ArticlePDF Available

Metalinguistic, Shmetalinguistic: The Phonology of Shmreduplication



Linguistics This paper explores English shm- reduplication and aims to answer questions that have previously been left open on the subject. Results from an online survey show emerging patterns that have not been adequately addressed in the literature. It will be shown that shm- reduplication targets prosodic landmarks, syllabic landmarks and the phrasal site of the target of reduplication. Instances of avoidance phenomena will also be discussed and analyzed. The results suggest that shm- reduplication is computed by a grammar that is formed through underdetermined exposure.
Metalinguistic, shmetalinguistic: the phonology of shm-
Andrew Nevins and Bert Vaux
MIT and Harvard University
1 Introduction
English shm-reduplication, as in Oedipus shmoedipus, is well known but
incompletely understood. McCarthy and Prince (1986:71) for instance
state that “the initial consonants, if any, are replaced by a specified
consonant or cluster” [in this case Σµ-]. Sanders 2000 similarly states “a
word is fully reduplicated, and the initial onset of the reduplicant is
replaced with schm-”. Such analyses leave many subcases of shm-
reduplication unexplained; what about cases where shm- does not replace
the entire onset, as in breakfast-shmreakfast (in the film Mall Rats)? What
about obscene, obschmene, where shm- overwrites a medial onset? What
about árcade, shmárcade, where the stress of the base word changes?
Despite frequent references in the literature (e.g. Feinsilver 1961,
McCarthy and Prince 1986, Zwicky and Pullum 1987, Pullum 1991,
Alderete et al. 1996, Ginzburg and Sag 2000), English shm-reduplication
and its attendant problems have yet to receive a thorough or satisfactory
linguistic study. To address these lacunae we conducted a detailed online
survey.1 Based on the principled variation that emerged, we characterize a
restricted set of morphophonological operations with implications for both
rules and representations as they may variably target prosodic landmarks
(first vs. stressed syllable), syllabic landmarks (onset vs. first consonant)
and the phrasal size of the target of reduplicative morphemes (first word
vs. entire phrase). Second, based on subjects’ treatment of s-, m-, schm-,
and sh-initial words, and words whose output is an existing schm- initial
word (e.g. Joe schmo), we argue that avoidance phenomena show a
regularity that, contra Pullum 1991 and Sanders 2000, cannot be dismissed
as metalinguistic.
2 How did shm-reduplication develop?
Though shm-reduplication is most familiar from English, individuals who
are familiar with it generally feel it to be of Yiddish origin. Southern
(forthcoming) suggests that shm-reduplication arose in Yiddish from a mix
1 The survey can still be taken at
of Turkic Echo m- and East Slavic sh-. The Oxford English Dictionary on
the other hand sees it as an English-internal development, “derived from
the numerous Yiddish words that begin with this sequence of sounds”. The
existence of early Yiddish forms in shm- supports the former theory over
the latter (cf. Weinreich (1980:623-4), who seems to think that the
construction goes back several centuries in Yiddish.) Southern cites in his
support Yiddish shmallig, employed in a manuscript of c.1600 to disparage
hallig ‘holy’. Yitskhok Niborski (personal communication) hypothesizes
that the archetype for shm-reduplication in Yiddish is the collocation tate
shmate ‘father shmather/rag’, which he states was already in use more than
150 years ago in European Yiddish communities. It would have been used,
he adds, by an embittered wife against the man who provided her with
children but not with an income. In this case shmate is an independent
lexical item meaning ‘rag’, but it may have provided the vehicle for
reanalysis as an echo formation.
Lockwood 1978 claims that shm-reduplication migrated into from Yiddish
into American English in the late 19th century, and became integrated into
common usage by the 1930s. The Oxford English Dictionary cites its first
known use in English as crisis-shmisis (Goller 1929:V.ii.215).
Today shm-reduplication enjoys a fairly wide distribution in English, and
is not limited to Yiddish or Jewish contexts, as shown by the following:
(1) “Pedro Schmedro. The third-place Red Sox may have their big gun
going to the mound this afternoon at the SkyDome, but as far as
the exuberant Blue Jays are concerned, it’s a case of too little,
too late.” (The Toronto Star, June 25, 2000)
(2) “Breakfast?! Breakfast shmreakfast, look at the score for God’s
sake. It’s only the second period and I’m winning twelve to two.
Breakfasts come and go, Rene, but Hartford, the Whale, they only
beat Vancouver maybe once or twice in a lifetime.” (Mallrats,
3 What does shm- mean?
Shm-reduplication resembles echo formations in other languages in being
used to downplay or deride a particular phrase (cf. Emeneau 1939 for
South Asian languages). As one survey respondent put it, applying shm-
reduplication to a form indicates “I care so little about [it] that I will
pronounce it flagrantly incorrectly, so there”. The dismissive sense of the
construction can also be employed modally, to reassure, to downplay a
situation or problem that is potentially overwhelming or threatening, or to
lighten a situation with humor by pretending to dismiss it.
The elements of a shm-reduplication generally take the form of a topic-
comment binomial pair, often with incomplete list intonation. As we have
discussed with Michael Wagner, it may be that the intonation of shm-
reduplication is crucial to deriving its meaning: the intonation of
enumerating a list of items that share a common property, whose second
item is in this case, a total nonsense word, may indicate that the first item
resides in a natural class with unimportant items.
4 The survey
The semantics and pragmatics of shm-reduplication are relatively clear, but
its phonological, morphological, and syntactic behavior are less so. In
order to elucidate the nature of these aspects of the phenomenon, we
conducted an online survey consisting of 55 questions designed to elicit
information on the treatment of variables including m-initial inputs, shm-
initial inputs, non-initial shm, glide-initial inputs, second-syllable stress,
ambisyllabicity, compounding, phrase structure of different types, lexical
blocking, and complex onsets. The order of the questions and of answers
was randomized for each survey respondent. Each question was multiple
choice, in the following general format:
(3) Person A: Who did that awful painting hanging in your
Person B (pointing at her husband): Umm, Lee did.
