The American Dialect Society: Duke University Press

The American Dialect Society
Stress in Reduplicative Compounds: Mish-Mash or Hocus-Pocus?

Author(s): John M. Dienhart
Source: American Speech, Vol. 74, No. 1 (Spring, 1999), pp. 3-37
Published by: Duke University Press
Stable URL: http://www.jstor.org/stable/455746 .
Accessed: 22/11/2014 18:07
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact [email protected].
The American Dialect Society and Duke University Press are collaborating with JSTOR to digitize, preserve
and extend access to American Speech.
http://www.jstor.org
This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

All use subject to JSTOR Terms and Conditions
STRESSIN REDUPLICATIVE
COMPOUNDS:MISH-MASHOR
HOCUS-POCUS?
JOHN M. DIENHART
Odense University
have been listed and classified in various ways.

For example, Jespersen (1974, 173-83) makes a three-way distinction
based on the form of the compound. Here are Jespersen's three classes,
illustrated with what I have selected as prototypical examples:
REDUPLICATIVE COMPOUNDS
1. kernel repeated with no change: boo-boo

2. kernel repeated with change of initial consonant: hocus-pocus
3. kernel repeated with change of vowel: mish-mashl
Observe thatJespersen's classification is based purely on the form2 of the

word. It is oblivious to such things as lexical category and stress pattern.
This tripartite division is also adopted by Flexner (1975, 605-06), who
labels these "firstorder," "second order," and "third order" reduplications,
respectively.3
A different vantage point is taken by Roger Kingdon in his very useful
book The Groundwork of English Stress (1967). Here we find that reduplica-
tive compounds enter into his larger class of Imitative Compounds, which
is divided into a number of subclasses on the basis of stress pattern and
number of components in the compound (186-87).
SoJespersen focuses on form, while Kingdon focuses on stress. A natural
question now arises: is there any correlation between the form of these
compounds and their stress patterns? I know of only one source that has
addressed this question: Bauer et al. (1980). The chapter on stress, which I
wrote for that book, outlined briefly the basic relationships, as I saw them
then, between form and stress in reduplicative compounds (188-89). This
paper develops those ideas further, providing a comprehensive investigation of the relationship between form and stress, while at the same time
illuminating an interesting array of other linguistic regularities in compounds of this type.
For ease of reference I shall henceforth refer to the three prototypical
classes of reduplicative compounds as follows:
Class1: the BOO-BOO class
Class2: the HOCUS-POCUSclass
Class3: the MISH-MASHclass
3

AMERICANSPEECH 74.1 (1999)

1. Two BASICPROBLEMS
This investigation faces two basic problems at the very outset. The first
involves determining class membership itself, while the second involves
determining the stress patterns for the items in each of the three classes.
Because I am interested in determining class membership INDEPENDENTLY
OF STRESS,the solution of the second problem is not to be invoked in the
solution of the first.
At first glance, determining class membership might seem quite straightforward. We shall see below that it is not.
1.1. DETERMINING
CLASSMEMBERSHIP.
Once the set of reduplicative com-
pounds is defined, there is no problem in distinguishing one (sub)class

from another. The classes are mutually exclusive and objectively definedfor any given member of the set, class assignment can be made solely on the
nature of the phonological relationship between the kernel and the
reduplicant. The basic form problem, however, is to determine what counts
as a "reduplicative compound" in the first place.
Consider first the BOO-BOOclass. If we are interested only in constructions of the form X1X2,where X, = X2, then this class contains not only such
baby talk as boo-boo,din-din, and pee-pee,but also animal noises like arf-arf
and oink-oink, other sounds such as bang-bang and boing-boing, words like
mama, papa, and yoyo, proper nouns like Lulu, Baden-Baden, and Walla
Walla,foreign borrowings like bonbonand couscous,as well as such repeated

lexical items as fifty-fifty,my-my,and so-so.
Similar variety can be found in connection with the other two classes.
Hocus-pocusis a prototypical example of the second class, as are boogie-woogie
and roly-poly.All these compounds are formed by altering the consonant
onset of the kernel and duplicating the rest. And at least one of the
elements in the compound is in itself nonsensical. But what about brain
drain and night light? These show the same phonological pattern as hocuspocus, but each of these compounds is made up of already existing lexical
items. Moreover, there is in these compounds a grammatical and semantic
relationship between the two elements: a brain drain is a draining away of
brainy people, and a night light is a light kept burning at night. No such
relationship holds in constructions such as hocus-pocusand boogie-woogie.
What about lexical items that are orthographically written as one word,
such as cookbook,
grandstand,and payday?Even more problematic are lexical
items where the notion of "compound" is dim, at best (e.g., bozo,hobo,kiwi,
and weenie). Once again, the same phonological relationship holds between the two "elements," yielding constructions of the type CVCV,where
the vowels are identical, but the consonants are different-the defining
characteristic of Class 2.

STRESS IN REDUPLICATIVECOMPOUNDS
Finally, consider the MISH-MASH class. In addition to mish-mash, the class

includes such prototypical examples as dilly-dally,knick-knack,and wishywashy. But what about see-saw,ship-shape,sing-song, and telltale? Or baby,
khaki, and Nina? In all these, the consonants are identical, but the vowels
are different, which is the defining characteristic of Class 3.
Should all the examples cited above be regarded as members of the set
of reduplicative compounds? If so, then there is no problem of definition.
If not, where do we draw the line?
Roughly speaking, this question can be pursued in terms of the "narrow
view," the "broad view," or the "verybroad view."
1.1.1. The Narrow View.The narrow view would permit only items in
which the kernel and/or the reduplicant involve nonsense forms. Unfortunately, nonsenseis a difficult term to define. For the moment, let us treat as
nonsense any form not appearing as an entry in a standard English dictio(the
nary. Then the narrow view would permit such items as jeepers-creepers
kernel is nonsense), fiddle-faddle (the reduplicant is nonsense), and arglebargle (both the kernel and the reduplicant are nonsense). It would not
allow items such as fifty-fifty,even-steven,and flip-flop.
For convenience, I shall use the label NONSENSE RHYME4 to refer to
compounds permitted by the narrow view. Such compounds appear to
spring from a native speaker's playfulness and creativity,while nonetheless
displaying a number of phonological regularities.
1.1.2. TheBroad View.The broad view would extend the set of reduplicative compounds by allowing, in addition to nonsense forms, forms in which
both elements are standard lexical items. This would open the set to such
items as fifty-fifty(Class 1), even-steven(Class 2), and flip-flop (Class 3).
To items of this type I shall assign the label LEXICAL RHYME.
1.1.3. The VeryBroad View.In the very broad view, kernels need not be
word forms at all (sensical or nonsensical). The reduplication of any
syllable(s) will do. This would further expand the set by including such
items as yoyo (Class 1), kiwi (Class 2), and Nina (Class 3).
Such items I shall refer to as SYLLABLE RHYME.
1.1.4. ReduplicativeCompoundsas a Cline.It is obvious that each of these
views increases the scope of set membership. As with so many other types of
categories, we appear to be dealing here with a cline. In this instance, the
cline runs from prototypical members of the NONSENSE RHYME type, through
LEXICAL RHYME, to the more peripheral SYLLABLE RHYME. The resulting cline
is displayed in table 1. It can easily be established that even this three-way
division is an oversimplification. The cline actually continues within each
of the three types of rhyme. For example, the nonsense rhymes can be

AMERICAN SPEECH 74.1 (1999)
TABLE
Reduplication as a Cline
Class 1 (BOO-BOO)
Class 2 (HOCUS-POCUS)
Class 3 (MISH-MASH)
NonsenseRhyme
hubba hubba
willy-nilly
wishy-washy
LexicalRhyme
my-my
brain drain
flip-flop
SyllableRhyme
lulu
kiwi
viva
further subdivided, depending on whether it is the kernel, the reduplicant,

or both that are nonsensical.
If we consider the lexical rhymes, I would rank even-stevenas more
prototypical than brain drainor night light.The difference, I think, is due to
the nature of the relationship between the kernel and the reduplicant.
Although even-stevenis made up of two lexical items, the meaning of the
compound is not predictable from the elements themselves. In the case of
brain drain and night light, however, there is a grammatical and semantic
relationship of some sort between the two elements.
Flexner (1975, 606) makes an effort to distinguish "rhymingterms" such
as brain drain and night light from reduplicative compounds proper. He
even takes the additional step of trying to separate "intentional rhyme" (fat
cat, zoot suit) from "unintentional
rhyme" (blackjack, grandstand). Though I
sympathize with Flexner's attempt to make these distinctions, they are very
difficult to carry out in practice, and Flexner himself is often inconsistent
in his classification,
listing such items as boogie-woogie,chitchat, fuddy-duddy,
and Hell's bellsboth as reduplications and as rhyming terms (645-47). In

hotshot,and nitwitare listed first under
addition, such items as double-trouble,
"unintentional rhyming terms" (646) and then again under "intentional
rhyming terms" (647).
Turning to syllable rhymes, I feel that, for example, picnic, though
admittedly peripheral, is a better example of a reduplicative compound
than kiwi, bozo, and baby.
The problem in delimiting the set of reduplicative compounds is clearly

linked to the difficulty inherent in the notion of "nonsense," on the one
hand, and "accidental rhyme"on the other. Can these notions be objectively
defined? If so, how? If not, by what means can we determine how much of
the cline should be included in the set of reduplicative compounds?
1.1.5. Problems with the VeryBroad View. The simplest solution of all would
be to adopt the very broad view-that is, admit the full range of constructions discussed above: nonsense rhymes, lexical rhymes, and syllable rhymes.
We would then have a very simple criterion for set membership: a reduplicative compound would be any sequence X1X2,where X2 is related to X1by
means of one of the three conditions stated earlier: Class 1, X2may be equal

STRESS IN REDUPLICATIVE COMPOUNDS
to XI (e.g., boo-boo,hubbahubba,my-my,lulu); Class 2, X2 may differ from XI

by having a different consonant onset (e.g., hocus-pocus,willy-nilly,brain
drain, kiwi); or Class 3, X2 may differ from XI by having a different vowel
peak (e.g., mish-mash,wishy-washy,
flip-flop, viva).
This very broad view of reduplicative compounds seems to be the one
taken by bothJespersen and Flexner-in theory, at least, if not in practice.
Listen first to Jespersen regarding the formation of items belonging to
Class 1 (BOO-BOO):
Repetitionof the same syllableor syllablescomes naturalto all human beings and
is found veryoften in languagesas a means of strengtheningan utterance.Whatis
repeated may be an ordinaryword as in the repeated interjections Come,come.t,
Hear, hear!, Well,well!.-Further such combinations as girly-girly,goody-goody,
pretty-
pretty,talkee-talkee....Babieswill repeat long stringsof identical syllableswithout

attaching any meaning to them, and parentswill often assign meaning to them,
and thus arisesthe fertile classof wordspapa,mama,etc. [1974, 173-74]
Note that Jespersen allows "nonsense" kernels (TALKEE-talkee),
lexical kernels (COME,come), and syllabic kernels (MAma).5
Flexner (1975, 605) provides a characterization of reduplication that is

both more formal and more general:
For this dictionary,a reduplication is considered any process whereby a word,
syllable,or sound is repeated as part of an additionalsyllablein a word or as an
additionalword or word element in a compound word or phrase.
So in theory, at least, Flexner takes the widest possible view in his definition
of reduplication. However, his theory seems to be at odds with his practice,
as we saw above-in his attempt to distinguish between "rhyming terms"
and reduplication.
The problem with the very broad approach is primarily that it lets in too
many members. Whereas examples like mama and yoyoare perhaps permissible, Zulu and viva seem rather distant from our prototypical examples.
Even more remote are words like baby, khaki, Nina, puppy, and weenie.
Significantly, no examples of this type appear in any of the lists supplied by
eitherJespersen or Flexner-though Flexner's definition of reduplication
clearly invites them. And this absence is understandable. If such items were
allowed into the set of reduplicative compounds, the concept of "reduplication" would be pushed to the limit, while at the same time doing severe
violence to the concept of "compound."
The reason that items such as babyand Nina seem so peripheral to the
class of reduplicative compounds has to do, I believe, with the fact that they
do not really involve any sort of reduplicative PROCESS.
The similarity between baand byin babyor between Ni and na in Nina is due to phonological
accident rather than to any intentional reduplication.

