Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Phonological and lexical factors in English
adjective gradation
UCLA Colloquium
Maria Gouskova
New York University
October 2, 2015
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Introduction
Phonological selection: affixes appear on bases that share a
phonological property
English comparatives: big/bigger, easy/easier,
productive/*productiver
English indefinites: a hat, an umbrella
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Introduction
Are the phonological generalizations part of the rule that
people know, or are they emergent?
People know a list defined by morphology—a sublexicon
Phonological generalizations are extracted from the list
through phonotactic learning
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Introduction
Sublexical phonotactic generalizaitons can be extended to
novel items (experiment 1)
Will discuss some new findings about the bracketing paradox
posed by “unhappier” (experiment 2), which present some
challenges both for lexical and for phonological accounts of the
selectional restriction
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
A phonological generalization about size?
English comparatives and superlatives:
Most bases for -er are monosyllabic (big/bigger, biggest)
A few disyllabic (easy/easier, easiest)—tend to end in [i]
Class of trisyllabic exceptions with un-: unhappier (cf.
*redundanter, *insipider, *seductiver)
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Background
Because -er bases are usually mono- or disyllabic, phonological bracketing
cannot follow semantic structure (Pesetsky 1985, Marantz 1988 et seq.).
Marantz 1988: un- requires its stem to to be a phonological wd; surface
syntactic bracketing differs from phonological bracketing much as with
clitics.
Semantics and syntax
Phonology
comp
adj
un
Pwd
er
Ft
adj
un
happy
[ [ un [happy ]adj ]
adj
er]
Pwd
Ft
er
hap py
[un [(happy)Ft er ]]
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
But!
Assumptions:
(i) cases like “unhappier”, “unfriendlier” are well-formed (Pesetsky
lists “unpleasanter” is acceptable)
(ii) restrictions are stated in phonological terms
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
How is this phonological restriction stated?
Hypothesis 1: phonological selection/subcategorization
frames
-er attaches to stems that are at most disyllabic
Rule refers to a Pwd that is a moraic trochee (McCarthy and
Prince 1986)
COMP $-er / [( )ft ]Pwd
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Phonological selection
Requires that base be phonologized before affix is realized. This
causes some syntactic difficulties (Embick 2007)
Syntactic difficulties aside, what to do with the bracketing paradox?
How to explain lexical exceptions that meet the phonological
criterion but do not take -er, such as teal-er, beige-er
Misses other phonological generalizations about -er and -est:
avoidance of r-final adjectives (e.g., purer, sourer, bitterer),
avoidance of final [A] (rawer), avoidance of clusters (moister) (see
Hilpert 2008, Mondorf 2009)
Formally odd: keeping track of the other side of the foot is non-local
(why not se(rene)-er?); Generalized Alignment does not allow
reference to monopod prosodic words (not primitive constituents)
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Alternative: Lexical selection + emergent phonotactics
Hypothesis 2: lexical selection. “-er” attaches to a
diacritically marked list of morphemes (Bobaljik 2012,
Gouskova et al. to appear).
The rule does not refer to phonology; the phonological
generalizations are extracted separately.
COMP $-er / {big, dumb, happy, soft, . . . }
Requires that there be a mechanism for learning phonological
generalizations over the list (next section)
Should work for un- without any embellishments—if the last
morpheme determines the suffix choice, prefixes should not
matter (spoiler: they do)
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Sublexicons: the intuition
General phonotactic learning:
“Does this word sound like other words of my language?”
Morpho-phonological learning:
“Does this word sound like other words that combine with this
affix?”
“Does the result of affixation sound like other words that
contain this affix?”