Person A: Lee did, _____! I know that you painted it, you
untalented fop!
A. Lee shmid
B. Shmee shmid
C. Shmee did
D. Nothing sounds good here
E. Other [ ]
Comments on this question [ ]
As of this writing, 190 individuals have completed the survey; our
discussion and analysis are based on the results from these. Since some
respondents did not complete the survey and the order of questions was
randomized each time the survey was administered, each question has
fewer than 190 data points.
We then administered a second survey of 200 individuals, to test additional
variables and establish a baseline for shm-reduplication in basic cases.2 All
200 individuals answered each of the five questions in this second survey.
5 Analysis
We begin our analysis with the phonological behavior of shm-
reduplication. We situate our interpretation of the reduplicative and
overwriting aspects of the phenomenon within the theory of Anchor Points
developed by Yu 2002 and Nevins & Vaux 2003, which in brief
hypothesizes the following:
(4) Phonological Rules cannot count.
(5) Phonological Rules can only refer to a restricted set of Anchor
In this theory, processes such as infixation and reduplication can only
target one of the following prominent landmarks (cf. Pierrehumbert and
Nair 1995):
(6) Anchor Points: 1st σ, 1st foot, 1st consonant, 1st vowel, stressed σ,
final σ
Each of these targets is attested for both reduplication and infixation:
(7) Infixation after first foot: Ulwa karas:-ka-mak ‘knee’
(McCarthy and Prince 1993)
(8) Reduplication after first foot: Yidiny mula-mula-ri ‘initiated
man’ (Dixon 1977, cited in McCarthy and Prince 1999)
(9) Infixation after Initial Consonant: Sundanese n-ar-aho ‘to
know’ (Robins 1959); Atayal k-maial ‘talk (actor focus)’
(Egerod 1965)
(10) Reduplication after Initial Consonant: Mangarrayi g-ab-abuji
‘old persons’ (Downing 2001)
(11) Infixation after first vowel: Katu ka-r-chet ‘dead’ (Costello
(12) Reduplication after first vowel: Ø (but Pima may provide an
(13) Infixation Before Stressed σ
σ: English e-goddamned-váporate
(McCarthy 1982)
2 This survey can be seen at
(14) Reduplication Before Stressed σ
σ: Chamorro hu-ga-gándo
‘playing’ (Topping 1973)
(15) Infixation before final σ
σ: Hua haru-
a-po ‘not slip’ (Haiman
(16) Reduplication before final σ
σ: Dakota spa-ha-ha ‘broken off’
(Shaw 1976)
As observed in Raimy 2000:69-70, in a representational model in which
reduplication is represented as an operation of introduction of precedence
relations, infixation and reduplication become quite similar; both involve
an anchor-point specification in the realization of a morphosyntactic
feature. We claim that infixation and reduplication can only target this set
of positions. Infixation is the introduction of a precedence relation
(including new segmental material) whose start point transitively precedes
its end point, while reduplication is the introduction of a precedence
relation (including new segmental material in cases of fixed segmentism)
whose start point is transitively preceded by its end point. Given their
representational similarity, one expects that these two processes can target
the same set of non-edge constituents. The logical conclusion for Yu is that
infixation (and presumably reduplication as well) must be stated as
inviolable, non-gradient Align constraints. We adopt this inventory of
anchor points (edge constituents and prominent constituents) as complete,
noting that restricting precedence-introducing relations to these points
prevents the grammar from “counting”, and disables infixation or
reduplication after, say, the third vowel in an octosyllabic word.3
We hypothesize that the set of anchor points provided by the above theory
accounts for the range of attested variation in shm-reduplication. That is,
speakers will vary as to where they insert shm-, but the range of variation
will be confined to the inventory of anchor points in (6). When a learner
hears bagel shmagel, there are many hypotheses compatible with this
single pattern (overwrite the first consonant, overwrite the first onset,
overwrite up to the nucleus, overwrite in the stressed syllable, etc.). The
underdetermined nature of this sort of data allows speakers to postulate a
range of rule targets, all of which are compatible with the data for simple
(i.e. one-consonant-onset with initial stress) forms, but reveal variation
when speakers are required to produce more complex forms.
3 Thus mitigating the concerns raised by Downing 2001 about the generative
space of precedence-based reduplication.
Turning next to mechanics, the fundamental operations for shm-
reduplication consist of three steps: (i) adding a new precedence relation
from the last segment of the input word to the sequence /Σµ-/, (ii) adding a
new precedence relation from the sequence /Σµ-/ to the anchor point (i.e.
postulated target) of the rule, and (iii) linearizing the structure. Step (ii) is
the locus of variation. The most common target of (ii) in shm-reduplication
is the first vowel. Thus, where denotes immediate precedence, most
speakers postulate the representation as in (17):
(17) bagel
The representation in (17) is within the precedence-based model of Raimy
2000 (see also Fitzpatrick & Nevins 2002). We adopt a precedence-based
model because it allows explicit formalization of the target of a
morphophonological rule resulting in reduplication. In this model, the
output token [bagel-shmagel] is phonologically ambiguous, in the way that
I saw the man with the telescope is—it has more than one possible
structural analysis. Generalization of the phonological realization of the
semantic target of shm-reduplication from one of these analyses of the
ambiguous string generates in the massive variation we document in this
paper, as exemplified in (18).
(18) variation by anchor point in shm-reduplication
a. targets first syllable versus stressed syllable
shmobscene (30% of respondents) vs. obshmene (28%)4
b. shm placed at target in stressed syllable versus second copy
starts with shm at target in stressed syllable
confusion conshmusion (34%) vs. confusion s(h)musion
c. targets material after first consonant versus first onset
breakfast s(h)mreakfast (10%) vs. breakfast
s(h)meakfast (87%)
4 47/128 respondents (37%) preferred no output for obscene; 1 respondent
favored shmene.