It follows, therefore, that we cannot simply adopt the very broad view of
set membership, simple as that would be.
Other kinds of problems arise if we try to adopt the narrow viewlimiting set membership to only those items which are "nonsensical."
1.1.6. Problems with the Narrow View. "Nonsensical" constructions seem to
be the ones most often cited when writers make reference to reduplicative
compounds. Thus Crystal (1997, 325), in his discussion of reduplication,
cites helter-skelterand shilly-shally as typical examples. Similarly, the lists
supplied by both Flexner and Jespersen are replete with examples such as
Herkimer-Jerkimerand ooly-drooly (Flexner 1975, 606) and driggle-draggle and
pindy-pandy (Jespersen 1974, 177). What is it that makes such forms "nonsensical"? Somewhat surprisingly, we shall discover that "nonsense" is not
easy to come to grips with.
In the first place, any given compound as a whole must be distinguished
from the elements (kernel and reduplicant) that make up the two parts of
the compound. Take, for example, the item higgledy-piggledy. Here are two
examples of the use of this compound that I have come across in recent
works of nonfiction:
rods and slabs were suspended from wires higgledy-piggledy throughout the
room [Pinker 1997, 215]
words are not just stacked higgledy-piggledy in our minds, like leaves on an
autumn bonfire [Aitchison 1994, 5]
It is clear that higgledy-piggledyis not nonsensical. We can easily assign a
word class to the compound in these contexts (it is an adverb). In addition,
we can assign a definite meaning to the compound. The American Heritage
Dictionary of the English Language (AHD 1969), for example, defines it as "in
utter disorder or confusion." So the nonsense in higgledy-piggledyis not in
the compound itself, but rather in the individual elements that make up
the compound. In this instance, both elements are nonsensical. In other
the kernel
instances, only one of the elements is nonsensical-either
or
the
MISH-mash)
reduplicant
holy
(JEEPERS-creepers,
(fiddle-FADDLE, MOLY).
in his
What is it that leads us to view such forms as nonsensical?Jespersen,
practical manner, takes it for granted that we all know an "ordinary word"
when we see one. Anything else is "meaningless," and hence nonsense:
It will be seen that in some cases one of the forms [kernel or reduplicant] exists as
an ordinary word, or even both may; . . . but in other cases the components are in
themselves meaningless. [Jespersen 1974, 177]
Let us delve a bit deeper into this distinction between "ordinary word"
and "meaningless" (or "nonsensical") forms. As a starting point, it is

STRESSIN REDUPLICATIVECOMPOUNDS
reasonable to assume that "ordinary words" are those that appear in

standard dictionaries. It would seem to follow logically, then, that "nonsense" forms are those that do NOTappear in standard dictionaries. But
this is neither a necessary nor a sufficient condition. It is not a necessary
condition, because some nonsense forms are made up of elements that do
appear in standard dictionaries. Examples are Huff-Duff, seabee,and Mayday (I shall return to these below). It is not a sufficient condition, because
there are many English words that are not in standard dictionaries. Dictionaries are alwaysincomplete, not least because new words are being coined
all the time. Consider, for example, such recent additions to the English
vocabulary as bitmap, download, e-mail, and Internet.Though these words
are not yet found in most standard dictionaries, they are certainly not
"nonsense" forms.
Another problem is that meanings and forms change in the course of
time, so that what was once meaningful and hence not "nonsense" may
gradually lose its meaning and become "nonsensical"-except perhaps to
etymologists. Consider, for example, a construction such as willy-nilly.Here
both the kernel and the reduplicant appear to be quite nonsensical. But for
someone familiar with the history of the English language, the compound
is full of sense: it derives from will ye, nill ye? 'will you or will you not?' A
related, but more opaque construction, is shilly-shally,which derives from
Shall I? Shall I?
Or consider abbreviated forms. Do items such as hi-fi 'high fidelity' and
lit-crit'literary criticism' involve "nonsense"forms? After all, the reduplicants
fi and critare not English words. Nonetheless, they are each endowed with
considerable meaning, being recognizable shortenings of existing lexical
items.
Consider next the examples Huff-Duff,seabee,and Mayday.At first glance
these compounds appear to be constructed from existing lexical items.
On closer inspection, however, they can be seen to have little or nothing
at all to do with those items. Thus Huff-Duff has no semantic connection
to either huff or duff. Instead it is a rhyming pronunciation of the string of
letters h.fd.f., which in turn stands for 'high frequency direction finder'.
Similarly, seabeelooks like a straightforward combination of two standard
lexical items, sea and bee.But, in fact, it is a literal rendering of the acronym
C.B.,which in turn is short for 'Construction Battalion', one of many such
volunteer branches of the Civil Engineer Corps of the US Navy (OED2).
The letter C is cleverly rendered orthographically as sea (rather than ceeor
see), thus appropriately bringing in naval associations. The letter B quickly
took on a new meaning, too. OED2provides the following quotation from a
1942 issue of the Army& NavyJournal:

10
Seabeesis the new name chosen to designate the new Naval Construction Regiments.... With the name an insignia has been adopted-a flying bee, fighting
mad. On its head it sports a sailor hat. In its fore hand or leg it clutches a spitting
Tommy Gun; in its amidship hand, a wrench, and in its aft hand, a carpenter's
hammer.
As in so many other instances, we see here the consequences of the natural
tendency of the human mind to impose meaning on relatively meaningless
elements.
Sometimes we can be fooled into thinking that we are dealing with
ordinary English words when in fact items have been borrowed from
foreign languages and then mapped onto existing English forms. A prime
example is the distress call, Mayday, whose English rendering conceals the
homophonous French origin of the construction: m'aider'help me!'.
A related type of example is the Japanese expression hara-kiri ('belly' +
'cut'), which, according to the OED2, entered the English language in this
form around 1850. The form is undergoing a phonological change in
modern English, so that for many speakers it has become a reduplicative
compound, hari-kari. The OED2 views this as "corrupt" and "erroneous,"
but I doubt that the process can be stopped. We even find another "corrupt" variant in the OED2, one which has resulted from a mapping of harakiri onto ordinary English lexical items: hurry-curry. Linguistic changes of
this type are inevitable. They represent yet another technique whereby
native speakers attempt to make sense of "nonsense."
An additional problem involves trying to determine what constitutes a
"compound." Mish-mash, I would argue, is a compound, whereas babyis not.
Therefore, claiming that a form such as mish (in mish-mash) is nonsensical is
not the same as claiming that ba (in baby) is nonsensical. I would prefer to
call the former "nonsensical," the latter "meaningless."6
In written texts, compounds can often be recognized by the existence of
an orthographic clue such as a hyphen (higgledy-piggledy) or a space (holy
moly) between the two elements. This allows us to view the forms as clear
instances of "compounding," even though one or both elements may be
nonsensical. What if such orthographic clues are absent? Then we tend to
resort to the notion of "lexical item." Thus cookbookand payday are compounds. But what about such items as crisscross, hobnob, nitwit, and picnic?
Are these instances of "syllable rhyme" (like baby) or examples of "nonsense
rhyme" (like hodgepodge and mish-mash)? Even if we limit our discussion to
the least likely candidate in this list, namely picnic, we are in for something
of a surprise. I believe that most people today would view picnic as a
form. But historically speaking it is a compound. The
noncompounded
AHD provides the following description of the origin of picnic: "French

11
piquenique,probably a reduplication (influenced by obsolete French nique,

a trifle) of piquer,to pick, peck." This French word entered the English
vocabulary around the middle of the eighteenth century. Early citations in
the OED2show that the French spelling piqueniquepersisted in English for
a number of years, alongside such Anglicized spellings as pic-nic (note the
hyphenation) and pic nic (written as two words). So apparently pic and nic
were earlier felt by English speakers to be a notch above simple meaningless syllables. It took roughly 120 years (from 1748 to 1868) for the word to
assume its present orthographic form, picnic, thereby obscuring its original
compound nature.
Here, then, is a theoretical problem that has consequences on two
fronts. On the one hand, there is a problem of consistently distinguishing
NONSENSE RHYME
RHYME bumps into
from LEXICAL
RHYME,
while on the other hand
NONSENSE
SYLLABLE RHYME.
The following section offers an approach to this problem that shifts the
focus from meaning to form. This can be done by transferring attention
from the semantic notion of "nonsense" to a consideration of phonological
"accident."
1.1.7. Accidental Similarity.I wish to devise some objective criteria for
distinguishing "accidental similarity"from similarity that results from "intention" of some sort. We can start by making the following generalization:
the longer the kernel, the less likely it is that the phonological similarity
between the kernel and the reduplicant is accidental. This is simple logic:
similarities between long strings are more significant (and hence less
accidental) than similarities between short strings. Thus, no one would
argue that accident is involved in the few instances we have where threesyllable kernels play against three-syllable reduplicants: higgledypiggledy
and niminy-piminy.The case is similar for most two-syllable kernels: argleand wishy-washy.Such constructions bear the
bargle,hanky-panky,ticky-tacky,
clear mark of intentional reduplication. This leads us to our first "filter,"
the SINGLE PHONE CONDITION.
1.1.7.1. Filter #1: The Single Phone Condition. The incidence of accidental "reduplication" is most common when the kernel is monosyllabic.
This is particularly the case with syllable rhyme and when the kernel has the
form CV, as in babyand bozo.In these examples, the full construction has
the form CV$CV,where each of the "elements" is a meaningless syllable.
The "reduplication" in such cases involves the repetition of only one
phone, either the C (as in baby)or the V (as in bozo).As noted earlier, such
items are peripheral indeed to the class of reduplicative compounds. To
block such items from set membership, I propose the following filter:

12

SINGLEPHONE CONDITION(SPC): This condition
denies
membership
to
monomorphemic constructions of the form CVCV, where the "reduplication" involves only a single phone. The repeated phone can be either C or V.
Thus the SPC filters out a large number of syllable rhymes (e.g., baby, bozo,
cookie, khaki, puppy, and Zulu).
However, the SPC does not block all instances of syllable rhyme. If a
syllable consists of more than two phones, the single phone condition
obviously does not apply. Thus forms such as chitchat, flubdub, hobnob,
humdrum, nitwit, picnic, and riffraffaffre not excluded from set membership.
or polymorphemic, both the
Whether these forms are monomorphemic
kernel and the reduplicant are longer than CV, so the SPC is not applicable.
Furthermore, because the SPC applies only to MONOMORPHEMIC constructions (that is, only to instances of syllable rhyme), it does not block items
such as boy toy, hi-fi, or see-saw, which are polymorphemic. However, in a few
polymorphemic items the reduplication is as accidental as it is in baby and
bozo. For these items a second filter is needed.
1.1.7.2. Filter #2: The Affix Condition. Accidental reduplication can
clearly involve morphemes as well as meaningless syllables-for
example,
in a word containing an affix of very general distribution that, in a particular instance, happens to be phonologically similar to the root to which it is
attached. Consider the English prefixes un- and dis-. These are productively
applied to a wide variety of roots: unloved, untie; dislike, disprove. But attach
un- to -done and dis- to -miss, and we end up with accidental rhymes that are,
phonologically speaking, candidates for membership in Class 2 (undone,
dismiss). Similar accidents involve suffixation: lowly, for example, fits the
very broad condition for inclusion in Class 3. However, the "reduplications" in items such as undone, dismiss, and lowly are as accidental as those
we encountered in words like babyand bozo. To remove them from the set of
reduplicative compounds, we can add the following filter:
AFFIXCONDITION(AC). This condition denies membership to any polymor-
phemic construction consisting of an affix and a root (e.g., dismiss, lowly),

where the form of the affix is not conditioned by the phonological makeup
of the root. Rather, the affix has general distribution, attaching itself freely to
a range of roots in addition to the one in question.
I am not, in general, trying to divorce the concept of reduplication from
that of affixation. In many languages, affixation is indeed a reduplicative
process. Crystal (1997, 325), in fact, accords to affixation the place of
honor in his characterization of reduplication, which he defines as follows:
A term in morphology for a process of repetition whereby the form of a prefix/
suffix reflects certain phonological characteristics of the root. This process may be

