Christiani Lehmanni inedita, publicanda, publicata
Pleonasm and hypercharacterization
huius textus situs retis mundialis
dies manuscripti postremum modificati
occasio orationis habitae
11. Internationale Morphologietagung Wien, 14.-17.02.2004
volumen publicationem continens
Booij, Geert E. & van Marle, Jaap (eds.), Yearbook of
Morphology 2005. Heidelberg: Springer
Pleonasm and hypercharacterization
University of Erfurt
Hypercharacterization is understood as pleonasm at the level of grammar. A scale of
strength of pleonasm is set up by the criteria of entailment, usualness and contrast. Hy-
percharacterized constructions in the areas of syntax, inflection and derivation are ana-
lyzed by these criteria. The theoretical basis of a satisfactory account is sought in a holis-
tic, rather than analytic, approach to linguistic structure, where an operator-operand struc-
ture is formed by considering the nature of the result, not of the operand. Data are drawn
from German, English and a couple of other languages. The most thorough in a number
of more or less sketchy case studies is concerned with German abstract nouns derived in
-ierung (section 3.3.1). This process is currently so productive that it is also used to hy-
percharacterize nouns that are already marked as nominalizations.
(German Übercharakterisierung) may be introduced per ostensionem:
it is visible in expressions such as those of the second column of T1.
T1. Stock examples of hypercharacterization
language hypercharacterized basic surplus element
der einzigste ‘the most only’ der einzige ‘the only’ superlative suffix –st
children, brethren childer, brether plural suffix –en
While it is easy, with the help of such examples, to understand the term and get a feeling for
the concept ‘hypercharacterization’, a precise definition is not so easy. The concept has, in
fact, never been formally defined. Most of the time it has been taken for granted, and often it
has been explicitly equated with neighboring concepts. The concepts against which it must be
delimited include pleonasm, tautology, redundancy, reinforcement and hypercorrection. Some
of these are well established in certain scientific disciplines, others are no clearer than hyper-
characterization itself. I will therefore
I thank Giorgio Banti, Dagmar Haumann, Johannes Helmbrecht, Nils Jahn, Yoko Nishina, Su-Rin Ryu and two
anonymous reviewers of the Yearbook of Morphology for helpful comments on the first draft and for a couple of
The word overcharacterization ‘exaggeration, caricature’ is not a technical term of linguistics.
Christian Lehmann, Pleonasm and hypercharacterization 2
1. start by defining pleonasm and delimiting it against neighboring concepts;
2. articulate the concept by reviewing a set of suggestive cases;
3. define hypercharacterization as a specific kind of pleonasm;
4. describe a set of cases of hypercharacterization within the framework outlined so far;
5. draw some conclusions which are of relevance for linguistic theory.
Since this procedure is not entirely deductive, but instead both based on an intuitive under-
standing of the concept of hypercharacterization and inspired by a variety of data, the defini-
tion resulting from it will be open to discussion and further refinement.
Pleonasm and hypercharacterization are absolutely pervasive at different levels of style
and at all the levels of the linguistic system, from discourse down to inflectional morphology
and even to phonology.
Moreover, pleonasm has obvious rhetorical and poetic functions
which would deserve a study of its own. In this paper, the approach is purely linguistic: the
structure and linguistic (communicative, semantic, grammatical) function of hypercharacter-
ized expressions in syntax and morphology will be studied.
From among the concepts akin to hypercharacterization in its semantic field, ‘hypercor-
rection’ must be separated out. Hypercorrection is the use of an expression X, in an attempt
to speak correctly, in a context C where the norm forbids it, the background being that X does
not occur at all in unconstrained colloquial speech, but is required by the norm in certain con-
texts other than C. Hypercorrection is frequent in situations where the speaker feels it would
be important to conform to the norm, for instance in language acquisition. A typical example
is Whom shall I say was calling? Hypercorrection has nothing to do with the topic under
The most general concept in our domain is redundancy. A message is redundant iff it con-
tains such elements which contribute nothing to the information not already conveyed by the
rest of the message. Repeating an utterance is redundant, and much of grammatical agree-
ment, as in German eine alte Eule (
) as compared to English an
old owl, is redundant.