5 47/113 respondents (42%) preferred no output for confusion; 13% preferred
For the purposes of comparison, we will address why we do not adopt the
shm-reduplication analysis of Alderete et. al 1999. In their model, bagel,
shm- and RED (the abstract instruction to reduplicate) are inputs to a
computation that is based on violation counting. Violations are counted
such that it is worse for the output to miss something in the input (MAXIO)
than it is for the reduplicant to contain something that is not in the base
(MAXBR); i.e. MAXIO >> MAXBR. The output [bagel-shmbagel] would
satisfy both of these constraints, but is rejected for phonotactic reasons. In
the computation of [bagel-shmagel] vs. [shmagel-shmagel], the latter loses
because it fails to include [b] from the input. However, with this same
computation, the optimal output for the input /eel/ is *[shmeel-shmeel].
Moreover, while the analysis predicts [wig-shmig] as optimal, 10/115 of
our respondents did not choose this option. The issue of how to deal with
all of the variation that we successfully model in the current paper is not
addressed, as Alderete et. al idealize to a unique output in English for shm-
reduplicating bagel.
In addition, the crosslinguistic generality of this model is questionable, as
Hindi v-reduplication yields [roti-voti] and never *[roti-vroti] for the input
/roti/, even though the latter output should be a penalty-minimization
dream come true. Finally, for proponents of the view that a simple re-
ranking of the constraints MAXIO >> MAXBR should produce an attested
reduplication pattern, it is demonstrable that their re-ranking will yield a
reduplicative system in which an input with three onset consonants will
never deliver shm- reduplication, and the optimal output for /string/ would
be [string-string]. As is more fully discussed in Nevins 2003, these
properties of Alderete et al.’s model all result from the fact that
computation is done in terms of counting the number of segments in
violation; in this case, because 3 is more than 2. Our model, as mentioned
above, is exempt from counting, and avoids such pathological predictions.
5.1 Complex Onsets
With these theoretical preliminaries in mind, we now turn to the data. One
of the first variations that we noticed in shm-reduplication is between those
who do and do not preserve elements of word-initial consonant clusters. In
order to determine the parameters of this variation, we included in our
survey four questions containing (or potentially producing) different types
of word-initial clusters: Cr (breakfast), CCr (street), Cl (floozie), and r
(rich). We also included two stimuli of this type in the follow-up survey,
broom and floss.
The treatment of Cr clusters was evenly split between the two options
mentioned above when no other factors interfered, as can be seen for
broom in (19).
(19) a. broom shmoom 39%
b. broom shmroom 37%
c. broom smroom 6%
e. no output 18%
The remaining Cr stimuli showed a greater preference for the r-less
(20) with r without r no output
a. breakfast 10% 87% 3%
b. street 4% 71% 22%
c. rich 3% 82% 13%
Cl clusters also display preference for the l-less treatment:
(21) with l without l no output
a. floss 16% 72% 11%
b. floozie 3% 88% 10%
The CR- cluster data in (19-21) suggest that speakers are most likely to
form the strongest generalization—target the segment immediately
following the first onset—when presented with underdetermined data of
the bagel shmagel type. Note, though, that a non-trivial percentage of
speakers select the more restrictive hypothesis that targets the segment
immediately following the first consonant.
5.2 Glides
Recently, several phonologists have debated an issue related to the
complex onset problem just discussed, the behavior of post-consonantal
glides. Barlow 2001 and Yip 2003 have suggested on the basis of Pig Latin
that English speakers may have two representations for a sequence like
[spju], one that syllabifies the [j] in the onset, and one that attaches it to the
nucleus. Nevins & Vaux 2003 and Idsardi & Raimy 2003, on the other
hand, have suggested that English post-consonantal [j] is always part of the
nucleus, and that the attested variation in reduplicative behavior results
from two different rule variables.
To investigate this issue we tested five words with appropriately situated
post-consonantal glides. Our results are tabulated in (22), where “+glide”
is a cover term for forms like union shmyoonion and wig shmwig, and “-
glide” represents forms that do not copy the glide, such as union
shmoonion and wig shmig.
(22) base form +glide -glide null output total
a. union 19% 75% 5% 114
b. use (verb) 8% 68% 22% 116
c. confusion 13% 38% 48% 986
d. wig 5% 92% 3% 114
e. dwarf 31% 63% 6% 199
We account for the variation in (22) by postulating that speakers
hypothesize two different targets for the reduplicative pointer: one group
adds it from /Σm/ to the first nuclear segment, giving e.g. union
shmyoonion, and the other adds a relation from shm- to the first vowel,
yielding e.g. union shmoonion. Speaker variation thus depends on the
target of the reduplicative pointer, not the representation of glides. This
analysis has the advantage over Barlow and Yip’s theory of employing a
uniform representation for all prevocalic glides, and explaining the fact
that speakers may treat /ju/ sequences differently in different language
games (cf. Idsardi & Raimy 2003)—the glide as a prenuclear segment
remains constant, but the game-particular rule contains a different anchor
point variable.
5.3 [+anterior] dissimilation
Having encountered individuals who employ sm- rather than shm- with
words containing a strident consonant {Σ τΣ Ζ δΖ σ ζ}, we included in the
survey several questions to test the extent of this phenomenon. Our
findings are summarized in (23), where “Σµ” is a cover term for forms like
witches shmitches, and “sm” represents forms that employ sm-, such as
witches smitches.
6 15 additional respondents said shmonfusion, which is not relevant to the
treatment of post-consonantal glides. These were not included in the
(23) base form Σ
µ sm null output total
a. witches 70% 17% 12% 114
b. Ashmont 70% 17% 13% 115
c. ash 66% 10%7 21% 115
d. massage 43%8 4% 48% 122
e. schnozz 66% 3% 27% 113
Why does Σ change to s precisely in forms containing one or more
stridents? We propose that it results from the rule of [anterior]
dissimilation represented in (24):
(24) [+strident] [+strident]
| |
| |
[-ant] [αant]
The system that produces ash shmass modifies (24) by delinking the
second [anterior] specification rather than the first.