13
found in Greek, where the initial consonant of the root is reduplicated in certain
grammatical contexts (perfective forms); e.g. /'lu:o:/, 'I loose', becomes /'leluka/, 'I
have loosed'. In English the nearest one gets to this is in 'reduplicative compound'
words, such as helter-skelter,
shilly-shally.
What makes Crystal's Greek example different from English examples
like dismiss, lowly, and undone is that the English affixes are not formed on
the basis of the root, but are simply general affixes attached to the root. The
ensuing rhyme in the English words is thus a linguistic accident rather than
the result of a linguistic process of reduplication.
1.1.8. Set Membership. The two conditions, SPC and AC, provide objective
criteria for filtering out some of the accidental instances of reduplication.
The resulting set of reduplicative compounds includes much more than
would be admitted by the narrow view (which would restrict us to nonsense
rhymes only), but considerably less than would be admitted by the very
broad view (which would allow all types of syllable rhymes as well).
The algorithm now established for set membership works as follows. It
starts by defining the broadest possible set, namely the set consisting of any
sequence X1X2, where X2 is related to X1 by being identical (boo-boo), by
differing in consonant onset (hocus-pocus), or by differing in vowel peak
(mish-mash). Entries are then filtered out (or assigned to the periphery) by
means of the SPC, which removes forms like baby, and the AC, which
removes forms like dismiss.
With the criteria for set membership in place, we can now turn to the
second problem mentioned earlier-the
problem of stress assignment.
STRESS.
Given well-defined criteria for set membership,
1.2. DETERMINING
how is the stress pattern for individual members of the set determined?
Again, this might, at first glance, seem like an easy problem to solve: ask a
native speaker or look the items up in a good dictionary. But anyone who
has worked with native speakers and dictionaries knows that variable responses are often the norm rather than the exception. Native speaker
intuition is notoriously difficult to pin down, not least when it comes to
making judgments about stress. Furthermore, many of the items to be
examined are not in every native speaker's vocabulary. How many people,
for example, know such forms as bandy-bandy, beer-beer,frish-frash, or yigyag?
While dictionaries are a big help, there are two problems here as well: (1)
not all the items under consideration can be found in standard dictionaries, and (2) in those cases where the items do appear, dictionaries sometimes differ from one another in their stress assignment (where stress is
assigned at all). This is not necessarily the fault of the lexicographers. Here,
for example, is what one of the most recent English pronouncing dictionaries tells its readers about its treatment of stress:

14
Stress patterns present one of the most difficult problems in a pronouncing

dictionary. One reason for this is that many polysyllabic words have more than one
possible stress pattern, and one must consider carefully which should be recommended. Secondly, the stress of many words changes in different contexts, and it is
necessary to indicate when this happens. [Jones 1997, xii]
To maintain some semblance of objectivity in stress assignment, I have
decided not to rely on my own judgments, but to take as my arbiter the
judgments made by one of the newest and largest of English dictionariesthe Random House Compact Unabridged Dictionary (RHCUD 1996). In addition to being one of the most comprehensive and up-to-date dictionaries
on the market, it offers a number of other useful features as well.7 In most
cases it provides an approximate date for when the word became part of the
English vocabulary; it supplies information about both American and
British English; and it comes with a fully searchable CD-ROM. This last
feature not only provides rapid access to the dictionary entries, it also
allows for searches that would be too time-consuming to be done by hand.8
This paper considers only the placement of primary stress, making no
claims about any other degrees of stress. Thus, when I say that a form such
as mish-mash is singly stressed (as opposed to doubly stressed), I mean that it
has only one primary stress; for the present discussion it is irrelevant
whether the other syllable has secondary or tertiary stress. What is relevant
is simply that it does not have primary stress.
We turn now to the data itself, taking each of the three classes in turn.
2. CLASS1: THE
BOO-BOO CLASS
The following lists contain all those examples of Class 1 (kernel repeated
with no change) that I found in the RHCUD,9 grouped into two subsets.
Consider first Class la: Single Stress:
ack-ack n [1935-40] Antiaircraft fire
baba n [1820-30] Spongelike cake
BB n [1870-75] Type of ammunition (AE)
bonbon n [1790-1800]
Candy
boo-boo n [1950-55] Error (AE baby talk)
boubou n [1960-65] Type of African garment
ca-ca n Feces (baby talk)
cha-cha n [1950-55] Type of dance
chin-chin n [1785-95] Light conversation
choo-choo n [1900-05] Train (baby talk)
chow-chow n [1785-95] Type of Chinese dog
couscous n [1590-1600] Type of food
cush-cush n [1870-75] Type of vine
dada n [1915-20] Twentieth-century group of artists

15
dik-dik n [1880-85] Type of African antelope

din-din n [1900-05] Dinner (baby talk)
doo-doo n Feces (baby talk)
dum-dum n [1895-1900] Type of bullet
frou-frou n [1865-70] Frills on clothing
grigri n [1755-65] Type of African drum
gris-gris n Variant of GRIGRI
guitguit n [1890-95] Type of tropical bird
juju n [1890-95] African amulet
khus-khus n [1800-10] Type of grass
lulu n [1855-60] Something outstanding
mama n [1545-55] Mother
no-no n [1940-45] Forbidden thing (AE)
papa n [1675-85] Father
pee-pee n [1840-50] Urine (baby talk)
pom-pom n [1740-50] Tuft (e.g., of feathers)
poo-poo n [1970-75] Feces (baby talk)
Sing-Sing n State prison in New York
sous-sous n Type of ballet jump
tam-tam n [1775-85] Gong
tom-tom n [1685-95] Drum
wee-wee n [1925-30] Urine (baby talk)
yoyo n [1915] Type of toy
The items in this list all have two common characteristics: they are nouns,
and they have monosyllabic kernels.
Class lb: Double Stress, on the other hand, is much more heterogeneous:
agar-agar n [1885-90] Gelatinlike product from certain seaweeds
Baden-Baden n City in southwest Germany
beri-beri n [1695-1705] Type of disease
Bio-Bio n River in Chile
blah-blah n [1920-25] Empty chatter (AE)
buddy-buddy adj [1960-65] Very friendly
bunyan-bunyan n [1835-45] Type of evergreen tree of Australia
chop-chop adv [1825-35] Quickly
divi-divi n [1825-35] Type of tropical American shrub
fifty-fifty adj [1910-15] Equally divided (AE)
ha-ha interj [before 1000] Used to express amusement or derision or to represent laughter
housey-housey n Type of game (British)
hubba hubba interj [1940-45] Used to express admiration
hula-hula n [1815-25] Type of Hawaiian dance
my-my interj Used to express surprise or dismay
never-never adj [1880-85] Imaginary
nonny-nonny Nonsense expression, esp in Elizabethan songs
pooh-pooh v [1820-30] Express disdain
rah-rah adj [1910-15] Very enthusiastic (AE)

16
so-so adj [1520-30] Indifferent

too-too adj [1890-95] Overly affected
tuco-tuco n [1825-35] Type of South American rodent
tut-tut interj [1585-95] Used to express contempt, disdain, or impatience;
tsk-tsk
twenty-twenty adj [1935-40] Having normal vision
Walla Walla n City in southeast Washington
wonga-wonga n [1890-95] Type of Australian vine
ye-ye adj [1960-65] Exuberant
ylang-ylang n [1875-80] Type of aromatic tree
yum yum interj [1880-85] Used to express enjoyment, esp in the taste of food
zero-zero adj [1935-40] Having zero visibility
Class lb contains a variety of word classes (including nouns): hula-hula
(noun), hubba hubba (interjection), buddy-buddy (adjective), chop-chop (adverb), and pooh-pooh (verb). Furthermore, the kernels are both monosyllabic
(so-so) and polysyllabic (fifty-fifty). Nonetheless, it is possible to characterize
the differences between Classes la and lb in the following generalization,
which establishes a clear relationship between form and stress:
IN CLASS 1 (BOO-BOO):When the compound is formed by
STRESSPLACEMENT
full repetition of the kernel, it is singly stressed (on the first element) if the
kernel is monosyllabic AND the compound is a noun (b6o-boo).It is doubly
stressed if the kernel is polysyllabic ORthe compound is not a noun (hulahuzla,yam yam).
It thus appears that stress placement
factors: syllable count and word class.
in this class is determined
by two
2.1. FILTERING BY THE AC AND THE SPC. The two filters, SPC and AC, have
very little effect on the membership of Class 1 because Class 1 constructions typically do not meet either of the two conditions for filtering. With
full repetition of the kernel, it is not likely that only a single phone will be
involved or that the construction will involve a root which is identical to an
affix.
However, one example in the RHCUD involves full repetition of a onephone kernel. This is the word o-o (1885-90), defined as a 'type of Hawaiian
bird'. Since the word is presumably monomorphemic
(despite the hyphen), it is filtered out by the SPC. It is perhaps worth noting, however, that
the stress is nonetheless predictable on the basis of the stress placement
rule for Class 1: since the kernel is monosyllabic and the compound is a
noun, it should be stressed on the first syllable. The RHCUD informs us that
this is indeed the case.
2.2. ICONICITY.Perhaps more than either of the other classes, the BOOBOO class displays a high degree of iconicity. AsJespersen

(1974, 174) has
17
pointed out, the full repetition of the base form often mirrors (or attempts
to mirror) repetition of sounds that are either "natural or produced by
human activity."
Many of these sounds are animal noises, such as arf-arf, baa-baa,gobbleand
gobble,meow-meow,moo-moo,oink-oink,peep-peep,quack-quack,tweet-tweet,
woof-woof.Such utterances function like interjections or adverbs (e.g., The
duck went qudck-qudck), and hence they are typically doubly stressed, thus
falling into subcategory lb. In baby talk, however, these forms are sometimes used as labels for the animals themselves, thus becoming nouns. As
predicted, they then take single stress and move into subcategory la (e.g.,
Lookat the littlequdck-quack).Dictionaries typically do not treat these noises
as "real"English words, and hence they are generally not listed-they can
not be found, for example, in the RHCUD, which is why I have not listed
them above.
Class 1 also contains a wide range of reduplicative compounds which are
imitations of noises stemming from sources other than animals. These, too,
are seldom found in standard dictionaries. Some common examples are
and knock-knock.
bang-bang,beep-beep,
Jespersen (1974, 175) supplies us with
quite an extensive list of sounds of this type. Each of his examples is taken
from an actual literary source, and in most of the cases he identifies the
instrument responsible for the sound itself: chip-chip (ax), chuff-chuff
(train), chug-chug(engine), chut-chut(car), click-click(needles), clank-clank
(horses), clock-clock(hansom-cab), clunk-clunk(oars in oarlocks), drip-drip
(water), hish-hish(rain), honk-honk(car horn), jug-jug (motorcycle), lock-lock
(oars), nick-nick(needles), pad-pad (bare feet), plod-plod(horses), plup-plup
(gas bubbles), puff-puff (train), ramp-ramp(sea), snip-snip (scissors), tap-tap
(knock at the door), thump-thump(crutch), tick-tick(clock), and ting-ting
(telephone). We can find a few additional noises in Flexner's list (1975,
and buzz-buzz.
645): boing-boing,boom-boom,
I have found a few exceptions to the Class 1 stress rule.
2.3. EXCEPTIONS.
These fall logically into two categories: misbehaving nominal forms and
misbehaving nonnominal forms.
2.3.1. MisbehavingNominals. Nouns can misbehave in one of two ways:

(1) they can be doubly stressed even though the kernel is monosyllabic,
and (2) they can be singly stressed even though the kernel is polysyllabic. I
have found only one example of the first type of exception: 16g-16g'the
logarithm of a logarithm'. Exceptions of the latter type are a bit more
frequent, though still rare. I have come across the following examples in
the RHCUD: bdndy-bandy'a kind of snake' (1925-30), mia-mia'an aboriginal hut' (1835-45), and willy-willy'an Australian cyclone' (1890-95).