However, a simple information-theoretical conception of redundancy does not lead us
very far in the analysis of linguistic structure. In particular, a simple-minded conception of
redundancy where ‘redundant’ implies ‘superfluous’ and therefore ‘useless’ would be inade-
quate. Redundancy fulfills functions at all levels of communication and grammatical struc-
ture. At the highest level (which is well recognized in information theory, too), redundancy
ensures understanding even under difficult communication conditions. At the level of com-
municative intentions, it may be employed to overwhelm or impress the receiver. Redundancy
may have poetic functions in the sense of Jakobson’s (1960) projection of paradigmatic rela-
tions onto the syntagmatic axis. And last not least, the combination of partly or wholly syn-
In phonemics, a phoneme is hypercharacterized if it differs from the closest less marked members of its subsys-
tem by more than one feature value, as e.g. in Ancient Greek the high round vowel opposed to /o/ was not /u/,
but /y/. On the syntagmatic dimension, Sherer 1994 applies the concept of hypercharacterization to syllable
Christian Lehmann, Pleonasm and hypercharacterization 3
onymous elements may fulfill various grammatical functions, as we shall see in section 3.1.
Thus, ‘redundant’ does not by any means entail ‘functionless’.
The concepts of pleonasm and tautology have been current in rhetoric, linguistics and
philosophy since antiquity. At the beginning, we can exclude the logical approach and with it
the meaning of the term tautology in propositional logic, where it refers to a proposition that
is always true independently of the truth values of its constituents, as e.g. It will rain or it will
not rain. In rhetoric and linguistics, the two terms have been treated as interchangeable and
been variously delimited against each other with about equal frequency. The following prop-
erties have usually played a criterial role:
• Tautology and pleonasm are kinds of redundancy.
• Both terms are used as nomina acti, referring to linguistic acts (specifically to rhetorical
and as nomina patientis, referring to expressions resulting from such acts.
• Both of them refer to complex expressions some of whose constituents bear some seman-
• Traditionally, the constituents in question are words or phrases.
• In the typical case, the similarity in question obtains between just two constituents.
• In the clearest and extreme case, one constituent is synonymous with the other. That case
is called tautology.
• In less extreme cases, the meaning of one constituent entails the meaning of the other
without being identical to it. ‘Pleonasm’ may either be restricted to this relationship or be
used as a cover term for both kinds of semantic relation.
Tautology may be illustrated by the examples in E1:
Apart from their semantic properties, many collocations like those of E1 also have poetic
qualities, to be seen in such features as alliteration and meter. Such expressions may in fact
even be motivated to some extent by the analogical model of binomials like kith and kin.
Pleonasm will here be used to include tautology as a special case. In general, a pleonastic
expression contains constituents – typically two – one of which implies – technically: entails
– the other. Thus, the meaning of the latter constituent is part of the meaning of the former.
For instance, the meaning of return is roughly ‘go back’. Return back is pleonastic because
the meaning of back is included in, or implied by, the meaning of return. This shared seman-
tic component in pleonastic expressions will be called the focal component.
We will drop the traditional implicit presupposition that the expressions in question are
words or phrases in order to be able to apply these concepts at the morphological level. Hy-
percharacterization will be defined in section 2.2.3 as pleonasm at the level of grammar.
Repetition may be regarded as a special kind of tautology where the relation between the
elements involved in the process is (type-)identity. We will have occasion to come back to it
in section 2.5. At this point, the relation between the concepts introduced so far may be visu-
alized as in S1:
Cf. Lausberg 1990, §§ 502, 604.