5.4 Non-initial stress
Most varieties of shm-reduplication require initial stress on both the base
and the reduplicant. What happens when the base does not have initial
stress? Canonical cases of shm-reduplication do not enable us to answer
this question, because they all involve bases with primary initial stress. It
turns out that speakers employ a variety of different strategies: if the initial
syllable bears a secondary stress most speakers promote it, as in économics
shméconomics; if it does not, as in arcáde, speakers split fairly evenly
between preserving the base stress in the reduplicant (arcáde shmarcáde,
40%) and overriding the base stress (árcade shmárcade, 47%). The
alteration of the base stress in the latter form is noteworthy; it is not clear
whether this is due to the rhythm rule or a desire for prosodic identity
between the based and the reduplicant.
We tested for potential influence of lexical competitors by including the
verb permít, which contrasts with its nominal counterpart only in terms of
7 10 respondents produced smash, and 2 produced shmass.
8 Responses grouped in this category were shmassage (39), shmage (7), and
mashmage (6).
stress, the former being final and the latter being initial (pérmit). Of the
107 individuals who answered this question, only 12 (11%) allowed the
verb to shm-reduplicate, whereas 105 (98%) allowed shm-reduplication of
the noun.
We also discovered that it is not uncommon to shm-reduplicate internally
on the stressed syllable in words with non-initial stress, as shown in (25):
(25) form initial % internal % null
a. confusion shmonfusion 13 (con)s(h)musion9 44 42%
b. obscene shmobscene 33 (ob)shmene10 31 40%
c. massage s(h)massage 36 (ma)s(h)mage11 11 48%
d. terrific shmerrific 67 tershmiffic12 5 26%
e. arcade shmarcade 87 (ar)shmade13 3 9%
Though not evident from the forms in (25), internal shm-reduplication is
not restricted to the second syllable; we have heard forms such as
regulations, regushmations and understandable, undershmandable. Since
in the latter case the stress coincides with a morpheme boundary, we
decided to include a form where the two do not align, unbelievable, and
found that 45% produced shmunbelievable, 38% preferred no output, 7%
chose unshmelievable, 6% picked unbeshmievable, and 3% produced
The last of these options shows not only that shm-reduplication for some
speakers refers to the stressed syllable, but also that the reduplicant may
begin from this syllable. Such forms have been produced by Spitzer 1952
(confúsion shmooshun) and Howard Stern (masságe shmage), among
Interestingly, a small number of individuals associate shm- to both the
initial and the stressed syllables, producing forms such as forbidden,
shmorshmidden, which appeared in an episode of the Fox program
9 34% produced conshm(y)oosion, 8% produced shmoosion, and 3% produced
10 36 tokens of obshmene, and 1 of shmene.
11 6 respondents gave mashmage; 7 gave shmage.
12 Note the preservation of the base /r/, suggesting that it is ambisyllabic for
these speakers.
13 2 respondents produced arshmade, and one produced shmade.
6 Morphology and syntax
Having examined the phonological variation that results from the
inventory of anchor points and the nature of phonological rule postulation,
we turn to variation in another domain: morphologically complex words.
As it is impossible that shm- reduplication happens in the lexicon”, that
leaves one place for word-formation to happen: in the syntax. Though
shm- reduplication has a phonological form that looks a bit different from
more concatenative modes of morpheme exponence, its composition in the
syntax is still achieved through the operation of Merge. The source of
variation that we observe arises from the fact that the morpheme of which
shm-reduplication is an exponent (call it ECHO, which is quite likely
featurally underspecified and derives its interpretation through pragmatics)
may attach at multiple levels in the syntactic tree.
As Lidz (2001) discusses, echo reduplication can apply at various points
within a syntactic structure. For example, Kannada gi-reduplication can
occur either inside or outside of the case-marker on a noun. Thus, the input
[baagil] `door’ can be merged with ECHO and then [[baagil] ECHO] can be
merged with the K(ase) morpheme -annu (26a), or [baagil] can be merged
with K and then [[baagil] K] can be merged with ECHO (26b).
(26) a. baagil-giigil-annu ‘doors-shmoors-acc.’
b. baagil-annu giigil-annu ‘doors-acc-shmoors-acc’
There is no semantic difference between these derivations; it constitutes a
straightforward case of two sequences of Merge with the same morpheme
inventory, resulting in radical variation once the structure is phonologically
spelled out.
The results of our survey demonstrate that variation in the site of merge of
ECHO in the syntax obtains in shm-reduplication as well. For instance,
given a phrasal input, our respondents varied as to whether shm-
reduplication applied to the first XP in the phrase (27b-c), or the most
deeply embedded (27a):
(27) Q16 Person A: Dear, you'd better stop squandering our money
on golf clubs. Jacob wants a laptop, and you owe him big-
time after embarrassing him in front of his friends by
getting Chumbawumba to play at his bar mitzvah.
Person B: Jacob wants a laptop, _____! That spoiled brat
has enough junk already, and Chumbawumba were ahead
of their time!
a. shm- targets laptop (9%)
b. shm- targets Jacob (5%)
c. shm- targets both Jacob and laptop (3%)
d. Ineffable for whole sentences (78%)
We also examined shm-reduplication in endocentric compounds:
(28) Q11 Person A: You know what we really need? A cookie
Person B: Cookie cutter, _____! We have enough kitchen
a. shm- targets cookie (65%)
b. shm- targets cutter (10%)
c. shm- targets both cookie and cutter (7%)
d. Ineffable for compounds of this type (19%)
It is notable that exocentric compounds (e.g. walkman) are significantly
more resistant to internal schm-reduplication (contrast (29a-b)), perhaps
indicating that these are not built by the syntax.14
(29) a. shm- targets man (1%)
b. shm- targets walk (93%)
c. shm- targets both walk and man (3%)
d. Ineffable for compounds of this type (3%)
The variation in (28), however, clearly demonstrates that ECHO can attach
in various places in compounds, and supports Lidz’s (2001) hypothesis
that morphological optionality is the result of different Merge sites.
14 Names, on the other hand, seem to allow a variation that we have little
explanation for in syntactic terms, e.g. Donald Shmumsfeld (26.5%),
Shmonald Rumsfeld (18.5%).