18
2.3.2. MisbehavingNonnominals.According to the stress rule, any comclass that is not nominal should typically be doubly
pound in the BOO-BOO
stressed. Consider, however, the following sentence: Thekittywee-weedon the
carpet.The verb wee-weeis singly stressed. It may be possible to account for
this by arguing that since there is a corresponding noun (wee-wee;see Class
la) with the same stress pattern, the verb adopts the nominal pattern.
Other constructed examples using nouns from Class la would appear to
support this hypothesis: Theyoung couplechd-chaedall night; The contestants
y6yoeduntil theirarms could takeno more.
Nonetheless, there appear to be a few "genuine" nonnominal exceptions to the double stress pattern. One such is the word gaga (1915-20),
which is an adjective meaning 'crazy'. Despite the fact that it is not a noun,
it has primary stress only on the first syllable.
Another exception is go-go'performing in a discotheque; full of energy'
(1960-65), which is singly stressed even though it is adjectival (e.g., go-go
girls, go-go funds, go-go generation). Similarly, hush-hush 'highly secret,
confidential' (1915-20), though adjectival, is marked as singly stressed in
the RHCUD (I suspect, however, that in a construction such as It was all very
hush-hushmany native speakers would have a doubly stressed hush-hish, as
predicted for an adjective in this class).
The reduplicative night-night (1895-1900) appears to have two stress
patterns. According to the RHCUD, it is singly rather than doubly stressed
when used as an adverb, whereas it is said to be doubly stressed when used
as an interjection.
2.4. VARIANTS.In a slight variant of the BOO-BOO class, the two elements
are not quite identical because an extra syllable or two gets added to the
first element. This extra material typically ends in -y: blankety-blank
(188590), clickety-click(1875-80), nighty-night(1875-80), and yackety-yak(194550). Certain numerals fall into this category as well: forty-four,sixty-six,
seventy-seven,eighty-eight,and ninety-nine.The stress pattern is still predictable, however: since the extra syllable (or syllables) is added to the kernel,
the kernel is clearly polysyllabic and hence the compounds are doubly
stressed.
Another type of variant involves such constructions as back-to-back,
blowhouse-to-house,
by-blow,by-and-by,
daybyday,half-and-half,heart-to-heart,
loop-theand word-forloop, man-to-man,one-to-one,out-and-out, through-and-through,
word.In cases like these, the extra syllable is a kind of independent infix
and is not felt to be part of the kernel. Once again, our generalizations
appear to hold: though the kernel in each of these compounds is monosyllabic, the compounds as such are nonnominal and hence are doubly

19
stressed. If such compounds are nominal, they are singly stressed: muck-amuck (1840-50) and so-and-so(1590-1600).
2.5. ADDITIONALITEMS.I have found
(primarily in Flexner's
lists) a
number of other items which fall into the BOO-BOO

class, but which are not
found in the RHCUD. Some of them (like the different kinds of noises cited
above) no doubt remain peripheral to the English language, while others
have probably had a very short life span and may no longer be part of
current English.
From Kingdon (1967, 187) come prettyprettyand talkee-talkee.Flexner
bee-bee,
(1975,645) gives us baddy-baddy,
bouncy-bouncy,
buckety-buckety,
beer-beer,
coo-coo,dee-dee,dust-dust,figfig, footie-footie,footsy-footsy,fuck-fuck,fuss-fuss,
gabble-gabble,
gee-gee,Gusy-Gusy,haba haba, ho-ho, mo-mo,mop mop, nice-nice,
she-she,talk-talk,yatata yatata, yen-yen,and yuk-yuk.
pip-pip,
pat-pat,
Since I have not found any of these items in the RHCUD, no stress
patterns are supplied. Therefore, this provides the reader with additional
data for testing the hypotheses given in this paper. Native speakers who
have any of these words in their English vocabulary can determine for
themselves where the primary stress falls, and then compare their stress
placement with the predictions made for this group.
3. CLASS2: THE HOCUS-POCUS CLASS
Membership in Class 2 involves an alteration at the beginning of the
kernel, whichJespersen (1974, 180) describes as "repetition with change of
initial consonants." Strictly speaking, however, such a formulation is not
quite general enough, since it excludes such items as itsy-bitsyand okey-dokey,
where a consonant is added rather than changed. Items of this type are in
fact not found injespersen's lists (1974, 180-83), so it is possible, but not
likely, that he intended to exclude items like these. But there is no reason to
exclude them, since they obey the same rules as hocus-pocusand palsywalsy.10The proper generalization for class membership is: "Modify the
onset of the opening syllable of the kernel."
As in the BOO-BOO
class, there are two basic types of stress pattern in the
HOCUS-POCUSgroup, so there are two natural subclasses.
Class 2a includes
items with single stress:

bedspread
big-wig n
boob tube
boy toy n
braindrain
nation
n [1835-45] Outer coveringfor a bed (AE)

[1725-35] Importantperson
n [1965-70] Televisionset
[1985-90] Female sex object
n [1960-65] Loss of trainedprofessionalsto another companyor

20
chalk talk n [1830-40] Lecture using a blackboard (AE)

clap-trap n [1720-30] Pretentious but insincere or empty language
cookbook n [1800-10] Book containing recipes for cooking (AE)
crumbum n [1950-55] Objectionable or worthless person
deadhead n [1570-80] Nonpaying passenger
dream team n Group of persons with high ability (e.g., in sports)
fag hag n [1965-70] Heterosexual female who seeks out male homosexuals
fan-tan n [1875-80] Type of card game
flubdub n [1885-90] Nonsense (AE)
fly-by n [1950-55] Flight of a spacecraft close to a celestial object (AE)
gang bang n [1950-55] Group rape (AE)
grandstand n [1835-45] Main seating area of a stadium
heat-treat v [1905-10] Subject metals to controlled heating and cooling
heyday n [1580-90] Period of greatest vigor or success
hobnob v [1595-1605] Associate on very friendly terms
hodgepodge n [1615-25] Heterogeneous mixture; jumble
hotchpotch n [1350-1400] Thick soup or stew
hotshot n [1595-1605]
Skillful and often vain person
hot spot n [1925-30] Area of known danger or instability (AE)
Huff-Duff n [WWII] Method of locating enemy submarines
humdrum adj [1545-55] Without change, monotonous
jetset n [1950-55] Fashionable, wealthy people who take jets to parties and
resorts
kid-vid n [1970-75] Television programs for children
lie-by n [1640-50] Roadside pulloff (e.g., for emergency repairs) (British)
lit-crit n Literary criticism
Mayday n [1925-30] International distress signal used by ships and aircraft
night light n [1640-50] Dim light kept burning at night
nitwit n [1920-25] Stupid or foolish person
no-show n [1940-45] Someone who has made a reservation but does not
show up
payday n [1520-30] Day on which wages are given
peetweet n [1830-40] Spotted sandpiper; species of bird (AE)
peewee adj [1885-90] Very small
peg leg n [1760-70] Artificial leg
pickwick n [1860-65] Picklike implement for putting out the wick of an oil
lamp
picnic n [1740-50] Meal eaten outdoors on an excursion
plumbum n [1910-15] Lead, the metal
pop-shop n [1765-75] Pawnshop
pop-top n [1965-70] Top of a can with a ring or tab used to open the can (AE)
powwow n [1615-25] Council of or with Indians (AE)
ragbag n [1810-20] Mixture or conglomeration
ragtag adj [1880-85] Ragged, shabby
ramstam adj [1780-90] Obstinate; headstrong
rum-dum n [1890-95] Stupid or ignorant person; drunkard
seabee n [1941] Member of the construction battalion of the US Navy
swapshop n Store where secondhand goods are traded or sold
thigh high n Garment that comes up to the knee (e.g., a stocking or a boot)

21
tie-dye v [1935-40] Dye fabric by means of a certain process

wing-ding n [1925-30] Noisy celebration (AE)
yoo-hoo interj [$1920-25] Used to get someone's attention (AE)
zoot suit n [1940-45] Type of suit popularized in the 1940s (AE)
Class 2b includes items with double stress:
airy-fairy adj Unrealistic
argle-bargle n [1870-75] Vigorous discussion or dispute (chiefly British)
argy-bargy n [1530-40] Variant of ARGLE-BARGLE
arsy-varsy adj [1530-40] In a backward or thoroughly mixed-up fashion
artsy-fartsy adj [1975-80] Arty
Variant of ARTSY-FARTSY
arty-farty adj [1975-80]
boogie-woogie n [1925-30] Form of instrumental blues (AE)
chiller-diller n Frightening or suspenseful story or film
Delhi-belly n [1960-65] Diarrhea
eensy-weensy adj Tiny (baby talk)
even-steven adj [1865-70] Having no debt on either side
fuddle-duddle v To be off; depart (Canadian)
handy-andy n Handyman (nineteenth century)
handy-dandy adj [1575-85] Convenient or useful
hanky-panky n [1835-45] Unethical behavior; illicit sexual relations
hari-kari n Ceremonial suicide
harum-scarum adj [1665-75] Reckless; rash; disorganized
helter-skelter adv [1585-95] In headlong and disorderly haste
herky-jerky adv [1955-60] Progressing in a jerky or irregular manner
In ajumbled or disorderly manner
higgledy-piggledy adv [1590-1600]
highty-tighty adj Pretentious; haughty
hocus-pocus interj [1615-25] Meaningless chant used in conjuring
hoity-toity adj [1660-70] Pretentious; haughty
hokey-pokey n [1840-50] Variant of HOCUS-POCUS;
trickery
holus-bolus adv [1840-50] Altogether; all at once
holy moly interj Used to express surprise
Sinuous quasi-Oriental dance
hootchy-kootchy n [1895-1900]
hotsy-totsy adj [1925-30] Just perfect
hully-gully n [1960-65] Type of dance
Humpty-Dumpty n Egg-shaped character in a nursery rhyme
hurry-scurry n [1725-35] Disorderly haste
itsy-bitsy adj [1890-95] Very small
itty-bitty adj [1890-95] Very small
jeepers-creepers interj [1925-30] Used to express surprise (AE)
killer-diller n [1935-40] Something having a devastating effect (AE)
Characterized by excessive elegance
lardy-dardy adj [1860-65]
(chiefly
British)
loosey-goosey adj [1965-70] Relaxed; calm
lovey-dovey adj [1810-20] Amorously affectionate
mumbo-jumbo n [1730-40] Meaningless incantation
namby-pamby adj [1726] Weak; indecisive; lacking in character
niminy-piminy adj [1795-1805] Affectedly delicate or refined; effeminate
nitty-gritty n [1960-65] Essential substance; crux (AE)

22
nolens volens adj Whether willing or not

okey-dokey interj [1930-35] Used to express agreement
okle-dokle
interj
(AE)
OKEY-DOKEY
palsy-walsy adj [1930-35] Friendly or appearing to be friendly in a hearty way

raggle-taggle adj [1900-05] Ragged; shabby
razzle-dazzle n [1890-95] Showiness
Short and plump
roly-poly adj [1595-1605]
stinky pinky n Type of word game
super-duper adj [1935-40] Extremely good; marvelous
teeny-weeny adj [1875-80] Tiny (baby talk)
teensy-weensy adj Tiny (baby talk)
Sweetheart
tootsy-wootsy n [1895-1900]
walkie-talkie n [1935-40] Combined transmitter and receiver (AE)
wheeler-dealer n [1950-55] Person who wheels and deals
willy-nilly adv [1600-10] In a disorganized manner
The distribution here is similar to, but not identical with, the pattern we
found in Class 1 (BOO-BOO). Class 2a, like Class la, contains only monosyllabic kernels, and all the forms have primary stress on the first element
only. Note, however, that Class 2a contains items other than nouns (though
nouns clearly predominate):
hobnob (verb), humdrum (adj.), and yoo-hoo
This
that word class is more or less irrelevant in
strongly
suggests
(interj.).
stress assignment for members of Class 2. This hypothesis is further supported by an examination of Class 2b: in all cases, the kernel is polysyllabic.
This differs from Class lb, where we found double stress on some items
with monosyllabic kernels-such
as blah-blah, my-my, and yum yum. So we
can formulate the following generalization based solely on syllable count:
IN CLASS2 (HOCUS-POCUS): When the compound is formed
STRESSPLACEMENT
by changing the onset of the kernel, it is generally singly stressed (on the first
element) if the kernel is monosyllabic (h6dgepodge)and doubly stressed if the
kernel is polysyllabic (h6cus-p6cus).Lexical category plays only a minor role.
On the basis of our X1X2

convention, there are a number of items (all listed in the RHCUD) that
meet the phonological criterion for inclusion in Class 2, but which must be
viewed as distinctly peripheral, since they appear to be instances of accidental rhyme. They can be filtered out by the AC and the SPC. The AC will
remove items such as dismiss and undone, while the SPC will filter out such
items as bozo, cooboo, Geechee, hobo, hoodoo, kiwi, kiyi, kriegie, kudu, loco, mojo,
Provo, tepee, voodoo, weenie, and Zulu.
In all these examples, the "kernel" is monosyllabic and hence the stress
placement rule cited above predicts that primary stress should fall on the
first syllable. In all but two cases, this prediction holds. The two exceptions
are the affixed forms, which are stressed on the second syllable: dismiss and
und6ne.
3.1.
FILTERING BY THE AC
AND THE SPC.


