it attaches to the word preceding a subject gap: the subordinator that in relative
clauses, or the last word of the matrix in content clause complements (in [iv] ‘ ’s going
to happen’ is complement of think: compare I think something absolutely terrible’s going
to happen, where the subject is present in the content clause). The elliptical construction
shown in [v] may have neither or nor instead of so: Ed isn’t going and neither’s/nor’s Sue.
With prenuclear non-subject interrogative phrases, the clear cases are those consisting
of a monosyllabic interrogative word, alone or followed by one of the emotive modifiers
the hell, on earth, ever, etc. Some speakers allow other interrogative phrases, as in [ix].
There is also variation with respect to comparative than in [x]. But [xi–xii] illustrate
constructions where cliticisation is excluded: following an adjunct in central position
(between the subject and the verb), and preposed constituents other than interrogatives
and the connectives of [v].
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 6.3 Incorporation of the infinitival marker to 1617
31 Aminority of speakers appear to allow the pronunciation /wɒnə/ in [12iib], but we would regard that not
as a case of morphological compounding but a matter of phonological reduction. The complex catenative
construction provides no evidence for a morphological explanation of the kind illustrated in [11].
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
1618 Chapter 18 Inflectional morphology and related matters
usta like it), the syncretism here reflecting the homonymy in the non-compounded I used to
like it and I didn’t use to like it. The /gɒtə/ of [9ii] is a past participle form governed by per-
fect have; the latter is often omitted, however, leading to the reanalysis of /gɒtə/ as an invariant
present tense form. The /ɔ
tə/ of [iv] is an invariant present tense form or a non-standard
plain form used in the negative with supportive do (! He didn’t oughta). Also invariant are
/gənə/ and /səpoυstə/, the former being part of an idiom headed by progressive be, the latter
a participial adjective likewise found only after be (cf. Ch. 16, §10.1.3).
6.4 Liaison
There are two instances in English phonology and morphology of transitional consonants
that are inserted between words, or parts of words, to avoid a hiatus between two vowels.
This phenomenon is widely known as liaison, a term taken over from the grammar of
French.
One case is that of linking and intrusive /r/ found exclusively in non-rhotic accents.
This is of morphological relevance in the formation of gerund-participles, as discussed in
§2.1.4. For the rest, it is a quite general phonological phenomenon, and it is not necessary
to add here to the account given in Ch. 1, §3.1.1. It is never reflected in the orthography.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
§ 6.4 Liaison 1619
it is unstressed. The [iia] words never take the liaison form: we have a hero, not ∗ an hero,
in accordance with the general rule. The initial /h/ of an unstressed syllable, however,
may optionally be dropped in connected speech, as in Did you see him?, this habitual
criminal, its historical development, and so on. Loss of the /h/ results in a word beginning
with a vowel, thus providing the context for the liaison form, again in accordance with
the rule: /ənəbitʃυəl kriminəl/.
This is unproblematic as far as speech is concerned, but in writing the status of
expressions like an habitual criminal, an heroic trek, an historical novel, an hysterical
outburst is less clear. Usage manuals generally agree that an is permissible, but not
obligatory, in such cases – which reflects the fact that /h/ is optionally, but not obligatorily,
omitted in speech. The manuals suggest, however, that the present trend is towards always
using a before words of type [15iib]: a habitual criminal, a heroic trek, a historical novel, a
hysterical outburst. A hotel is often mentioned as a special case, with the suggestion that
an hotel is now old-fashioned and to be avoided.
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019
Downloaded from https://www.cambridge.org/core. National University of Singapore (NUS), on 28 Jan 2022 at 01:59:26, subject to the Cambridge Core terms of use, available
at https://www.cambridge.org/core/terms. https://doi.org/10.1017/9781316423530.019