In an interesting paper, Travis 2001 proposes that syntactic reduplication
(e.g. I need a DOCTOR doctor) involves movement of a copy to check a
feature. Presumably, one copy of the XP stays in base position, while a
higher one moves to the head with the uninterpretable feature. However,
we would like to point out that shm-reduplication can never appear in
argument position (in fact, in stark contrast to e.g. Hindi v- echo
(30) a. *I don’t want to go to Europe, Shmeurope
b. Europe, Shmeurope, who wants to go there!
(31) Optimality, schmoptimality—who needs *(it)?
We suggest that shm-reduplication involves obligatory movement to a
Topic phrase (above the position of wh-movement, as shown by (30)) from
an argument position, often with obligatory resumption (as in (31)) in the
base position.
7 Coprecedence, and allomorphy
Shm-reduplication has received some discussion as an example of an
avoidance phenomenon. McCarthy and Prince 1986:68 observe that “even
English speakers with little experience of the phenomenon…report that
words already beginning with the cluster shm cannot enter into this pattern:
*shmaltz-shmaltz (with the intended reading).” They then claim that
speakers confronted with shm-initial words replace the overwriting -m-
with -p- (shmaltz shpaltz), but our study reveals a far greater range of
variants, as shown in (32) and (33):
(32) base % shm- % null output % other
a. Schmidt 3 76 21
b. schmooze 5 64 31
c. schmuck 5 67 28
(33) alternative overwriting sequences
a. schmuck: shluck, shnuck, schnook, fluck, shpuck, vluck,
shmluck, shuck, shfuck, shvuck, smuck, fuck, shmukster, my
ass, (Bronx cheer)
b. schmooze: shnooze, flooze, shmooze, shpooze, shlooze,
vlooze, shplooze, shmmooze, mooze, wooze, commooze,
my ass
Some have suggested that shm-reduplication is infelicitous with shm-initial
words because the latter already contain the derisive semantic connotations
shm-reduplication is intended to produce. This hypothesis fails to explain
the fact that words lacking these semantics, such as Schmidt, show the
same avoidance effects.
We propose instead to link this phenomenon to the cross-linguistic fact
about most (but not all—cf. Bengali -) instances of echo reduplication
that when the fixed segmentism would result in identical base and
reduplicant, avoidance occurs, as exemplified in (34)-(36).
(34) Hindi aam vaam, paani vaani, vakil
(35) Turkish kitap mitap, *masa masa
(36) Turkish cip ciliz, dop doluz, dim dik (dissimilation with
obstruents; cf. Kelepir 1999)
A blanket statement that Antifaithfulness or non-identity is required does
not explain why it is precisely the fixed segment that dissimilates: a
presumable solution to an Antifaithfulness requirement for shm-
reduplication would be schmear, schmeak15 , but Turkish, Hindi, and
English speakers always dissimilate the fixed segment, as we saw in (34).
A theoretical question that arises, however, is how long-distance
dissimilation of a potentially unbounded sort can be stated in local terms.
In fact, as Fitzpatrick and Nevins 2002 shows, the precedence-based model
offers a quite natural derived relation, coprecedence, defined in (37).
(37) When AC and BC, then A and B are coprecedents
(denoted by AB)
The environment for dissimilation is exemplified for Turkish (38) and
Hindi (39):
15 To carry forth the spirit of the penalty-minimization view to its logical
conclusion, we put the epenthetic segment at the end of the word, minimally
disrupting continuity, and choose /k/ as the epenthetic segment, since we
know, from the KEKU theorem (Fitzpatrick, Nevins, and Vaux 2003), that /k/
is as good a candidate for the most unmarked consonant as /t/.
(38) Turkish m-reduplication (cf. (35))
a. masa
b. Ineffable structure: mm
(39) Hindi Representative Plurality (cf. (34))
a. vakil
b. Dissimilation rule: v Σ / v_
The statement of the rule for shm-reduplication is less systematic than it is
in Hindi. While the structural change clearly differs for our respondents,
the structural description for the rule, delivered by the formalism in (37), is
virtually invariant: dissimilate when two Σµ sequences coprecede.
Avoidance allomorphy looks like long-distance dissimilation. But when
two v’s are coprecedent this is a local relationship, and provides a context
for dissimilation. Shm-dissimilation occurs not as metalinguistic correction
(pace Pullum 1991, Sanders 2000), but due to a simple coprecedence-
based allomorphy rule.
8 Ineffability
The ineffability that we observe in many sub-cases of shm-reduplication,
most notably with shm-initial words, raises another serious theoretical
issue. As Gerfen 2001 observes, “a fundamental axiom of OT is that all
constraints are violable”. The idea behind this generalization is that a
grammar always picks the best of a set of candidate outputs, an idea that
stands in explicit and intentional contrast to the inviolable constraints
employed in pre-OT phonology. Absolute ungrammaticality directly
contravenes OT’s central principle of Violability. Absolute
ungrammaticality moreover plays a central role in human language,
appearing robustly in all known phonological systems.
Prince and Smolensky 1993 propose to generate certain cases of
ineffability by combining a Null Parse candidate, which is mysteriously
stipulated to satisfy all well-formedness and faithfulness constraints, with a
constraint MPARSE, which assigns a violation only to the Null Parse
candidate. This analysis fails to capture the intuition that (for example)
shm-reduplication of schmuck produces no output, rather than a
phonetically null output. Orgun and Sprouse 1999 demonstrate moreover
that the Null Parse analysis creates unresolvable ordering paradoxes in
Turkish. They conclude that grammars must be able to contain inviolable
constraints, which requires that we abandon the pivotal OT notions of
Violability and Emergence of the Unmarked. In contrast, the theory we
employ in this paper directly accounts for ineffability in terms of
inviolable constraints.
But what exactly motivates the inviolable constraint involved in echo
reduplication? We tentatively suggest that nonrecoverability is involved: if
one pronounces a form such as schmidt schmidt, luck shmuck, or jews
schmooze (the name of Rob Reiner’s talk show in the film Primary Colors)
it may not be sufficiently clear that shm-reduplication has occurred. This
analysis may be able to help explain the different degrees of lexical
blocking observed between forms where the output competes with a well-
known word, such as joe shmoe (cf. schmo), and forms where the
competing word is not well-known, as in ear shmear (cf. shmeer); the
former case was ineffable for 34% of respondents, whereas the latter was
only so for 10% of respondents.