3.2. SOME INTERESTINGFEATURESOF CLASS 2 MEMBERS.Having
23
determined
the basic rule for stress assignment in Class 2, let us examine some of the
other interesting features of this class-in particular features dealing with
the semantic, phonological, and syntactic nature of certain members.
3.2.1. Semantic Characteristics. As he so often does, Jespersen (1974) has
some pertinent observations to make. He notes that in members of this
class the second part of the compound
is felt as a playful appendix to the first. These formations have as a rule a less serious
character than those in the [MISH-MASHclass]; many of them distinctly belong to the
nursery, where it is customary in speaking to children to vary names and other
words on the pattern of Georgy-Porgy.
This childish practice explains the universal
tendency to have an initial labial consonant in the repeated syllables. [180]
Jespersen's "playful appendix" (the reduplicant)
NONSENSERHYME (as opposed
creates what I have labeled
to LEXICALRHYMEand SYLLABLERHYME). This
playful, nonsensical dimension is particularly apparent in compounds with

a polysyllabic kernel (Class 2b). As noted earlier, the "playfulness" (or
"nonsense") is not limited to the reduplicant alone, as witnessed by such
examples as argle-bargle, hanky-panky, hotsy-totsy, namby-pamby, and tootsywootsy.
3.2.2. Phonological Characteristics.Jespersen points out that the reduplicant
in "nonsense rhymes".
often starts with a labial consonant-particularly
Again, Jespersen has a valid point. Here are some examples taken from
Classes 2a and 2b (and supplemented by additional items found in Jespersen
1974, 180-82, but not in the RHCUD):
1. with /p/: hanky-panky,higgledy-piggledy,
hocus-pocus,hodgepodge,hokey-pokey,
namby-pamby,niminy-piminy,roly-poly,and stinkypinky (Jespersen: blambyHeezlum Peezlum, hockerypamby, Charlie-parlie,clatter-patter,Georgy-Porgy,
wifey-pifey,wimmenypockery,liony-piony,nosy-posy,Riminipimini, rosey-posey,
pimmeny,and wobbly-pobbly)
2. with /b/: argle-bargle,argy-bargy,crumbum,holus-bolus, itsy-bitsy,plumbum,
and
and ragbag (Jespersen: hubble-bubble,
hunkum-bunkum,rumble-bumble,
tootelusbootelus)
3. with /m/: holy moly (Jespersen: cagmag, Clydie-Mydie,Hogen Mogen, hugger
mugger,tosy-mosy,and tuzzy-muzzy).
4. with /w/: boogie-woogie,
nitwit,palsy-walsy,peetweet,peewee,pickwick,
eensy-weensy,
and tootsy-wootsy
(Jespersen: Andy Wandy,kicky-whicky,
powwow,teensy-weensy,
kiddie-widdie,kissie-wissie,peesy-weesy,
piggie-wiggie,pinkie-winkie,popsy-wopsy,
tirly-whirly,titter-wittertootsumswootsums,and
snuggly-wuggly,Sweetie-Weetie,
twisty-wisty)
Jespersen (1974, 181) adds that the reduplicant always starts with /w/ if
the kernel starts with /p/. Examples from the above list are palsy-walsy, peesy-

24
weesy, peetweet, peewee, pickwick, piggie-wiggie, pinkie-winkie, popsy-wopsy, and
powwow.Note, however, that there are a few exceptions tojespersen's claim.

Class 2a contains the following counterexamples: payday, peg leg, picnic,
plumbum, and pop-top.In all fairness to Jespersen, it should be added,
RHYME.
however, that none of these fall into the category of NONSENSE
5. with /f/: airy-fairyand artsy-fartsy(Jespersen: grumpyfrumpy)
and nolensvolens(no examplesare given byJespersen)

6. with /v/: arsy-varsy
3.2.3. Syntactic Characteristics. In The Language Instinct (1995),
Steven
Pinker makes an interesting observation about the relative order of the two
members in reduplicative compounds of the HOCUS-POCUS
type. First he asks
the following question:
Why do we say razzle-dazzleinstead of dazzle-razzle?Why super-duper,helter-skelter,
walkieharum-scarum,hocus-pocus,willy-nilly,hully-gully,roly-poly,holy-moly,herky-jerky,
talkie, namby-pamby,mumbo-jumbo,loosey-goosey,wing-ding, wham-bam, hobnob,
razzamatazz,and rub-a-dub-dub?
Then he supplies the following answer:

The word beginning with the less obstruent consonant alwayscomes before the
word beginning with the more obstruentconsonant. [170]
The point Pinker is making, which seems to me to be perfectly valid, is
based on the general classification of the sounds of natural language into
an obstruency (or sonorant) cline." Thus vowels are less obstruent (that
is, less obstructed) than consonants. And within the consonants, sonorants
(e.g., /1, r, n/) are less obstruent than fricatives (e.g., /f, s, z/), which, in
turn, are less obstruent than stops (e.g., /p, d, k/). Thus, in an item such
as razzle-dazzle,the less obstruent /r/ precedes the more obstruent /d/.12
Pinker does not elaborate further on why the less obstruent sound
should precede the more obstruent one. An answer is perhaps to be found
in the general principle of syllable sonority thatJespersen developed years
ago. Part of this theory involves the claim that as a syllable closes it moves
from less obstruent to more obstruent. Thus, in such English words as
dwarf, help, sand, we see that /r, 1, n/ precede, respectively, /f, p, d/. This is
the same pattern that Pinker has noted for the larger structures of reduplicative compounds. There may be a kind of iconicity here, in that the
closing of syllables and the "closing" of reduplicative compounds mirror
the closing of the oral cavity at the end of an utterance.
3.3. EXCEPTIONS. I have come across several exceptions to the rule for
stress placement in Class 2. These are of four types: (1) doubly stressed
items with monosyllabic kernels, (2) singly stressed items with polysyllabic
kernels, (3) items with primary stress only on the reduplicant, and (4)
items with variable stress.

25
3.3.1. DoublyStressedItemswithMonosyllabicKernels.Some instances of this

type of exception are nouns: dryfly 'artificial fly for use on the surface of the
water' (1840-50), fat cat 'wealthy person' (AE; 1925-30), grayjay 'type of
bird' (1935-40), half-staff'half-mast' (1595-1605), hi-fi 'high fidelity' (AE;
1945-50; given in the RHCUD as doubly stressed for both noun and
adjective; I suspect that many speakers have single stress here: hi-fi), ill will
'hostile feeling' (1250-1300), Joe Blow 'average citizen' (1935-40), lay day
'day in which a vessel is delayed in port' (1835-45), pall-mall'type of game;
alley on which the game was played' (1560-70), plainJane 'a drab, unattractive girl or woman' (1935-40), sci-fi'science fiction' (1950-55), and wen-yen
'formal, literary variety of written classical Chinese'.
Most of these are examples of LEXICAL RHYME-consisting here of an
adjective (e.g., fat) plus a noun (e.g., cat). The double stress pattern would
be expected for phrases that are not (yet) full compounds (contrast, for
with bldckb6ard,or WhiteHousewith whiteh6use). Class
example, bldckboard
2a contains several examples of the predictable pattern-where an adjective + noun compound is singly stressed (on the monosyllabic kernel), as it
is in blackboard:big-wig,deadhead,grandstand, h6t shot,and h6t spot.
A few other instances of this type (doubly stressed monosyllables) are
nonnominals: d6wnt6wn'adv to or in the main business section of a city'
(AE; 1825-35), Hell's bells'interjindicates vexation or surprise' (1910-15),
n6-g6'adj inoperative, canceled' (1865-70), and pdll-mdll'adv in disorderly,
headlong haste' (1570-80).
3.3.2. Singly StressedItems with PolysyllabicKernels.Examples of this type
are typically nouns: culturevulture'person with an excessive or pretentious
interest in the arts' (1945-50), finder bender'collision between vehicles in
which there is only minor damage' (1960-65), fuddy-duddy 'stuffy, oldfashioned, fussy person' (1900-05), genderbender'someone, such as a crossdresser, who blurs the differences between the sexes' (1980-85), hubblebubble'simple form of the hookah; uproar' (1625-35), htgger-mugger'disor-
der or confusion' (1520-30), p6oper-scooper

'scoop for cleaning up after an
animal that has defecated on a street or sidewalk' (1970-75), silly billy
'clownish person' (1840-50),
and tittle-tattle 'gossip or foolish chatter'
(1520-30). In all these cases the stress is on the kernel only.

I have found one example of this pattern in a nonnominal: h6nky-tonky
'adj characteristic of a honky tonk' (AE; 1890-95). The RHCUDmarks this
as singly stressed, though I suspect it may be doubly stressed for many
speakers (and hence not an exception).
3.3.3. Itemswith PrimaryStresson theReduplicantOnly.These include booh6o 'v to blubber; n sound of blubbering' (1515-25), fa-ld 'n text or refrain
in old songs' (1585-95), fal-lal 'n bit of finery' (1700-10), Locof6co 'n New

26
York City radical Democratic faction' (1835), te-hge'interjexpressing laughter; n titter; v to titter' (1250-1300), tra-ld 'n nonsense syllables sung as a
refrain, representing gaiety' (1815-25), and yo-h6 'interj used to attract
attention' (1760-70). We can also mention here two items filtered out by
the AC: dismissand und6ne (the fact that the stress pattern is at odds with
the general pattern for Class 2 as a whole is further evidence that these two
words are only accidentally related to the other members of the class).
I have found one polysyllabic example of this type: abba-ddbba("perhaps
taken from the nonsensical refrain of a popular song, 'Aba Daba Honeymoon' (1904)," RHCUD).
3.3.4. Itemswith VariableStress.There are a few items that appear to have
variable stress patterns. Two examples formed from monosyllabic kernels
are bowwow(1570-80) and squeegee(1835-45). According to the RHCUD,
the former can be either singly stressed (b6wwow) or doubly stressed
(b6ww6w),whereas the latter can be singly stressed on either the kernel
(squgegee)or the reduplicant (squeegee).We are not told in what contexts the
different stress patterns appear, but interestingly the variation appears to
be found within one and the same word class, since the RHCUD classifies
both words simply as nouns.
Examples involving polysyllabic kernels include dilly-dally'to waste time'
(1735-45), hurdy-gurdy'barrel organ' (1740-50), hurly-burly'n noisy disorder and confusion; adj tumultuous' (1520-30), and ricky-ticky
'ragtime beat'
to
the
all
of
these
can
be
both
RHCUD,
(1935-40). According
doubly and
singly stressed (the single stress falling on the kernel).
3.4. VARIANTS.At least two minor variants can be viewed as peripheral to
Class 2. One involves a plural form (rare); the other involves an epenthetic
syllable between the two parts of the compound.
3.4.1. Plurals. The item heebie-jeebies
(1905-10) ends in -s, even though it
does not appear on the kernel. This is clearly the plural marker (a single
heebie-jeebie being no doubt as rare as a single measle or a mump). As
expected, it is doubly stressed, since the kernel is polysyllabic.
3.4.2. EpentheticSyllable.Another variation consists of items that, though
they are constructed by varying the onset, have an epenthetic syllable
between the two elements of the compound. This infix is commonly -a-:
chuck-a-luck(also chuick-luck;1830-40), chug-a-lug (1955-60), ding-a-ling
(1930-35), and rub-a-dub(1780-90). These are all singly stressed (on the
kernel), as predicted-since the kernel is monosyllabic.
But there are a few exceptions. The following items, though they do not
make use of -a-, nonetheless have an epenthetic infix and are doubly