Given this, the title of this paper is slightly ill-formed; it should read: ‘Pleonasm and, in particular, hyperchar-
Christian Lehmann, Pleonasm and hypercharacterization 4
S1. Conceptual field of pleonasm
We may now refine the definition of pleonasm:
An expression E
, is pleonastic iff it contains a meaning component F that is
included in the meaning of more than one E
Typically, F is the intension of one E
and properly included in the intension of E
, reduces to a binary construction E
The notion of a component Q being properly included in a meaning P can be explicated as
semantic entailment: P (x) entails Q (x). This formula would directly fit such examples as
Essential (x) → wesentlich (x), but would not apply in a straightforward way to others such as
return (x) → back (x). We will therefore assume that the pleonastic character of an expression
is tested by an implication E
. For instance, return (x) → return back (x),
and example (x) → specific example (x).
Now for any construction E
, the entailment E
is unusual and defines its
pleonastic character, while the reverse entailment E
is always valid provided the
is at all semantically compositional. Now ((p → q) & (q → p)) ↔ (p ↔
q). In other words, what we have is synonymy of a pleonastic construction with one of its
members. In this light, the difference between a pleonastic and a tautological construction
consists in the fact that in a pleonastic construction, one member is synonymous with the con-
struction, while in a tautological construction, each member is synonymous with the construc-
In ancient rhetoric, the hyperonym for pleonasm is adiectio, i.e. the addition of linguistic
material. Its opposite is detractio, the suppression of linguistic material,
which we may trans-
late by pregnancy (conciseness). The publicity slogan in E2 is a relatively recent example:
‘Germany’s most [common] credit card’
one of whose manifestations is ellipsis
Christian Lehmann, Pleonasm and hypercharacterization 5
As we shall see below, pleonasm and hypercharacterization are of interest to the grammarian
because they may be at the origin of new grammatical structure. Pregnancy is the opposite in
this respect, too: in order to be viable, it must rely on established linguistic structure and ex-
ploit it to the utmost. Pregnancy will not occupy us further here.
Finally, a methodological remark must be made. In section 2, many examples of pleonas-
tic expressions are adduced without individual analysis. Some of them have both pleonastic
and plain uses. For instance, repeat means ‘do something for the nth time, with n > 1’. There-
fore, the literal meaning of repeat again is ‘do something for the nth time, with n > 2’. If the
expression is used in this sense, it is not pleonastic. Purists will restrict their use of it to this
sense. However, it suffices to observe actual speech (or to do an internet search) in order to
become aware that most uses of repeat again actually mean ‘do something for the second
time’, being thus included in the meaning of repeat. Consequently, while it may be observed
that several of the examples below are not necessarily pleonastic, this does not invalidate the
point that they do have pleonastic uses; and that is all that is necessary for the argument.
2.2. Structural types of pleonastic constructions
Since pleonasm is a purely semantic (or stylistic/rhetorical) concept, it implies very little
about the structure of pleonastic expressions. These are therefore structurally quite heteroge-
neous. At least the following criteria are useful in their classification.
2.2.1. Grammatical level of the pleonastic construction
The principal distinction here is between
• a syntactic construction, as in resulting effect,
• and a word (form), as in German bestmöglichst ‘best (most) possible’.
Of course, different syntactic levels may be distinguished, if necessary; for instance, see with
one’s eyes is a verbal, resulting effect is a nominal. There could, in principle, be pleonastic
sentences, too, like this whale is a mammal; but they probably occur chiefly as examples of
analytic sentences in logic books.
2.2.2. Nature of the elements expressing the focal component
The criterion of section 2.2.1 may be applied again to the focal component of a pleonastic
expression itself. Stepping down the hierarchy of grammatical levels, it may be expressed by
• a syntagma, as in fly through the air,
• a word (form), as in return back,
• a stem, as in German Eichbaum ‘oak tree’,
• a derivational morpheme, as in German Reformierung ‘reform’,
• an inflectional morpheme, as in spaghettis.
In the prototypical pleonastic construction, the focal component is expressed twice, once by a
dedicated unit (underlined in the above examples) whose meaning is exhausted by the focal
component, once as part of the meaning of another unit. It is, however, not excluded that the
focal component is represented by a dedicated unit more than once. For instance, in OE chil-
dren, plural is expressed by each of the suffixes –(e)r and -en. Consequently, this parameter
may be applied separately to each of the occurrences of the focal component, leading by itself
Christian Lehmann, Pleonasm and hypercharacterization 6
to a cross-classification of pleonastic constructions. At the morphological level, naturally tau-
tologies of derivational and of inflectional morphemes are of special interest.