“Sounds like” means “is phonotactically likely”
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Sublexicons
Learning starts with general phonotactics, segmentation into
phonological words, some basic morphological segmentation
Phonotactically conditioned alternations (æR@m/@th Am1k) can
be learned from distributions (Calamaro and Jarosz 2015)
Eventually, learner encounters inconsistencies that cannot be
reduced to general phonotactics.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Inconsistency triggers sublexicon separation
Example: English -er attaches to mono- and disyllables, but not
trisyllables. But the deverbal nominalizer -er attaches to trisyllables
(e.g., malínger-er, discóver-er, ánalyz-er)
Some argue that comparatives show the emergence of general
constraints in English (Mondorf 2009)
móre bíg: violates *Clash
beaútifuler: violates *Lapse
But why is this lifted for verbs? malíngerer has a lapse
Solution: sort the lexicon and restart phonotactic learning for
subsets. Keep the insight about constraints, but confine it to
sublexicons.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Sublexical phonotactic learning
Selectional restrictions of morphemes are learned from sets of
phonological words defined morphologically or phonologically
(Albright and Hayes 2003, Becker and Gouskova to appear,
Gouskova et al. to appear, Becker and Allen submitted)
But the rules are stated over morphemes (Halle and Marantz 1993,
Embick and Marantz 2008, inter alia)
Learning input: {[bIg], [bIgô]}, {[sævi], [sævi@ô] . . .}
" big, sick, . . .} "
Rules: COMP $-er / {savvy,
Generalizations over the list {savvy, big, sick. . .} — the
sublexicon for COMP $-er — are learned the same way
phonotactic grammars are learned (Hayes and Wilson 2008).
Should notice that initial unstressed syllables are underattested
in list; ditto for long lapses
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Sublexical phonotactic learning
Gouskova et al. (to appear): Evidence that people use
sublexical phonotactics to decide which allomorph to put on
wugs in cases of suppletion
That paper discusses ifferences between this theory and other
theories of generalization, such as the Minimal Generalization
Learner
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Phonological generalizations about -er takers are fragile:
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Phonological generalizations about -er takers are fragile:
stupider
vapider, rapider (and cf. “stupidest”)
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Phonological generalizations about -er takers are fragile:
stupider vapider, rapider (and cf. “stupidest”)
Your turn:
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Phonological generalizations about -er takers are fragile:
stupider vapider, rapider (and cf. “stupidest”)
Your turn:
funnest, funner
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Phonological generalizations about -er takers are fragile:
stupider vapider, rapider (and cf. “stupidest”)
Your turn:
funnest, funner
mauvest, mauver
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Phonological generalizations about -er takers are fragile:
stupider vapider, rapider (and cf. “stupidest”)
Your turn:
funnest, funner
mauvest, mauver
realest, realer
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Phonological generalizations about -er takers are fragile:
stupider vapider, rapider (and cf. “stupidest”)
Your turn:
funnest, funner
mauvest, mauver
realest, realer
tealest, tealer
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Phonological generalizations about -er takers are fragile:
stupider vapider, rapider (and cf. “stupidest”)
Your turn:
funnest, funner
mauvest, mauver
realest, realer
tealest, tealer
Explanation: your sublexicon is not the same as other people’s. This has
nothing to do with a rule that mentions a phonological context
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Usage changes a lot over time:
“The country is a vast deal pleasanter, is it not, Mr Bingley?”
. . . it was the only provision for well-educated young women of small
fortune, and however uncertain of giving happiness, must be their
pleasantest preservative from want. – Pride and Prejudice, Jane Austen
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
While “pleasanter” is on the decline, “stupider” appears to be on the rise:
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Some observations that this view explains
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Explanation
Explanation: the content of sublexicons changes over time
because it depends on usage.
But whether an item can join the sublexicon still depends on
its well-formedness wrt the sublexical phonotactic grammar.
So let’s try to figure out what the phonological generalizations
are
Also prove that general phonological factors cannot explain
what -er attaches to
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
An experiment
Task: nonce word rating study
Suzy Ahn and I presented ⇡300 nonce words as either
adjectives or verbs, in isolation and combined with -er.
Items represented a “phonotactic spectrum” based on English
lexical comparatives. E.g., “zice”, “ezzep”, “fudaddow” (with
the aid of the UCLA Phonotactic Learner, Hayes and Wilson
2008)
Participants: 267 Mechanical Turkers, divided appx evenly
between adjective and verb condition
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
The task: adjective condition
Please push the “play” button to listen to the adjective:
I know you think I am chike. ⌘
Rate the adjective chike:
awful [ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] good
Now listen to the comparative form:
But you are even chiker. ⌘
How likely would an English speaker be to say chiker in this sentence?
