CLASS I & CLASS II
AFFIXES IN
ENGLISH
 Dorothy Siegel in her 1979 PhD
dissertation  Topics in English
morphology pointed out differences in
phonological and morphological
properties of two types of affixes in
English, i.e. Class I and Class II affixes.
The affixes were distinguished also by
the type of boundary they introduced:
morpheme (stem) boundary + ( plus
symbol) in Class I, or word-boundary #
( cross-hatch symbol) in Class II af.
 Class I suffixes: +ion, +ity, +y, +al, +ic,
+ate, +ous, +ive
Class I prefixes: re+ (in reduce), con+,
de+ (in deduce), sub+, in+, en+, be+
Class II suffixes: #ness, #less, #hood,
#ful, #ly, #y, #like
Class II prefixes: re# (in rewrite), sub#,
un#, non#, de# (in derail), semi#, anti#
Differences in the phonological behaviour
of Class I and Class II affixes:
Class I suffixes may cause a stress shift
in the derivational base to which they
attach. Class II suffixes are stress-neutral:
productive  productivity (Class I +ity)
productive  productiveness(Class II #ness)
fragile  fragility (Class I +ity)
fragile  fragileness (Class II #ness)
 Some Class I prefixes may attract the
stress fromthebasewordto
themselves, e.g.:
finite / faInaIt/ - infinite /'InfIn´t/
marine  submarine
 Class I affixes may cause vocalic and
consonantal alternations in the bases
to which they attach:
Vocalic alternations in the base
/aI/~/I/ fragile  fragility
Consonantal alternations
/t/ ~ /s/ (Spirantisation)
democrat - democracy
 Class I affixes often show allomorphy.
Ex.: the adjective-forming suffix  al
exhibits also the variant  ar (when
attached to the stem ending in /l/):
cause (n.) + -al causal (adj.)
inflection (n.) + -al inflectional (adj.)
Vs: pole (n.) + -ar polar (adj.)
module (n.) + -ar modular (adj.)
NB: -al may cause a stress shift, e.g.
government  governmental.
 Allomorphy of Class I prefix con+:
contain, conceive
vs. compel, commit
Allomorphy of Class I negative prefix :
in-: inedible, inconceivable,
im-: impossible, immobile, impatient
ir-: irrational, irrelevant
il-: illegible, illegal
 Class I affixes can attach to bound
morphemes (i.e. bound roots or bound
stems). Class II affixes attach to words.
in+ept (Class I in+ and bound root *ept).
Cf. *unept (Class II un#)
leg+al (bound root *leg and Class I +al)
Tortu+ous (Class I +ous). Cf: *tortulike,
*tortuful (Class II #like, #ful)
 Class I affixes typically occur closer to
the root than Class II affixes,
[[fastidi+ous]#ness] fastidiousness
(bound stem - Class I suf. - Class II suf.)
Class II affixes are not expected to follow
Class I affixes:
*successfulity (Class II #ful  Class I
+ity) vs. successfulness (Class II #ful 
Class II #ness)
 Class I prefix is not attached  outside
of Class II suffix:
success#ful - *[in+[[success]#ful]]
(impossible derivation:
[Class I pref. [[root] Class II suf.]]
success  successful - *insuccessful)
vs. [un#[[success]#ful]] (success 
successful  unsuccessful)
[Class II pref. [[root] Class II suf.]]
 Siegel (1979) proposed the so-called
Level Ordering Hypothesis:
Class I and Class II affixation take place
in separate blocks. Class I affixation
processes occur first, preceding stress
placement rules. Class II affixation
processes apply after stress rules.
 CLASS I AFFIXATION
STRESS RULES
CLASS II AFFIXATION
 Problems for the level ordering:
occurrence of Class II affixes closer to
the root (or stem) than Class I affixes:
a/ [[[root] Class II suffix] Class I suffix]
readability [[[read]#able]+ity]
(read readable readability)
palatalisation [[[[palate]+al]#ize]+ation]
(palate palatal palatalize
palatalization)
b/ [[Class II prefix [root]] Class I suffix]
untruth [[un#[true]]+th]
(true untrue untruth)
ungrammaticality
[[un#[grammat+ical]]+ity]
grammatical ungrammatical
ungrammaticality
underestimation
[[under#[estim+ate]]+ion]
(estimate underestimate
underestimation)
 Comment by Aronoff and Sridhar:
we should treat the Level Ordering
Hypothesis as expressing a tendency,
rather than a strict principle in English.
This tendency does not take into accout
the preferences for some affixes to
occur together, e.g.:
#ize+ation (velarization)
#able+ity (readability)
#ment+al (governmental)
un# & .+ity (unpopularity)
 Aronoff and Sridhar propose to divide
English affixes into Stem affixes and
Word affixes. Stem affixes (e.g.  al,
in-, -ous, -ity, -ible) attach typically to
bound stems/roots, but can also attach
to words, e.g.  al in leg+al (attached to
a bound root) and in governmental
(attached to the word).
Word affixes (e.g.  ful, -able, -ness)
can attach only to words, and not to
bound roots, e.g. *eptness (cf. inept),
*legable (cf. legible).
 Word affixes cannot affect stress
placement because they behave like
phonological clitics attached to the
lexical word.
Ex. of phonological clitics: unstressed
function words in English (e.g. articles,
object pronouns).
Compare: an 'apple  (un('like))
'write us  (('write)-er)
 When a Word affix is followed by a
Stem affix, both affixes form a
phonological word, as in:
compartmentalization (related to
com'part  to divide into parts )
[[[[[compart]#ment]+al]#ize]+ation]
Primary, secondary and tertiary stresses:
('compart) ('mental) (i'zation)
Cf. the stress pattern in a sentence:
('combat) ('mental) (e'lation)