2.2.3. Relation between elements containing the focal component
Given that the dedicated unit and the unit including the focal component are members of a
construction, they are in some structural relation. At the higher levels of grammar, this will be
one of the generic syntactic relations of
• sociation, as in German mit Fug und Recht ‘with full right’,
• government, as in dream a dream,
• modification, as in return back.
In a sociative pleonastic construction, the two related elements are generally synonymous.
The construction is then a tautology, as in E1. In a governing construction, the dependent is
by definition selected by the head. The meaning of the latter then includes a selection restric-
tion that embodies a hyperonym of the dependent. For instance, the meaning of mow includes
as a selection restriction a component that represents (an area covered by) a uniform collec-
tion of plants of a certain shape, which is a hyperonym to such nouns as lawn. This is also true
for such cognate object constructions as dream a horrible dream, sleep a restful sleep. Gov-
erning constructions, cognate or otherwise, are generally not regarded as pleonastic,
because the verbal selection restriction is unavoidable. There is, however, an extreme variant
of the cognate object construction where the object is not further specified: In Korean, verbal
concepts such as ‘sleep’ and ‘dream’ are obligatorily rendered by a cognate object construc-
tion of the form ‘sleep a sleep’, ‘dream a dream’ (S.-R. Ryu p.c.). As far as the semantic rela-
tion between the two units is concerned, it seems to be a matter of definition whether it should
be regarded as pleonastic or tautological.
The core of pleonastic constructions is constituted by modificative constructions. Typical
examples have one of the following syntactic structures:
• a nominal consisting of a head noun containing, and an adjective attribute expressing, the
focal component, as in original source, free gift; German die wesentlichen Essentials ‘the
• a verbal consisting of a verb containing, and an adverbial expressing, the focal compo-
nent, as in fly through the air, return back;
• an adjectival consisting of an adjective containing, and an adverbial expressing, the focal
component, as in potentially capable, more than unique.
These constructions have a modifier in common that is syntactically optional and semantically
redundant. It is, however, the modifier, not the head, that codes the focal component more
The preceding classification is restricted to the syntactic level, i.e. it is a subclassification
of the first class of section 2.2.1. For present purposes, it does not seem necessary to take up
the issue of grammatical relations at the word level; the categorial distinctions introduced in
section 2.2.2 will suffice.
However, Lausberg (1990:328) cites a Roman author who does subsume this construction (in Latin) under
Most of the English examples in section 2 are from www.wordexplorations.com/pleonasm.html as of
10.02.2004. Some of them may already be found, in their Latin or Greek version, in ancient treatments of rheto-
Christian Lehmann, Pleonasm and hypercharacterization 7
This discussion amounts to a recognition that in the prototypical pleonastic construction,
the dedicated unit modifies the unit that properly includes the focal component. We will treat
this as an empirical generalization over many examples from English, German and a couple of
other European languages, based on the semantic definition of pleonasm given in section 2.1.
Although pleonastic constructions are typically modificative, it is probably wise not to elevate
this to the status of a definitory criterion, because then the concept would consist of purely
semantic and purely structural criteria which seem to be essentially independent.
Intuitively, hypercharacterization is pleonasm at the level of grammar. We can now re-
fer this to the structural distinctions introduced in section 2.2.2. Hypercharacterization may
then be defined as that kind of pleonasm where the focal component is expressed by an inflec-
tional or derivational morpheme. This is taken as criterial no matter whether this morpheme
can be identified as the surplus element in the construction and whether the other occurrence
of the focal component in the construction takes the form of a dedicated unit, too.
The methodological upshot of section 2.2 is, then:
• The classification of 2.2.1 is a prerequisite for the classification of section 2.2.3.