not likely [ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] pretty likely
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
The task: verb condition
Please push the “play” button to listen to the verb:
Flo recently learned how to chike. ⌘
Rate the verb chike:
awful [ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] good
Now listen to the noun:
She’s an amateur chiker. ⌘
How likely would an English speaker be to call someone who chikes a chiker?
not likely [ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] pretty likely
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
Predictions
Bare forms: Verbs are more restrictive phonotactically than
nouns/adjectives, so bare V should be rated lower than bare
Adj
-er forms:
General phonotactics: there should not be a difference between
V-er and Adj-er
Sublexical phonotactics: Adj-er should be worse than V-er
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
The items
To make the stimulus list, we searched CELEX (Baayen et al. 1993)
for the verbs that appeared with the nominalizer -er, and for
adjectives that appeared with the comparative -er
For matches, we got transcriptions from the Carnegie-Mellon
University pronunciation dictionary
Trained the UCLA Phonotactic Learner separately on each list
The learner generates nonce words in the course of figuring out
what is un- and under-attested in the list.
Sampled from this list, whittled down to non-words by checking
against the CMU dictionary
Had a phonetician pronounce the list for us, rejected items she
couldn’t handle
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
But you are even fudaddower (black)
Barbara likes to fudaddow (gray)
She is quite the fudaddower (gray)
5
3
low
mid
high
People's rating of Adj vs. Verb,
grouped by violations in adjective base sublexical grammar
Maria Gouskova
Phonological and lexical factors in English adjective gradation
1
worst
2
3
2
1
worst
Adj-er
Verb-er
4
5
Adj
Verb
best
You think I am fudaddow (black)
4
best
The results
low
mid
high
People's rating of Adj-er vs. Verb-er,
grouped by violations in adjective-er sublexical grammar
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
Some more refined predictions to test statistically
If the rule specifies the insertion context as [(´ )Ft ]Pwd, then
trochees should not be rated lower than monosyllables
If the rule only specifies the prosodic shape, then phonotactic
violations in the -er base sublexicon should not contribute anything
on top of that
If the rating is based on general well-formedness factors, then
presentation as a verb vs. adjective should not matter—and there
should not be an interaction between sublexical phonotactic
violations and how the item was presented.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
Analysis
lmer(er-rating ⇠ pros.shape + sublex.violations * adj-vs-verb + (1|frame) +
(1+adj-vs-verb|base) + (1+pros.shape+sublex.violations|participant)
Estimate
Std Error
t
p
3.621533
0.082106
44.11
<0.0000
trochee
-0.871253
0.096966
-8.99
<0.0000
iamb
-0.815502
0.090273
-9.03
<0.0000
Intercept(=mono, adj, 0 viol.)
trisyllabic
-1.074316
0.117345
-9.16
<0.0000
phonotactic violations in er-base sublexicon
-0.023351
0.005043
-4.63
<0.0000
presented as verb
0.158822
0.084607
1.88
0.0605
er-base-violations : presented as verb
0.005994
0.002871
2.09
0.0368
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
Discussion
Theory of the task: people run each nonce word through the
sublexical phonotactic grammar for bases and derived words
Score is assigned based on how well the nonce word does
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
An experiment: nominalizer vs. comparative -er
Discussion
People use sublexical phonotactic knowledge to generalize from
lexical comparatives to nonce words—not just general phonotactic
knowledge about English
More than just “is it a foot?”
But these are monomorphemic nonce words—nonce roots, really
What about morphologically complex words?
Theoretical issue: what is the content of the sublexicons? Is it
“words” or morphemes?
Empirical question: “unfriendlier” sounds better than it should if the
restriction is on disyllables; it sounds worse than it should if
“friendly” is in the sublexicon for -er and simple adjacency
determines comparative formation
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Is selection local?
In some versions of Distributed Morphology, a morpheme’s
realization is determined by string-local overt morphemes (Embick
2010). In x-y-z, y can determine shape of z, but x cannot. (though
see Myler to appear)
A simple case of local allomorph selection: English n.
In English, -ity is not as productive as -ness except after -able
and -al (Baayen and Renouf 1996)
googleable googleability ?googleableness
theatrical
theatricality ?theatricalness
Analysis (Embick and p
Marantz p
2008):
n $ -ity/
Roots ( Atroc, Curious. . .); [a, able], [a, al]
n $ -ness
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Do affixes select -er?