We have shown, through our survey, that speakers have clear and
consistent linguistic intuitions, suggesting that shm-reduplication is
computed in the grammar, and that the the systematic responses of
speakers to these forms show that this is not a metalinguistic phenomenon
to be dismissed, and that the notion of metalinguistic is vacuous here, as
speakers’ strategies still manipulate phonological objects. We have
suggested that an individual’s grammar for shm-reduplication is generally
formed through underdetermined exposure, and we have identified several
distinct subtypes of shm- reduplication, and shown that, as the variation is
principled, it supports a theory of anchor points, as developed in Yu 2002
and Nevins and Vaux 2003.
Having demonstrated the importance of shm-reduplication as a source of
data for models of phonological representation, word formation, and the
status of ineffability in a grammatical computation, we would like to point
out that the importance of shm-reduplication extends even beyond its
illustration of the points we have discussed here. At this same conference,
exactly twenty years ago, Alexis Manaster-Ramer demonstrated that the
copying dependency in what he called “reshmuplication” constituted one
of the first bona fide examples that the status of natural language within a
complexity hierarchy could be conclusively derided as context-free,
We would like to thank all of the indivuals who took our survey, and Toby
Ayer, Sarah Benor, George Bergman, Susan Brunka, Justin Cooper, David
Diaz, Noam Elkies, Elissa Flagg, Rebecca Starr, and Donca Steriade for
helpful comments on earlier versions of this paper. Special thanks to
Charlie Cheever, Marc Prud’hommeaux, Jeff Surette, and Dorothy Weiss
for help with designing the survey and tabulating and analysing the data.
Alderete, John, Jill Beckman, Laura Benua, Amalia Gnanadesikan, John
McCarthy, and Suzanne Urbanczyk. 1996. Reduplication and Segmental
Unmarkedness. Manuscript, University of Massachusetts, Amherst. [Different
than the version later published in LI, which does not include the section on schm-
reduplication. Available as ROA 134-0996.]
Alderete, John, Jill Beckman, Laura Benua, Amalia Gnanadesikan, John
McCarthy, and Suzanne Urbanczyk. 1999. Reduplication with fixed segmentism.
Linguistic Inquiry 30.3: 327-365.
Barlow, Jessica. 2001. Individual differences in the production of initial consonant
sequences in Pig Latin. Lingua 111:667-696.
Costello, Nancy A. 1998. Affixes in Katu of the Lao P. D. R. Mon-Khmer Studies
28: 31-42.
Dixon, Robert. 1977. A grammar of Yidiny. Cambridge: Cambridge University
Downing, L. 2001. Review of Raimy (2000). Phonology, Volume 18, Issue 03.
December 2001. pp 445-451
Egerod, S. 1965. Verb inflection in Atayal. Lingua 15: 251-282.
Emeneau, Murray B. . Another example of the echo-word motif in Dravidian folk
tales. Journal of the American Oriental Society 59.503-5, 1939.
Feinsilver, Lillian. 1961. On Yiddish Shm-. American Speech 36: 302-3.
Fitzpatrick, Justin and Andrew Nevins. 2002. Phonological Occurrences:
Relations and Copying. Paper presented at NAPhC 2.
Fitzpatrick, Justin, Andrew Nevins, and Bert Vaux. 2003. KEKU: On relative
privilege of data in assessing relative markedness. Manuscript, MIT.
Gerfen, Chip. 2001. A critical view of licensing by cue: codas and obstruents in
Eastern Andalusian Spanish. In Lombardi, L. (ed). 2001. Segmental Phonology in
Optimality Theory. Cambridge: Cambridge UP.183-205.
Ginzburg, Jonathan and Ivan Sag. 2000. Interrogative Investigations: the form,
meaning and use of English interrogative constructions. Stanford: CSLI.
Goller, I. 1929. Five Books of Mr. Moses.
Haiman, John. 1980. Hua: A Papuan Language of the Eastern Highlands of New
Guinea. Amsterdam: John Benjamins.
Idsardi, William and Eric Raimy. 2003. Phonological representations and the
Delphic Oracle. Manuscript, University of Delaware and Swarthmore College.
(Presented at the University of Toronto)
Kelepir, M. 1999. Turkish emphatic reduplication and antifaithfulness.
Proceedings of ConSOLE 7, University of Bergen, Norway.
Lidz, Jeff. 2001. Echo Reduplication in Kannada and the theory of word
formation. The Linguistic Review 18: 375-394.
Lockwood, William B. Herabsetzendes shm- im Jiddischen. KZ 92.300-5, 1978.
Manaster-Ramer, Alexis. 1983. The soft formal underbelly of theoretical syntax.
CLS 19:256-262.
McCarthy, John. 1982. Prosodic Structure and Expletive Infixation. Language
58.3: 574-590.
McCarthy, John and Alan Prince. 1986. Prosodic Morphology 1986.
McCarthy, John and Alan Prince. 1993. Generalized alignment. In Geert Booij and
Jaap van Marle (eds.) Yearbook of morphology 1993, 79-153. Dordrecht: Kluwer
McCarthy, John and Alan Prince. 1999. Faithfulness and Reduplicative Identity.
In Kager, René, Harry van der Hulst, and Wim Zonneveld, eds., The prosody-
morphology interface. Cambridge: Cambridge University Press.
Nevins, Andrew. 2003. Overwriting: All or Nothing? Manuscript, MIT.
Nevins, Andrew and Bert Vaux. 2003. Underdetermination in language games:
Survey analysis of Pig Latin dialects. Paper presented at the LSA.
Orgun, Cemil Orhan and Ronald Sprouse. 1999. From MPARSE to CONTROL:
deriving ungrammaticality. Phonology 16.2: 191-224.