• The classification of 2.2.2 is presupposed for the delimitation of hypercharacterization.
• The classification of section 2.2.3 yields an empirical generalization over pleonastic con-
As a consequence of this, hypercharacterized constructions will be just as heterogeneous
structurally as pleonastic constructions in general. This should be kept in mind for section 3.
2.3. Asymmetry in pleonasm
Given a tautological expression E
each make an equal contribution to
the overall tautological character of the expression. If E
are linked by a sociative rela-
tion, we can choose any one of them at random, omit the others and still have the same total
meaning. For instance, we can easily reduce useless and unnecessary to either useless or un-
necessary. In this sense, binary tautologies are symmetric.
Now the question arises whether non-tautological pleonastic expressions are semantically
symmetric in the sense that the focal component can be omitted either in the head or in the
modifier, or whether they are asymmetric in the sense of having a legitimate core and a super-
fluous periphery. Since non-tautological pleonastic expressions generally have a dependency
structure, they are structurally asymmetric, so that one can leave out the modifier, but one
cannot simply leave out the head. We will therefore assume that the methodological counter-
part to leaving out the dependent in an expression such as wesentliche Essentials is to replace
the head by a hyperonym that does not contain the focal component, e.g. wesentliche Punkte
‘essential points’. Semantically, then, pleonasm might be symmetric in the sense explained.
An examination of a large set of data – some of which are adduced in section 2.4 – shows
that the procedure of replacing the head by an appropriate hyperonym is not viable in many
cases because there is no such hyperonym. Exactly the same, potentially capable, original
source illustrate this point. On the other hand, omission of the modifier is always possible
both syntactically and semantically. It is also the simpler procedure. I will therefore assume
The modificative nature of pleonastic constructions is, in fact, methodologically ambivalent: It is here treated
as an empirical generalization over a phenomenon whose concept does not entail it. However, as Lausberg
(1990, § 502) shows, already in ancient rhetoric some authors defined pleonasm with respect to modificative
Christian Lehmann, Pleonasm and hypercharacterization 8
that non-tautological pleonastic expressions are asymmetric not only structurally – by virtue
of their dependency structure – but also semantically in the sense that the syntactic head is the
semantic core and the syntactic modifier is the surplus element that renders the expression
pleonastic. In other words, the implication used to operationalize the definition of pleonasm
will be directed in this way: ‘head → dependent’ or else ‘head → head + dependent’. We will
see in section 2.4.6 that this assumption is not entirely unproblematic.
2.4. The motivation for pleonasm
Everywhere in linguistic structure, a movement descending the levels of structure from dis-
course down to the morpheme correlates with a decrease in the freedom of selection and com-
bination of the units of those levels. At the highest level, these operations are motivated by
semantic, stylistic, pragmatic etc. considerations, i.e. by considerations concerning the cogni-
tive and communicative aims the speaker is pursuing. At the lowest level, such motivations no
longer exert any influence, because it is the linguistic system that dictates them. If hyperchar-
acterization differs from other kinds of pleonasm only by the lower level at which it plays, it
is foreseeable that there will be a variety of extra-structural motivations for pleonasm in gen-
eral, which will be relevant only in a diluted and weakened form for hypercharacterization.
Given a predicate that may be true of its argument to different degrees, there may be a default
value of that predicate for that class of arguments, and there may be particular individuals that
the predicate is true of to a higher degree or even to the highest conceivable degree. To ex-
press such a situation, ascription of that predicate to that argument may be intensified. E3 con-
tains some relevant examples.
The expressions chosen for intensification are often based on exaggeration. Probably, some-
body calling a person bärenstark is not committed to a bet that that person could stand a test
against a bear. But the concept of intensification is indifferent to the validity of such literal
interpretations. It suffices that bärenstark is not synonymous with stark, but assigns its argu-
ment a position on the relevant scale that is above the default. This kind of intensification will
be called polar extreme enforcement.