Suppose we combine this notion of locality with lexical selection (affixes come
with a list of morphemes)
Bare adjectives can clearly select -er: big⇠bigger, tidy⇠tidier, stupid⇠stupider,
etc.
What about adjectives formed with -ly and -y?
man-ly
manlier
craz-y
like-ly
likelier
eas-y
friend-ly
friendlier
ros-y
time-ly
timelier
fiddl-y
shape-ly
shapelier
savor-y
love-ly
lovelier
old-person-y
Adjectival -ly appears to select for monosyllables as well, and contributes many
of the disyllabic adjectives to the -er base list.
But -y is much more productive, occuring on phrase-sized constituents, and it
clearly forms at least some adjectives that can take -er
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Does un- affect the acceptability of comparatives?
Suppose the rules are:
COMP $-er / -ly, ROOTS {busy, big, savvy, sick, etc.}
a $-ly / ROOTS {friend, man, god, gain, wield. . .}
Phonotactic generalizations: Roots in COMP sublexicon are at
most disyllabic; roots in -ly sublexicon are at most
monosyllabic. The disyllabicity of most comparative adjectives
with -er is emergent
If this is how the rule is stated, prefixes shouldn’t affect
suffixation
So do they?
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Experimental design
The task and the participants
Mechanical Turkers in Canada and the US, N=90
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Experimental design
The stimuli
Adjectives were presented with and without un-, in comparative and superlative
contexts.
(un)busy
(un)earthly
(un)grabby
(un)lovely
(un)savvy
(un)timely
(un)canny
(un)easy
(un)happy
(un)manly
(un)comely
(un)costly
(un)crazy
(un)friendly
(un)healthy
(un)ruly
(un)shapely
(un)gainly
(un)holy
(un)saintly
(un)sightly
(un)godly
(un)likely
(un)savory
(un)tidy
Maria Gouskova
Phonological and lexical factors in English adjective gradation
(un)seemly
(un)wieldy
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Experimental design
The stimuli
Some occur mostly without un- (red), some mostly with (blue). Most occur
with and without un- (magenta).
The red and blue items have a frequency of 0 in the English Lexicon Project.
Others (e.g., “canny”, “seemly”) have very low frequencies in the corpus.
Included to ensure no floor or ceiling effect for ratings.
unbusy
(un)earthly
(un)grabby
unlovely
unsavvy
(un)timely
(un)canny
(un)easy
uncomely
(un)friendly
(un)happy
(un)manly
(un)seemly
wieldy
(un)healthy
ruly
unshapely
uncostly
gainly
(un)holy
unsaintly
sightly
uncrazy
godly
(un)likely
(un)savory
(un)tidy
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Experimental design
The frames: an example
Pick a {manlier/more manly} hobby than stamp collecting.
My hobby is the {manliest/most manly} of all hobbies.
Stamp collection is an even {unmanlier/more unmanly}
hobby than your last one.
Stamp collection is the {unmanliest/most unmanly} hobby
ever.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Experimental design
Controls
30 MTurkers were asked to rate the bare adjectives (“canny”,
“unbusy”, etc.), to obtain baseline ratings (no frames)
40 MTurkers were asked to rate frames with bland
comparatives and superlatives (“good/better/best”,
“hard/harder/hardest”, etc.)
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Results
Overall trends
5
worst
1st bean: ppl like adjective-er better than un-adjective-er.
2nd bean: ppl like more adj better than more un-adj.
1st, 2nd vs 3rd, 4th: people like superlatives better than comparatives
best
1
2
3
4
un-adj
adj
er
more
est
most
Ratings for comparative
and superlative adjectives with and without unMaria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Results
Analysis
Linear mixed effects model fit in R using lme4 with the most complex random
effect structure that allowed for a model to be fit
Est
SE
t
p
(Intercept)
2.163259
0.09
23.735
0.0000
log frequency in ELP
0.184009
0.008
22.472
0.0000
periphrastic (more/most adj)
0.309622
4.487
0.0000
superlative (most/-est)
0.220289
0.069
3.192
0.0014
un-prefixed
0.069
-0.392239
0.072
-5.480
0.0000
base rating (1–5) vs. freq.