Pierrehumbert, J. and R. Nair (1995) Word Games and Syllable Structure,
Language and Speech, p. 78-116.
Prince, Alan & Paul Smolensky (1993): Optimality Theory: Constraint Interaction
in Generative Grammar. Rutgers University Center for Cognitive Science
Technical Report 2.
Pullum, Geoffrey. 1991. The great Eskimo vocabulary hoax, and other irreverent
essays on the study of language. Chicago: University of Chicago Press.
Raimy, Eric. 2000. The Phonology and Morphology of Reduplication. Berlin:
Mouton de Gruyter.
Raimy, Eric. 2000. Remarks on backcopying. Linguistic Inquiry 31: 541-552.
Sanders, Nathan. 2000. Intra-Representational Correspondence and the
Realization of Empty Morphemes. Manuscript, UC Santa Cruz.
Shaw, Pat. 1976. Theoretical issues in Dakota phonology and morphology.
Doctoral dissertation,
University of Toronto.
Southern, Mark. forthcoming. Contagious Couplings: Yiddish shm- and the
contact-driven transmission of expressives. Westport, Conn.: Greenwood
Publishing Group.
Spitzer, Leo. 1952. Confusion shmooshun. Journal of English and Germanic
Philology 51: 226-33.
Steinmetz, Sol. 1986. A Century of Yiddish in America: Yiddish and English.
Topping, Donald. 1973. Chamorro Reference Grammar. Honolulu: University of
Hawaii Press.
Travis, Lisa. 2001. The syntax of reduplication. In Minjoo Kim and Uri Strauss,
eds., Proceedings of the North Eastern Linguistics Society Annual Meeting (NELS
31), 455-469. Amherst, Mass.: GLSA.
Weinreich, Max. 1980. History of the Yiddish language. Chicago: University of
Chicago Press.
Yip, Moira. 2003. Casting doubt on the Onset/Rime distinction. To appear in
Yu, Alan. 2002. Understanding infixes as infixes. Paper presented at the Second
North American Phonology Conference, Montreal.
Zwicky, Arnold and Geoff Pullum. 1987. Plain morphology and expressive
morphology. Proceedings of the Berkeley Linguistic Society.
... First, to match the fricative-and stop-rises for place of articulation and voicing, we were forced to use ʃm-initial onsets for rises (the only other alternatives, /ð/ or /θ/, are both highly marked). Note, ʃm onsets are marginally attested, given their preservation in loanwords (e.g., schmooze) and participation in reduplication (Nevins & Vaux, 2003) – consequently some English speakers might be familiar with this combination. As we demonstrate in the General Discussion, however, our results are inconsistent with a familiarity explanation. ...
... One explanation is presented by the possibility that some of the onsets employed in our experiments are, in fact, attested in the lexicon of our participants. Specifically, one-half of the fricative-initial onsets with a small rise began with " shm " – an onset cluster that is marginally attested in English, as it is preserved in loanwords (e.g., schmooze), and takes part in productive reduplication (Nevins & Vaux, 2003, e.g., homework-shmomework). As explained earlier (under Method, Section 2.1), several considerations have led us to favor shm-onsets as the best exponent of their category (the category of voiceless coronal obstruents). ...
Full-text available
Across languages, stop-sonorant onsets are preferred to fricative-sonorant ones (e.g., pna ≻ fna), suggesting that stop-initial onsets are better formed. Here, we ask whether this preference is active in the linguistic competence of English speakers. To address this question, we compare stop- and fricative-nasal onsets (e.g., pnik vs. fnik) to matched obstruent-obstruent controls (e.g., ptik vs. fsik, respectively). Past research has shown that (a) stop-stop onsets (e.g., ptik) are dispreferred to stop-nasal onsets (e.g., pnik); and (b) dispreferred onsets tend to be misidentified (e.g., ptik → ptik). We thus reasoned that, if fricative-nasal onsets (e.g., fnik) are worse formed relative to stop-nasal ones (e.g., pnik), then fnik-type onsets should be more vulnerable to misidentification, hence, their advantage over obstruent-obstruent controls (e.g., fsik) should be attenuated. Consequently, when compared to the obstruent-obstruent baseline (e.g., ptik, fsik), misidentification should be less prevalent in stop-nasal onsets (e.g., pnik) compared to fricative-nasal ones (e.g., fnik). The results of three experiments are consistent with this prediction. Our findings suggest that English speakers possess linguistic preferences that mirror the distribution of onset clusters across languages.
... Reduplication Another seemingly 'exotic' morphological process is reduplication, where some portion of a word is repeated to make a new form. Again, English makes limited use of this mechanisms, for example, in words such as fancyschmancy (Nevins & Vaux, 2003). Other languages use reduplication much more productively, and in some cases the portion of the word that is repeated is not a syllabic or morphological unit. ...
Full-text available
Phonemes play a central role in traditional theories as units of speech perception and access codes to lexical representations. Phonemes have two essential properties: they are ‘segment-sized’ (the size of a consonant or vowel) and abstract (a single phoneme may be have different acoustic realisations). Nevertheless, there is a long history of challenging the phoneme hypothesis, with some theorists arguing for differently sized phonological units (e.g. features or syllables) and others rejecting abstract codes in favour of representations that encode detailed acoustic properties of the stimulus. The phoneme hypothesis is the minority view today. We defend the phoneme hypothesis in two complementary ways. First, we show that rejection of phonemes is based on a flawed interpretation of empirical findings. For example, it is commonly argued that the failure to find acoustic invariances for phonemes rules out phonemes. However, the lack of invariance is only a problem on the assumption that speech perception is a bottom-up process. If learned sublexical codes are modified by top-down constraints (which they are), then this argument loses all force. Second, we provide strong positive evidence for phonemes on the basis of linguistic data. Almost all findings that are taken (incorrectly) as evidence against phonemes are based on psycholinguistic studies of single words. However, phonemes were first introduced in linguistics, and the best evidence for phonemes comes from linguistic analyses of complex word forms and sentences. In short, the rejection of phonemes is based on a false analysis and a too-narrow consideration of the relevant data.