This analysis implies that polar extreme enforcement is not a kind of pleonasm in the
sense defined in section 2.1. It is nevertheless necessary to start our treatment of motivations
for pleonasm with intensification, because the conditions for intensification are often loos-
ened. That is, intensification often treats predicates as gradable that are inherently absolute.
We are coming to this in the next section.
Like most kinds of redundancy, pleonasm is often regarded as bad style. Ancient rhetoric did,
in fact, classify it as a kind of solecism (Lausberg 1990, §502). It is also true that unwitting
Christian Lehmann, Pleonasm and hypercharacterization 9
pleonasm violates the Gricean maxim of quantity and may insofar be irritating. On the other
hand, many pleonastic expressions are evidently no unwitting slips, but are meant to lend em-
phasis to the message. Examples of such emphatic pleonasm are given in E4; moreover, all
of the examples of tautology given in E1 are motivated by emphasis.
The emphatic character of the expressions in E4 is verifiable by a test: in all of them, the
modifier may receive emphatic stress.
Analyzing the examples in E4, we see at once that in most of them the modifier is an in-
tensifier. Emphatic pleonasm may be subdivided as follows:
1. Default confirmation: In E4.a, the head may be interpreted more or less liberally. How-
ever, what the modifier says is the default interpretation of the head, anyway, and insofar
the expression is pleonastic. The intensifier confirms this default interpretation, forestal-
ling a possible moderate interpretation of the head.
2. Insistence on focal component: In E4.b, the meaning of the head is absolute in the sense
that it applies to something in a yes-or-no fashion rather than to some extent. Conse-
quently, the intensifier cannot do more than underline the significance of what the head
Many heads in emphatic pleonasm admit of a less-than-perfect reading and, correspond-
ingly, of an attenuative modifier. Thus, expressions like inexact replica, almost the same, sur-
rounded on almost all sides are unobjectionable. They presuppose the possibility of canceling
the perfect interpretation of the head. In this perspective, intensification has a purely semantic
justification in the cases of section 2.4.1 (E3); it is semantically motivated to some extent in
‘default confirmation’ (E4.a) and only stylistically motivated in ‘insistence on focal compo-
nent’ (E4.b). In other words, the three varieties of intensive and emphatic pleonasm seen so
far may be ordered on a scale from purely semantic to stylistic motivation as in S2:
S2. Motivation of intensive and emphatic pleonasm
polar extreme enforcement > default confirmation > insistence on focal component.
Insistence on the focal component is legitimated by analogy to default confirmation, and de-
fault confirmation is legitimated by analogy to polar extreme enforcement. Emphatic pleo-
nasm sails under the flag of polar extreme enforcement. The latter is just a kind of intensifica-
tion, which, in itself, is not (yet) pleonastic. Thus, S2 symbolizes the emergence of pleonasm;
its central position may be taken to mark the pole of incipient, unobtrusive pleonasm.
Functional sentence perspective is gradual in many ways. One of these is the fact that the dif-
ference between thematic and rhematic material is greater at higher levels of syntactic com-
plexity and shrinks down to the lowest level, viz. the level of the word form. Now if I have a
sentence in which the focal component is to be rhematic, this will not be sufficiently repre-
sented by the word of whose meaning it is but a component. The modifier codes the focal
Christian Lehmann, Pleonasm and hypercharacterization 10
component separately so that it can receive rhematic status in the utterance. This is the typical
motivation for expressions such as those of E5.
Here again, the focal component may, in some cases, be a defeasible implication of the mean-
ing of the head noun. Thus, the concepts of a big baby, of an intermediate source or of a
Danaans’ gift are not self-contradictory. As in the default confirmation variety of emphatic
pleonasm, the modifier here makes explicit a component that is part of the default interpreta-
tion of the head.
On the other hand, non-pleonastic uses of some of the phrases in E5 are possible. One
might construct a text that meaningfully opposes potentially capable to actually capable. The
point here is that these phrases are generally used in a pleonastic fashion where potentially
capable is not opposed to actually capable, but just means capable.
The set of examples in E6 shows another motivation for pleonasm:
If you are not sure whether the head actually possesses the focal component, you play it safe
by expressing the component separately in a modifier. We will call this safety pleonasm.