0.560163
0.052
10.678
0.0000
periphrastic : un-prefixed
0.623759
0.08
7.824
0.0000
superlative : un-prefixed
0.254297
0.08
3.192
0.0014
-0.177912
0.08
-2.235
0.0254
periphrastic : superl. (most)
collinearity: max VIF = 3.2 (predictors are reasonably independent)
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Results
Analysis
People disprefer adjectives with un-;
But they prefer un- adjectives with more and most;
“more unfriendly”
“unfriendlier”; “friendlier”
Maria Gouskova
Phonological and lexical factors in English adjective gradation
”unfriendlier”
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Results
5
worst
Individual adjectives: “friendly”
best
1
2
3
4
unfriendly
friendly
er
more
est
most
Ratings for '(un)friendly'
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Results
5
worst
Individual adjectives: “likely”
best
1
2
3
4
unlikely
likely
er
more
est
most
Ratings for '(un)likely'
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Results
5
worst
Individual adjectives: “tidy”
best
1
2
3
4
untidy
tidy
er
more
est
most
Ratings for '(un)tidy'
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Results
5
worst
Individual adjectives: “happy”
best
1
2
3
4
unhappy
happy
er
more
est
most
Ratings for '(un)happy'
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Results
5
worst
Individual adjectives: “canny”
best
1
2
3
4
uncanny
canny
er
more
est
most
Ratings for '(un)canny'
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Discussion
Does -ly select -er/-est?
No difference between -ly and -y in the study
Doesn’t matter if -y is morphemic, eas-y vs. busy
Adjectives in -ly are likelier to get high ratings when presented with -er, but not
hugely so
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Discussion
Does un- make comparatives and superlatives worse?
YES.
Sometimes, the difference is small, and sometimes it is fairly dramatic
(untidier more untidy, untidiest
most untidy)
So there isn’t really a bracketing paradox. But now we need to explain why unmakes comparatives worse.
Possibility: “un-” is less compatible with the comparative meaning
Then “more un-” should also be bad, and it generally isn’t
Matushansky (2012) argues -er and more forms are semantically distinct,
and Embick (2007) shows why any adjective can combine with more in
meta comparatives.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Discussion
Back to the theories: phonological selection
Phonological selection: -er/-est attach to a maximally disyllabic foot
Theory of the task: people take the suffixed form and check whether the base is
disyllabic; if so, higher rating
Explains why “unfriendlier” is rated worse than “friendlier”—but only if the
phonological bracketing is [un-friend-ly]-er
Fails to explain the fine-grained distinctions that people made in the
fudaddow/fudaddower experiment
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Discussion
Back to the theories: Lexical selection
-er/-est attach after listed morphemes
The rule:
COMP $-er / -ly, ROOTS {busy, happy, savvy, tidy, canny. . .}
Theory of the task: when people encounter a comparative/superlative they
haven’t heard, they run it through the sublexical phonotactic grammar
Trisyllables are far less likely in these sublexicons than mono- and disyllables
Doesn’t explain why “unhappier” gets lower ratings than “happier”—but there is
another story
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Effects of complexity on ratings
Boyd (2007): “more Adj” easier to process
People prefer “more Adj” forms to “-er” forms in complex environments.
The cop was angrier than the sailor.
The cop was angrier to hear that the band was breaking up than the sailor.
The cop was more angry than the sailor.
The cop was more angry to hear that the band was breaking up than the sailor.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Effects of complexity on ratings
Boyd (2007)
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Effects of complexity on ratings
Negation+comparison
Thus, “unfriendlier” sounds better than “productiver” given its
trisyllabicity because “friendly” is in the sublexicon, and the
grammar lets it through
It sounds worse than “friendlier” or “friendliest” because both
comparatives and negation (un-) add to the processing
difficulties that lead people to rate words like “unfriendlier”
lower.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Conclusions
Most of the trisyllabic comparatives and superlatives with the un- prefix are
rated lower than disyllabic equivalents—as one might expect from their
phonotactics
This is consistent with a theory that the affixation rule checks the phonological
size of the base, but that theory has other problems
Alternatives:
sublexicon checking of phonological words
semantic incompatibility with uneffects of complexity in rating experiments
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Conclusions
The theory presented here maintains that phonological
selection is always lexical
Insertion rules refer to morphemes, not words
Phonological generalizations are extracted from sublexicons via
phonotactic learning
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Thanks
Exp 1 was done in collaboration with Suzy Ahn (NYU)
Thanks to Lisa Davidson for recording audio stimuli for that
experiment
This research is supported in part by NSF BCS-1224652
Thanks to the members of NYU’s Morphology Research
Group, audiences at Roots IV, and CUNY
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Albright, Adam, and Bruce Hayes. 2003. Rules vs. analogy in English past tenses: A
computational/experimental study. Cognition 90:119–161.