27
stressed, despite the fact that the kernels are monosyllabic: lt-DE-dd (1880(1895-1900), and h6iPOLl6i(1815-25).
85), rdzzMAtdzz
If the epenthetic syllable is -y-,constructions do not necessarily adhere to
the stress pattern for Class 2 items. Orthographically, the -y-appears to be
attached to the kernel. Strictly speaking, this makes the kernel polysyllabic
and hence double stress would be expected. And indeed the RHCUD does
give double stress in the noun ricky-tick(1935-40), but not in h6nky-tonk
(1890-95), rinky-dink(1910-15), and rinky-tink (1960-65), all three of
which can be both noun and adjective. As expected, double stress occurs in
a construction such as hitherand thither,where the kernel is polysyllabic.
3.5. ADDITIONAL
ITEMS.In the section dealing with phonological
aspects
of Class 2, a number of items fromJespersen not found in the RHCUDwere

given. This section provides some additional items gleaned from Kingdon,
Jespersen, and Flexner.
Kingdon (1967) gives stress patterns for elements in his lists (whereas
Jespersen and Flexner do not). In Kingdon the following additional Class 2
items can be found (186-87): ftzzy-wuzzy,h6otst6ots,rumbletumble,and rat
tdt.Strikingly, all four items are exceptions to the stress patterns established
for this class: h6otst6otsand rdttdtare given as doubly stressed even though
the kernel is monosyllabic, while ftuzzy-wuzzy
and rumbletumbleare said to be
singly stressed, though the kernel is polysyllabic. I suspect that both fuzzywuzzyand rumbletumbleare doubly stressed for many speakers, thus adhering to the predicted stress pattern. Native speakers of English can draw
their own conclusions, as well as test the stress rule further by means of the
following items from Jespersen and Flexner.
FromJespersen (1974, 180-82):
Bizzy-Izzy,canny-nanny, cherry-derry,
flybie-skybie,
fusty-rusty,hiddyflibber-gibber,
giddy, highty-tighty,hobjob,hockerty-crockerty,
hoddy-doddy,hoity-doity,hotsy-totsy,
huftie-tuftie,humperdee-clumperdee,
molly-dolly,ram-stram,ram-tam, randem-tandem, rantum-scantum,ranty-tanty.
Flexner has many additional examples (none of which are listed in the
RHCUD). Here are items from his lists, classified according to his own
distinctions between "rhymingterms" and reduplicative compounds proper.
The items are all taken from his "Appendix" to the Dictionaryof American
Slang (1975):
"Second order reduplications" (645-46): beddie-weddie,
boz-woz,
boogily-woogily,
ducky-wucky,footsie-wootsie,
dizzy-wizzy,
fusty-dusty,handsy-wandsy,herkimer-jerkimer,
hipper-dipper,holly-golly,hooper-dooper,
hoovus-goovus,hunkie dunkie, jeezy-peezy,
jobsie-wobsie,nasty-wasty,petsy-wetsy,ping-wing, poolsie-woolsie,poopsie-woopsie,
racket-jacket,
rangle-dangle,rep-dep,rinky-dinky,row-dow,rusty-dusty,sacky-dacky,

28
whooperslangwhang, soogie moogie, spooky-wooky,starvie-warvie, thriller-diller;,

dooper,yockdock.
"Unintentional rhyming terms" (646): back-slack,double-trouble,hot squat,
hustle-bustle,lead-head,rootin'-tootin'.
"Intentional rhyming terms" (646-47): beatfeet, bee'sknees, cakessakes, cellercurlgirl, datebait, date mate,dizzieLizzie,dizzie-wizzy,
double
smeller,chopper-copper,
trouble,drape shape,face lace, fancy-nancy,fem-sem,fly-guy,fly pie, frame dame,
freak-beak,glad lad, glad pad, gruesometwosome,handsomeransom, hen-pen,hick
dick, jug-mug, ken-ten, khaki-wacky,large charge, legal beagle, loud shroud, lush
mush, molehole, mopchop,neatereater;,
passion ration,pen yen, ping-wing,ptomain
reetpleat, screwyLouie,
domain, rabbithabit, racketjacket, ram bam, rasher-splasher;,
shrewd dude, shy guy, slick chick, sockfrock, squawky-talkie,sticky icky, stubble
trouble,stuff cuff, sudsy-dudsy,town clown, WAChack, wavy navy, zootsnoot.
Flexner also provides an appendix to his "Appendix," where we find the
following additional "intentional rhymes" (763): cheat sheet, cop shop, fag bag,
jakeflake, kick stick, no-go, panman, peekfreak, ricky-ticky,scrub club, tragic magic,
white flight.
4. CLASS 3: THE MISH-MASHCLASS
In this class, the reduplication involves a change of vowel in the kernel.

The stress pattern for members in this class appears to be the most straightforward of all three classes, as the following list will make clear:
chiffchaff n [1770-80] Type of bird
chitchat n [1700-10] Light conversation
clip-clop n [1880-85] Sound of horses' hooves on pavement
crisscross v [1810-20] Move back and forth over something
dillydally v [1735-45] Waste time; loiter
ding-dong n [1550-60] Sound of a bell
fiddle-faddle n [1570-80] Nonsense
flicflac n [1850-55] Type of dance step
flim-flam n [1530-40] Trick or deception
flip-flap n [1655-65] Sudden or unexpected reversal
flip-flop n [1655-65] Sudden or unexpected reversal
Senseless chatter
gibble-gabble n [1590-1600]
heehaw n [1805-15] Braying sound made by a donkey (AE)
hip-hop n [1985-90] Subculture of big-city teenagers
kit-cat n Any of the portraits of members of the Kit Cat Club, painted 1702-17
knick-knack n [1610-20] Ornamental trinket
mingle-mangle n [1540-50] Jumbled or confused mixture
mish-mash n [1425-75] Confused mess
mish-mosh
[1425-75]
Variant of MISH-MASH
ping-pong n [1900-05] Table tennis

rickrack n [1880-85] Narrow zig-zag ribbon
riff-raff n [1425-75] Disrespectable people
(AE)

29
riprap n [1570-80] Broken stones used for a wall or foundation

say-so n [1630-40] One's personal statement or assertion
see-saw n [1630-40] Plank on which people at opposite ends go up and down
Show indecision
shilly-shally v [1690-1700]
ship-shape adj [1635-45] In good order
sing-song n [1600-10] Monotonous rhythmical cadence
skimble-skamble adj [1590-1600]
Rambling; confused; nonsensical
or trifling talk or writing
n
[1665-75]
Meaningless
slipslop
n Variant of SEE-SAW
teeter-totter
ticktack n [1540-50] Repetitive sound, such as ticking or tapping
tick-tock n [1840-50] Alternating ticking sound
ticky-tacky adj [1960-65] Shoddy; flimsy
tittle-tattle n [1520-30] Gossip; foolish chatter
trick-track n [1645-55] Variety of backgammon
Odd or fanciful object
whim-wham n [1490-1500]
wigwag v [1575-85] Move to and fro
wish-wash n [1780-90] Foolish talk or writing
wishy-washy adj [1685-95] Lacking in decisiveness
zig-zag adj [1705-15] Characterized by sharp turns first to one side, then to
the other
While most of the entries in this list are nouns, and most contain monosyllabic kernels, these are not necessary conditions for class membership.
Several of the entries are nonnominal as well as polysyllabic (e.g., dillydally,
fiddle-faddle, and wishy-washy, respectively verb, interjection, and adjective).
So there appears to be a very straightforward generalization for stress
placement in reduplicative items of this type:
STRESSPLACEMENTIN CLASS3 (MISH-MASH): When the compound
is formed by
changing the vowel (mish-mash),it is singly stressed (on the first element)regardless of lexical category and number of syllables in the kernel.
4.1. FILTERINGBYTHE AC AND THE SPC. As in the case of Class 2, a number
of words adhere to our X1X2 convention and hence formally fall into Class
3, but must be viewed as distinctly peripheral since the "reduplications"
appear to be accidental. Such forms can be filtered out by the AC and the
SPC. The AC removes forms like lowly and fulfill, while the SPC removes
such items as baby, bobby,booby, cookie, daddy, dido, khaki, kooky, lolly, Mamie,
mummy, nanny, Nina, poppy, puppy, and viva. Even these items, however,
abide by the stress placement rule for Class 3. The only exception is the
word fulfill, which is stressed on the second syllable. In all the other cases,
primary stress falls on the first syllable, as predicted.
Both Jespersen and

4.2. PHONOLOGICALAND SYNTACTICCHARACTERISTICS.
Pinker have some interesting observations relating to the nature of the
vowel alternation and the relative order of the two elements in compounds
of this type.

30
Jespersen (1974, 176) observes that the most frequent type of vowel
change in this group is the one in mish-mash.Other items in Class 3
displaying this vowel alternation pattern include chiffchaff,chitchat, dillyknick-knack,mingledally,fiddle-faddle,flicflac, flim-flam,flip-flap, gibble-gabble,
ticky-tacky,tittle-tattle,
mangle, rickrack,riff-raff, shilly-shally,skimble-skamble,
trick-track,whim-wham,wigwag,and zig-zag.
This vowel pattern, saysJespersen (1974, 176), is "found in all parts of
the world-in all Gothonic and Romanic languages, in Greek, Lithuanian,
Turkish, Magyar, Bantu, etc." He even offers an explanation for this alternation: "Youbegin with what is light and indicates littleness and nearness
and end with the opposite.... The duller and more open sound is also
musically best adapted for the conclusion."
Pinker (1995, 167-68) makes the claim in even more general terms,
relating the vowel alternation more clearly to the concept of me-here-now
versus you-there-then:
and not faddle-fiddle?
Haveyou everwonderedwhywe sayfiddle-faddle
Whyis it pingrather than pong-pingand patter-pitter?
Why dribsand drabs,
pong and pitter-patter
rather than vice versa? Why can't a kitchen be span and spic?... The answer is that
the vowels for which the tongue is high and in the front of the mouth always come
before the vowels for which the tongue is low and in the back. No one knows why
they are aligned in this order, but it seems to be a kind of syllogism from two other
oddities. The first is that words that connote me-here-now tend to have higher and
fronter vowels than words that connote distance from "me": me versus you, here
versus there,this versus that. The second is that words that connote me-here-now
tend to come before words that connote literal or metaphorical distance from "me"
... hereand there. . ., thisand that,nowand then.