... Another expressive cluster, widely used mainly in American English, is [∫m]. The phonaestheme originated in European Yiddish 150 years ago in echo-formations like Russian/Polish tate shmate 'father shmather/rag', poezye-schmoezye 'poetry and stuff — I couldn't care less', or German gelt-shmelt 'money-who cares?' (Southern 2005, Nevins & Vaux 2003). The phrases are derived via a process involving reduplication of the base word and the replacement of the initial part of the reduplicant (segment, syllable or syllable-like element) with the initial schm-/shm-. ...
This book contains new work by prominent phonologists and it goes to the heart of current debates in phonological and linguistic theory: should the explanation of phonological variety be constraint or rule-based and, in the light of the resolution of this question, how in the mind does phonology interface with other components of the grammar? The book includes contributions from leading proponents of both sides of the argument and an introduction setting out the history, nature, and more general linguistic implications of current phonological theory.
Exponence is the mapping of morphosyntactic structure to phonological representations, a research area which is not only the traditional bone of contention between phonology and morphology, but also approached in fundamentally diverse ways in different theoretical frameworks such as Optimality Theory and Distributed Morphology: by morphological rules carrying out complex phonological operations, highly abstract morphophonological representations, and/or by phonological constraints which are sensitive to morphological information. This volume presents a synopsis of the state-of-the-art in research on exponence, based on a novel conception: Every chapter systematically discusses a specific aspect of exponence from the point of view of current theoretical morphology, but also from a theoretical phonology perspective. Topics include nonconcatenative morphology, allomorphy, iconicity, dissimilation and truncation processes. Two detailed chapters formulate a new coherent research program for exponence which integrates the central insights of the last decades and provides important new challenges for years to come. © Editorial matter and organization Jochen Trommer 2012. © The Chapters their several authors 2012. All rights reserved.
This volume endeavors to bridge one of the gaps between linguistic theory and the biological sciences by presenting a comprehensive view of phonology which simultaneously addresses linguists and those who from other fields who would like to make contact with phonological theory. It proposes a new theory of phonological computation using representations and operations informed by a broader biolinguistic perspective, breaking the human language externalization system into component parts and investigating their possible origins in cognitive abilities found throughout the animal kingdom. Issues discussed include phonology in evolutionary perspective, the role of phonology within a Minimalist conception of the language faculty, phonological operations and representations, arguments for parallel cyclicity across linguistic modules, the order of operations at the syntax/phonology interface, diachronic phonology, the role of language acquisition in language change, and the sources of linguistic variation.
Full-text available
Students often struggle to understand linguistic concepts through examples of language data provided in class or in texts. Presented with ambiguous information, students frequently respond that they do not 'get it'. The solution is to find an example of humour that relies on the targeted ambiguity. Once they laugh at the joke, they have tacitly understood the concept, and then it is only a matter of explaining why they found it funny. Utilizing cartoons and jokes illustrating linguistic concepts, this book makes it easy to understand these concepts, while keeping the reader's attention and interest. Organized like a course textbook in linguistics, it covers all the major topics in a typical linguistics survey course, including communication systems, phonetics and phonology, morphemes, words, phrases, sentences, language use, discourses, child language acquisition and language variation, while avoiding technical terminology.
Phonological generalizations that are not absolute, but rather statistic, are a traditional problem for formal phonological theory. This paper gives an example of such a phenomenon and then proposes how such a skewed lexical distribution can be understood in terms of language learning in optimality theory. If lexicon optimization is extended minimally so as to incorporate the fact that the learner is confronted with (random, phonetic) variation, individual words may change, in order to start satisfying lower ranked markedness constraints. If sufficient words have undergone such a change, this may eventually lead to constraint reranking and therefore grammatical change.
This article examines the morphosemantic and phonological structure of Russian rhyming paired words such as gogol j -mogol j , plaksa-vaksa, and devočka-pripevočka, building on the research of earlier linguists into this phenomenon. The perspective taken here is that these are echo-words formed by a correspondence relationship that simultaneously produces both rhyme and reduplication. Active linguistic subprocesses used to form rhyming paired words in Russian include fixed segment overwriting and retrieval of the reduplicant from the lexicon instead of simple copying from the base. Five structural types of rhyming paired words are distinguished, and an analysis of the segmental and prosodic nature of base-reduplicant identity in such words follows. The article also notes similarities with echo-formations in other languages, as well as some phonological constraints that have been proposed as producing such echo-formations.
Full-text available
Overwriting is modeled in Optimality Theory as a competition for a position within the derivational base (Alderete et al. 1999, Ussishkin 1997). Faithfulness constraints that are evaluated on the basis of segment counting predict a typology of languages in which (a) optimization dictates that the relative size of the affixal material determines whether it will win out and overwrite the base, and (b) optimization ensures that if both the affix and base material can surface without incurring phonotactic violations, this should be optimal. Both predictions are wrong. Hebrew denominal verb formation and Hindi echo reduplication demonstrate cases of nonconcatenative derivation in which overwriting is better understood as rule-induced change.
Full-text available
This article demonstrates that Echo Reduplication in Kannada applies equally to words, subparts of words and entire syntactic phrases. Such a rule is problematic for a lexicalist grammatical architecture because it treats the internal structure of words as parallel to the internal structure of sentences, suggesting that there is no clear division between word structure and phrase structure. Numerous variants of the lexicalist architecture are examined and rejected on empirical grounds. The data are analyzed within the framework of Distributed Morphology, in which syntactic structure provides the input to the morphological component. The article goes on to demonstrate that some pieces of morphological structure realize pieces of consistent structure whereas others arise due to featural properties of the syntactic environment of the root.
Full-text available
This article presents a rule-based serial model of reduplication that is empirically and conceptually capable of handling backcopying effects in reduplication, contrary to McCarthy and Prince's (1995) claim that serial models of phonology are incapable of adequately accounting for this phenomenon. The model of reduplication presented here claims that reduplication is the result of loops in underlying temporal precedence structures of segments in formatives and reduces over- and underapplication effects in reduplication to cases of opacity.