Baayen, R Harald, Richard Piepenbrock, and Rijn van H. 1993. The {CELEX} lexical data base.
{CD-ROM}.
Baayen, R Harald, and Antoinette Renouf. 1996. Chronicling the Times: Productive lexical innovations
in an English newspaper. Language 69–96.
Becker, Michael, and Blake Allen. submitted. Learning alternations from surface forms with sublexical
phonology. Phonology, URL http://ling.auf.net/lingbuzz/002503.
Becker, Michael, and Maria Gouskova. to appear. Source-oriented generalizations as grammar inference
in Russian vowel deletion. Linguistic Inquiry 47.
Bobaljik, Jonathan. 2012. Universals in comparative morphology . Cambridge, MA: MIT Press.
Boyd, Jeremy Kenyon. 2007. Comparatively speaking: A psycholinguistic study of optionality in
grammar. Doctoral dissertation, University of California, San Diego, La Jolla, CA.
Calamaro, Shira, and Gaja Jarosz. 2015. Learning general phonological rules from distributional
information: A computational model. Cognitive science 39:647–666.
Embick, David. 2007. Blocking effects and analytic/synthetic alternations. Natural Language &
Linguistic Theory 25:1–37.
Embick, David. 2010. Localism versus globalism in morphology and phonology . Cambridge, MA: MIT
Press.
Embick, David, and Alec Marantz. 2008. Architecture and blocking. Linguistic Inquiry 39:1–53.
Gouskova, Maria, Sofya Kasyanenko, and Luiza Newlin-Łukowicz. to appear. Selectional restrictions as
phonotactics over sublexicons. Lingua URL http://ling.auf.net/lingbuzz/002673.
Halle, Morris, and Alec Marantz. 1993. Distributed Morphology. In The view from building 20. Essays in
honor of Sylvain Bromberger , ed. Kenneth Hale and Samuel Jay Keyser. Cambridge, MA: MIT Press.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University
Introduction
English gradation: background
Sublexicons
The bracketing paradox
The end
References
Hayes, Bruce, and Colin Wilson. 2008. A maximum entropy model of phonotactics and phonotactic
learning. Linguistic Inquiry 39:379–440.
Hilpert, Martin. 2008. The English comparative—language structure and language use. English
Language and Linguistics 12:395–417.
Marantz, Alec. 1988. Clitics, morphological merger, and the mapping to phonological structure. In
Theoretical morphology , ed. Michael Hammond and Michael Noonan, 253–270. San Diego:
Academic Press.
Matushansky, Ora. 2012. More or better: On the derivation of synthetic comparatives and superlatives
in English. Ms., UiL OTS/Utrecht University/CNRS/Université Paris-8.
McCarthy, John J., and Alan Prince. 1986. Prosodic morphology. Technical Report #32, Rutgers
University Center for Cognitive Science, version of 1996.
Mondorf, Britta. 2009. More support for more-support: The role of processing constraints on the choice
between synthetic and analytic comparative forms, volume 4. Amsterdam: John Benjamins
Publishing.
Myler, Neil. to appear. Exceptions to the mirror principle and morphophonological "action at a
distance": The role of "word"-internal phrasal movement and spell out. In The structure of words at
the interfaces, ed. Heather Newell, Maire Noonan, Glyne Piggot, and Lisa Travis. Oxford University
Press.
Pesetsky, David. 1985. Morphology and logical form. Linguistic Inquiry 16:193–246.
Maria Gouskova
Phonological and lexical factors in English adjective gradation
New York University