4.3. EXCEPTIONS.The stress pattern for the MISH-MASHclass is highly
regular with only a handful of exceptions. These involve a few cases of

double stress and variable stress.
4.3.1. DoubleStress.In the RHCUD I have come across the following three
items which are doubly stressed: heigh-ho'exclamation of surprise, exultation, or boredom' (1545-55), pi-pd 'Chinese lute' (1840-50), and wu-wii
'Taoist expression referring to an action which accords with nature'. To this
small list can be added King K6ng.The explanation for this last exception
may be that, being a proper name, it is stressed in the same way as, for
which, as we saw, is an exception to
example, Jo6hnSmith (cf. also J6e Blo6w,
the Class 2 stress pattern).
4.3.2. VariableStress.According to the RHCUD,variable stress is found in
the case of hoo-ha(1930-35), which is pronounced h6o-haas a noun ('uproarious commotion') and hoo-ht as an interjection (expressing mock
surprise or excitement). Additionally, tip-top(1695-1705) is singly stressed

31
as a noun ('summit'), is doubly stressed as an adverb ('very well'), and can

take either of these two types of stress when used as an adjective ('situated
at the very top').
4.4. VARIANTS.As in Classes 1 and 2, a variant form is constructed
by
infixing one or more extra syllables. If the infix is -a-, the stress appears to
conform to the basic MISH-MASH
pattern-that is, primary stress falls on the
first syllable of the compound: bric-a-brac(1830-40) and pitapat (1515-25).
However, another infix (-ety-) attaches itself to the kernel and results in
double stress for the compound as a whole: clickety-cldck(1875-80), clinketycldnk (1900-05),
and clippety-cl6p (1925-30).
ITEMS.Not surprisingly, it is possible to find a number of

4.5. ADDITIONAL
items of the Class 3 type which are not listed in the RHCUD.Kingdon (1967,
186) gives us snipsnap, which he claims is doubly stressed, and hence an
exception
to the MISH-MASHpattern. Jespersen
and Flexner provide many
more-with no marking of stress, however. So here again the native speaker

has additional data to use for testing stress predictions:
click-clack,clink-clank,clinkum-clankum,
chick-chuck,clatter-clutter,
blabber-blubber,
driggle-draggle,drip-drop,fidclip-clap,clish-clash,dimberdamber,dribble-drabble,
frish-frash,higgle-haggle,jig-jog,
fad, fix-fax, flimmery-flammery,
flipperty-flopperty,
nid-nod, niddlejiggety-joggety,
jiggle-joggle,jingle-jangle,kittle-cattle,nibble-nobble,
noddle, niffy-naffy,pick-pack,pindy-pandy,pinkie-pankie,pishery-pasherie,pishyscribble-scrabble,
rickety-rackety,
pashy,prittle-prattle,ribble-rabble,
shiffle-shuffle,slipsplatter-splutter,
pery-sloppery,spatter-sputter,
squib-squab,stick-stock,tisty-tosty,titwhittie-whattie.
twiddle-twoddle,
triddle-traddle,
[Jespersen
whimsy-whamsy,
ter-totter,
1974, 176-80]
bibble-babble,
(sic, perhaps an error for chitter-chatter),
bing-bang,chatter-chitter
ding-dang, drizzle-drazzle,
fizz-fuzz,jim-jam,jing-jang, past post, yigyag,ying-yang.
[Flexner 1975, 646]
5. CONCLUSION
This paper has examined three basic processes for producing reduplicative compounds:
Class 1, repetition with no alteration of the kernel (BOO-
BOO); Class 2, repetition
with change
of kernel
onset
(HOCUS-POCUS); and
In addition, it
Class 3, repetition with change of kernel peak (MISH-MASH).
has been proposed that the set of reduplicative compounds is best viewed as
RHYMES
RHYMES
at the other.
a cline, with NONSENSE
at one pole and SYLLABLE
An intermediate position is taken by items involving LEXICAL
RHYMES.In
order to filter out obvious cases of accidental reduplications, two conditions were formulated: the SINGLEPHONECONDITION
(SPC) and the AFFIX
CONDITION
The
SPC
filters
out
words
such
as
baby and bozo, while the
(AC).
AC filters out words like dismiss and lowly.

32

5.1. PATTERNSOF FORMATION.Semantic, phonological,
and syntactic pat-
terns are observable in the makeup of some of the compounds.

5.1.1. SemanticPatterns.Two semantic notions are particularly interesting. The first involves "nonsense," the second involves onomatopoeia.
While "nonsense rhymes" play a significant role in the discussion of reduplicative compounds, the notion of "nonsense" itself is difficult to define
with any rigor. Often, if we probe deeply enough, we find a good deal of
sense hiding behind seeming nonsense. This is true, for example, in the
case of abbreviations (din-din, lit-crit)and acronyms (Huff-Duff,seabee).It is
also true if we go back in time: what appears to be nonsense today may have
been perfectly sensible at an earlier stage of the English language (shillyshally, willy-nilly).Also significant is the role played by onomatopoeia (or
iconicity). This is particularly notable in Class 1 (ack-ack,choo-choo),but it
also appears to a limited extent in Class 2 (peetweet,yoo-hoo)and Class 3 (clipclop, ding-dong).
5.1.2. PhonologicalPatterns. It is quite common for the reduplicant in
nonsense rhymes in Class 2 to start with a bilabial consonant. This can be /p/
(hanky-panky),/b/ (itsy-bitsy),/m/ (holymoly),/If (airy-fairy),/v/ (arsy-varsy),or
/w/ (boogie-woogie).Furthermore, the reduplicant nearly always starts with
/w/ if the kernel starts with /p/ (piggie-wiggie).
5.1.3. SyntacticPatterns.In Class 3 constructions many compounds contain two front vowels (rendered orthographically by i and a). The standard
order within such compounds is for the i-element to precede the a-element
(fiddle-faddle,mish-mash,pitter-patter).
A number of reduplicative compounds have been bor5.2. BORROWING.
rowed into the English language, with or without change. Some of these
involved reduplication in the original language (bonbon,picnic), whereas
others were modified after borrowing-with the result that the modified
form became a reduplicative compound in English (Mayday,hari-kari).
ANDPRODUCTIVITY.
5.3. PERSISTENCE
Given the wide variety of techniques
and processes for the formation of reduplicative compounds, it is not

surprising that the English language has so many of them. The items from
the RHCUD listed above total 220 items (Class 1: 67; Class 2: 112; Class 3:
41). Adding to these the even larger number cited fromJespersen, Kingdon,
and Flexner gives well over 500 examples of reduplicative compounds. And
the list is not complete. Even more surprising, perhaps, is the longevity that
many of these items display.
5.3.1. Old Compounds.While it may be the case that numerous reduplicative compounds are nonce constructions that come and go in the course of

33
a generation or less (e.g., buckety-buckety,hoots toots, niffy-naffy), many of

these compounds have been around for a very long time. This is true for all
three classes. Here is a list of those RHCUD items which have been available
to English speakers and writers for more than two centuries.
Class 1: ha-ha (before 1000), so-so(1520-30), mama (1545-55), tut-tut(158595), couscous (1590-1600), papa (1675-85), tom-tom(1685-95), beri-beri
(1695-1705), pom-pom(1740-50), grigri (1755-65), tam-tam (1775-85),
chin-chin (1785-95), chow-chow(1785-95)
Class 2: hotchpotch(1350-1400), payday (1520-30), argy-bargy(1530-40), arsyvarsy (1530-40), humdrum (1545-55), deadhead (1570-80), handy-dandy
(1590(1575-85), heyday(1580-90), helter-skelter
(1585-95), higgledy-piggledy
1600), hobnob (1595-1605), hotshot (1595-1605), roly-poly(1595-1605),
willy-nilly(1600-10), hodgepodge(1615-25), hocus-pocus(1615-25), powwow
(1615-25), lie-by (1640-50), night light (1640-50), hoity-toity(1660-70),
harum-scarum(1665-75), clap-trap (1720-30), bigLwig(1725-35), hurryscurry (1725-35), namby-pamby(1726), mumbo-jumbo(1730-40), picnic
(1740-50), peg leg (1760-70), ramstam(1780-90)
Class 3: mish-mash(1425-75), mish-mosh(1425-75), riff-raff(1425-75), whimwham(1490-1500), tittle-tattle(1520-30), flim-flam (1530-40), mingle-mangle
(1540-50), ticktack(1540-50), ding,dong (1550-60), fiddle-faddle (1570(1590-1600), skimble80), riprap(1570-80), wigwag(1575-85), gibble-gabble
skamble (1590-1600), singsong (1600-10), knick-knack(1610-20), say-so
(1630-40), see-saw(1630-40), ship-shape(1635-45), trick-track(1645-55),
(1685flip-flap (1655-65), flip-flop(1655-65), slipslop(1665-75), wishy-washy
95), shilly-shally(1690-1700), chitchat (1700-10), zig-zag (1705-15), dillydally (1735-45), chiffchaff(1770-80), wish-wash(1780-90)
By comparing the number of items in each of these lists with the number of
items in the full lists, we get a rough estimate of the productivity of the
1: 13/67 = 19%;
processes involved in the formation of each class-Class
Class 2: 29/112 = 26%; Class 3: 30/41 = 73%. These tallies indicate that
Class 3 (MISH-MASH) is the least productive class. Approximately two out of
every three members of this class are more than 200 years old. This overall
picture is reinforced by an examination of the members added to each of
the classes in more recent times, as will be seen in the next section.
5.3.2. New Compounds. The following items entered the English language
during the twentieth century (for reasons which will become apparent
shortly, I have retained the distinction between subclass 2a and subclass
2b):
Class 1: choo-choo(1900-05), din-din (1900-05), fifty-fifty(1910-15), rah-rah
(1910-15), yoyo(1915), dada (1915-20), blah-blah(1920-25), wee-wee(1925(1935-40), zero-zero(1935-40), hubba
30), ack-ack(1935-40), twenty-twenty
hubba (1940-45), no-no (1940-45), boo-boo(1950-55), cha-cha(1950-55),
boubou(1960-65), buddy-buddy(1960-65), ye-ye(1960-65), poo-poo(197075)

34
Class 2a: heat-treat(1905-10), plumbum (1910-15), nitwit (1920-25), yoo-hoo

hot spot (1925-30),
($1920-25),
Mayday (1925-30),
wing-ding (1925-30),
tie-dye(1935-40), Huff-Duff (WWII), no-show(1940-45), zoot suit (194045), seabee(1941), hi-fi (1945-50), crumbum(1950-55), gang bang (195055), jetset (1950-55), brain drain (1960-65), boobtube (1965-70), fag hag
(1965-70),
pop-top (1965-70),
kid-vid (1970-75),
boy toy (1985-90)
Class 2b: raggle-taggle(1900-05), boogie-woogie(1925-30), hotsy-totsy(192530), jeepers-creepers

(1925-30), okey-dokey
(1930-35), palsy-walsy(1930-35),
killer-diller(1935-40), super-duper(1935-40), walkie-talkie(1935-40), wheelerdealer (1950-55), herky-jerky
(1955-60), Delhi-belly(1960-65), hully-gully
(1960-65), nitty-gritty(1960-65), loosey-goosey
(1965-70), artsy-fartsy(197580), arty-farty(1975-80)
Class 3: ping-pong (1900-05),
ticky-tacky (1960-65),
hip-hop (1985-90)
These twentieth-century lists indicate once again that Class 3 (MISH-MASH)is

the least productive of the three classes-Class
1: 19/67 = 28%; Class 2:
39/112 = 35%; Class 3: 3/41 = 7%. Observe that Class 2, which is the largest
class, also has the highest percentage of new members. But as the above
sublists for Class 2 indicate, two different processes predominate in the
subclasses. Lexical rhyme (e.g., boy toy) seems to be the predominant
process where monosyllabic kernels are concerned, whereas nonsense
rhyme (e.g., boogie-woogie)predominates if the kernel is polysyllabic. This is
perhaps not too surprising, given that it is easier to find rhyming pairs
among existing English monosyllabic words than among polysyllabic ones.
To create polysyllabic rhymes, it is therefore often necessary to resort to
"nonsense" forms.
5.3.3. The American Contribution. As is clear from the numerous occurrences of "AE" in the lists, American English has contributed its share of
innovations to the collection of reduplicative compounds. Here again,
Class 2 is the most productive, Class 3 the least:
Class 1: BB (1870-75),fifty-fifty (1910-15), rah-rah(1910-15), blah-blah(192025), no-no (1940-45),
boo-boo (1950-55),
cha-cha (1950-55)
Class 2: powwow(1615-25), cookbook(1800-10), voodoo(1810-20), chalk talk

(1830-40),
peetweet (1830-40),
bedspread (1835-45),
flubdub (1885-90),
yoo-hoo (.1920-25), boogie-woogie(1925-30), hot spot (1925-30), jeepers(1930creepers(1925-30), mojo(1925-30), wing-ding(1925-30), okey-dokey
35), killer-diller(1935-40), walkie-talkie(1935-40), zootsuit (1940-45), hi-fi
(1945-50), fly-by (1950-55), gang bang (1950-55), nitty-gritty(1960-65),
pop-top(1965-70)
Class 3: heehaw(1805-15), rickrack(1880-85)
5.3.4. Widespread Use. Regardless of the fact that the three classes of
reduplicative compounds show different degrees of productivity, there can
be no denying that all three classes are alive and well in the English
not just in colloquial or informal speech and writing.
language today-and

35
Earlier I supplied two examples of the use of higgledy-piggledyfrom modern

works of nonfiction. Here are examples of the current use of three other
reduplicative compounds, one from each class:
Class 1: "Perhaps the most consistent critic of the Internet is computer
expert Clifford Stoll, author of the antimanifesto Silicon Snake Oil. Stoll
POOH-POOHS
the claims that the Internet will one day swallow up all forms
of human interaction." [Kaku 1997, 50]
Class 2: "Many explanations of behavior have an AIRYFAIRY
feel to them
because they explain psychological phenomena in terms of other, equally
mysterious psychological phenomena." [Pinker 1997, 85]
Class 3: "After a while, [Hopfield] began to realize that the field of artificial
intelligence had few, if any, organizing principles. It was a loose HODGEPODGE of interesting but disjointed tidbits of knowledge." [Kaku 1997, 83]
5.4. SUMMARYOF THE STRESSRULES. In addition
to providing
an empirical
basis for the observations detailed above, the word lists have enabled us to
attain the primary goal described at the outset of this paper-namely
the
uncovering of regularities in stress assignment for each of the three classes
of reduplicative compounds. We have learned that primary stress is predictable with a high degree of reliability according to the following three rules:
STRESSPLACEMENT
IN CLASS1 (BOO-BOO): When the compound
is formed by
full repetition of the kernel, it is singly stressed (on the first element) if the
kernel is monosyllabic ANDthe compound is a noun (b6o-boo).It is doubly
stressed if the kernel is polysyllabic ORthe compound is not a noun (hulahula, ytimyum).
STRESSPLACEMENT
IN CLASS2 (HOCUS-POCUS): When the compound
is formed
by changing the onset of the kernel, it is generally singly stressed (on the first
element) if the kernel is monosyllabic (h6dgepodge),and doubly stressed if
the kernel is polysyllabic (h6cus-p6cus).Lexical category plays only a minor
role.
IN CLASS3 (MISH-MASH): When the compound
STRESSPLACEMENT
is formed by
changing the vowel (mish-mash),it is singly stressed (on the first element)regardless of lexical category and number of syllables in the kernel.
In conclusion, it has been established that the stressing of English reduplicative compounds is itself neither a mish-mash nor a case of hocus-pocus,
but rather a striking instance of regularity in language.
NOTES
1. FollowingJespersen, I use the label KERNEL

to refer to the first element of the
reduplicative compound (e.g., boo-,hocus-,mish-). I shall follow Crystal (1997, 326)
in using the label REDUPLICANT
for the second element (e.g., -boo,-pocus,-mash).
to refer to the segmented phonological makeup of these
2. I use the word FORM
reduplicative compounds, not as a synonym for wordclass or lexical category.

36
3. I am indebted to an anonymous reviewer of an earlier version of this article

for drawing my attention to the material provided by Wentworth and Flexner's
Dictionary of AmericanSlang (DAS 1975). The reviewer correctly surmised that my
discussion of reduplicative compounds would benefit from the use of a larger
empirical base. In addition to data from the DAS, I have drawn primarily on
material in Kingdon (1967), Jespersen (1974), and the computerized version of
the RandomHouse CompactUnabridgedDictionary(RHCUD 1996).
4. Throughout this paper I shall use the label RHYMEin a very broad sense. It
covers all three types of phonological parallelism between the kernel and the
reduplicant seen in the three classes discussed here: Class 1, full identity (BOO-BOO);
Class 2, kernel repeated with change of consonant onset (HOCUS-POCUS);
and Class
3, kernel repeated with change of vowel peak (MISH-MASH). In the more traditional
sense, of course, only members of Class 2 would be considered rhymes.
5. It could be argued that mamaconsists of a lexical kernel (ma), rather than a
syllabic kernel. As we shall see, it is not always easy to draw the line between
nonsense, lexical, and syllabic kernels.
6. Lurking in the shadows here is the tantalizing question of whether a "nonsense" form can be a morpheme. Is a construction such as argle-bargle,
for example,
monomorphemic or polymorphemic? Since both argle and bargleare "nonsense"
forms, it is theoretically disturbing to view them as morphemes-if we adopt the
traditional view that morphemes are the smallest MEANINGFUL units in language. But
if they are not morphemes, what are they? It appears that it could be useful to have
a theoretical construct occupying an intermediate position between the morpheme and the syllable.
7. The RHCUD (1996, vii) reports, "This is the first unabridged American
dictionary to list the dates of entry into the language of vocabulary items, and
similarly to record which words are of specifically American origin." We are also
informed that a "particular effort has been made to specify the contribution to
English vocabulary of non-European languages." This is reflected in some of the
"exotic" entries in the lists which we will be examining.
8. For example, one can search all word definitions for the occurrence of, e.g.,
"rhyming compound," "redupl.," "imit.,"or "babytalk."One can also devise various
types of search strings using variables. Knowing, for example, that many reduplicative compounds contain kernels ending in the letter -y, one can search for occurrences of *y-*y,which will yield such Class 1 items as buddy-buddy,
fifty-fifty,and mymy. It is even more productive in finding instances of Class 3, such as eensy-weensy,
fuddy-duddy,and hanky-panky.
9. The data in this paper was collected as follows. First I found all relevant
examples in Jespersen (1974), Kingdon (1967), and DAS (1975). I then looked
for these examples in the RHCUD,using the CD-ROM. Every one of the examples
found in the RHCUDwent into the lists-provided that stress was supplied by the
RHCUD. Items from Jespersen, Kingdon, and the DAS that were not listed in the
RHCUD (or that were in the RHCUD but without stress assignment) have also
been included in this paper, but not in the lists themselves. Instead they are
included separately (without stress assignment) under the heading "Additional
Items."
But the word lists contain additional items as well. Once I had exhausted the lists
supplied by Kingdon, Jespersen, and the DAS, I continued the search on the CD-

37
ROM, using various look-up strategies such as those described in the preceding
note. Thus the main characteristic of the data is that it contains only items actually
listed, with stress assignment, in the RHCUD.
In the word lists, the RHCUDdates (e.g., 1935-40) indicate approximately when
the item entered the English language. If the date is preceded by "+",it means that
the item in question is probably earlier than the date given. No date is given in the
word lists if the RHCUDlists no date. AE indicates an American English innovation.
10. Flexner (1975, 645) apparently agrees, since he treats items like okey-dokey
as
standard members of his "second order" reduplications.
11. The concepts of sonority and obstruency are converses. Maximal sonority is
equivalent to minimal obstruency.
12. An interesting question arises in examples such as holy moly and hobnob,
where /h/ precedes, respectively, /m/ and /n/. If English /h/ is viewed as a fricative,
then these items violate Pinker's principle, since we would have a fricative preceding a sonorant. There is a considerable literature, however, which discusses the
special behavior of /h/ in many natural languages and argues for assigning this
sound to a special category (see, for example, Kenstowicz 1994, 456-61).
REFERENCES
Aitchison, Jean. 1994. Wordsin the Mind: An Introductionto the Mental Lexicon.2nd
ed. Oxford: Blackwell.
AHD. AmericanHeritageDictionaryof theEnglish Language. 1969. Ed. William Morris.
New York:American Heritage.
Bauer, Laurie,John M. Dienhart, Hans H. Hartvigson, and LeifKvistgaardJakobsen.
1980. AmericanEnglish Pronunciation.Copenhagen: Gyldendal.
Crystal, David. 1997. A Dictionary of Linguistics and Phonetics. 4th ed. Oxford:
Blackwell.
DAS. Dictionary of American Slang. 1975. Comp. and ed. Harold Wentworth and
Stuart Berg Flexner. 2nd supplemented ed. New York:Crowell.
Flexner, Stuart Berg. 1975. "Introduction to the Appendix" and "Appendix." DAS
596-608, 609-55.
Jespersen, Otto. 1974. A ModernEnglish Grammaron HistoricalPrinciples. Part VI:
Morphology.London: Allen.
Jones, Daniel. 1997. English PronouncingDictionary. 15th ed. Ed. Peter Roach and
James Hartman. Cambridge: Cambridge UP.
Kaku, Michio. 1997. Visions:How ScienceWill Revolutionizethe Twenty-FirstCentury.
New York:Anchor.
Kenstowicz, Michael. 1994. Phonology in Generative Grammar.Cambridge, MA:
Blackwell.
Kingdon, Roger. 1967. The Groundworkof English Stress.London: Longman. (First
published in 1958.)
OED2. OxfordEnglish Dictionary.1989. 2nd ed. 20 vols. Oxford: Clarendon.
Pinker, Steven. 1995. TheLanguage Instinct. London: Penguin.
. 1997. How theMind Works.New York:Norton.
RHCUD. Random House CompactUnabridgedDictionary. 1996. Special 2nd ed. with
CD-ROM. New York:Random.


The American Dialect Society: Duke University Press

Uploaded by

Copyright:

Available Formats

The American Dialect Society: Duke University Press

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The American Dialect Society: Duke University Press

Uploaded by

Copyright:

Available Formats

The American Dialect Society

Stress in Reduplicative Compounds: Mish-Mash or Hocus-Pocus?

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

have been listed and classified in various ways.

1. kernel repeated with no change: boo-boo

Observe thatJespersen's classification is based purely on the form2 of the

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICANSPEECH 74.1 (1999)

pounds is defined, there is no problem in distinguishing one (sub)class

Walla,foreign borrowings like bonbonand couscous,as well as such repeated

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

Finally, consider the MISH-MASH class. In addition to mish-mash, the class

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICAN SPEECH 74.1 (1999)

further subdivided, depending on whether it is the kernel, the reduplicant,

rhyme" (blackjack, grandstand). Though I

listing such items as boogie-woogie,chitchat, fuddy-duddy,

and Hell's bellsboth as reduplications and as rhyming terms (645-47). In

The problem in delimiting the set of reduplicative compounds is clearly

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

STRESS IN REDUPLICATIVE COMPOUNDS

to XI (e.g., boo-boo,hubbahubba,my-my,lulu); Class 2, X2 may differ from XI

pretty,talkee-talkee....Babieswill repeat long stringsof identical syllableswithout

Flexner (1975, 605) provides a characterization of reduplication that is

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICANSPEECH 74.1 (1999)

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

reasonable to assume that "ordinary words" are those that appear in

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICAN SPEECH 74.1 (1999)

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

piquenique,probably a reduplication (influenced by obsolete French nique,

while on the other hand

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICAN SPEECH 74.1 (1999)

phemic construction consisting of an affix and a root (e.g., dismiss, lowly),

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICAN SPEECH 74.1 (1999)

Stress patterns present one of the most difficult problems in a pronouncing

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

STRESS IN REDUPLICATIVE COMPOUNDS

dik-dik n [1880-85] Type of African antelope

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICAN SPEECH 74.1 (1999)

so-so adj [1520-30] Indifferent

in this class is determined

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

(1974, 174) has

2.3.1. MisbehavingNominals. Nouns can misbehave in one of two ways:

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICANSPEECH 74.1 (1999)

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

number of other items which fall into the BOO-BOO

items with single stress:

n [1835-45] Outer coveringfor a bed (AE)

This content downloaded from 193.6.50.23 on Sat, 22 Nov 2014 18:07:52 PM

AMERICAN SPEECH 74.1 (1999)