There are several fields in which safety pleonasm appears to be commonly operative. An es-
pecially important one is loanwords, as in T2.
T2. Safety pleonasm in loan words
language expression comment
English Rio Grande river Spanish rio ‘river’
English Sahara desert Arabic sahara ‘desert’
English Mount Fujiyama Japanese yama ‘mountain’
Italian Mongibello Sicilian mon = Arabic gebel ‘mountain’
German die La-Ola-Welle Spanish la ola = German die Welle ‘the wave’
In a speech community, there is variation with respect to command of the donor language of
loans. Those that borrow an expression may be assumed to have some knowledge of the
meaning and even structure of the loan. To other speakers of the recipient language, the struc-
ture of such foreign names is either unknown or irrelevant. For these, rio is not another word
for ‘river’, but part of the proper name Rio Grande. To this extent, such formations are not
really pleonastic in the recipient language.
In general, safety pleonasm is a symptom of instability of variation, at the level of the in-
dividual or of the speech community. For some speakers, handwritten manuscript is clearly
pleonastic, while for others it is not, but just means ‘handwritten paper’. Safety pleonasm
A Google search (06.05.2005) for potentially capable yields 40,400 pages, 20 of which oppose it to actually
capable and one to capable.
It follows the (German proverbial) maxim doppelt genäht hält besser ‘double-stitched lasts longer’.
Christian Lehmann, Pleonasm and hypercharacterization 11
therefore indicates that at least part of the speech community does not feel that the base of the
expression is (already) characterized for the focal component.
Yet other examples evince a desire to equip a naked noun, verb or adjective with a companion
so that it need not stand alone. The word alone seems too weak. E7 contains a couple of rele-
In some cases, this horror vacui may be motivated purely phonologically, by reasons of
rhythmic euphony. Observe also that several of the heads are monosyllabic.
may be called phatic pleonasm. The modifier is not stressed and in most cases cannot even
be stressed because there is no possible contrast.
A pleonastic combination may become usual to the extent that it is less marked than its non-
pleonastic counterpart which lacks the modifier. The expansion of repeat / wiederholen to
repeat again / noch einmal wiederholen is almost an automatism. To the extent that there is a
rule that requires that modifier and head agree in the focal component, we have a kind of se-
mantic concord at the syntactic level. That some such mechanism must be operative becomes
more plausible if this rule manifests itself at the morphological level. This may be seen in the
following two sets of examples.
A variant of the pleonastic nominal appears in diminutive expressions of the kind illus-
trated in E8 – E10:
‘Gyricon’s SmartPaper essentially consists in small dichromic mini-balls embedded in
a thin flexible plastic layer. These mini-balls turn ...’
‘With its 1024 x 576 small mini-mirrors, this [chip] is attuned to the PAL norm which
is standard in this country.’
‘small islets situated in front’
Such examples share with the foregoing types the fact that the focal component is expressed
more explicitly by the syntactic modifier than by the head. However, something similar to
Cf. Malkiel 1957f:79, 98f on rhythmic aspects of hypercharacterized Spanish pieses ‘feets’, Löfstedt 1933 on
the idea that a word may be perceived as too short for what it signifies, and Haiman 1985 for theoretical aspects
of quantitative isomorphism.
Christian Lehmann, Pleonasm and hypercharacterization 12
agreement appears to be operative in such a combination, in that once we have chosen the
adjective klein as a modifier, diminution of the head noun is almost an automatism.
therefore call this variant concord pleonasm. In this and the following case, concord pleo-
nasm manifests itself at the morphological level: The focal component is not just a semantic
feature of the lexical meaning of the noun, but expressed separately by the diminutive mor-
pheme. Because of this, either the syntactic or the morphological modifier are freely omis-
sible, with little difference in meaning.
The same phenomenon may be illustrated with female sex marking in German. E11 illus-
trates the range of phenomena relevant here.
‘In 1884, Sofja Kovalevska in Stockholm becomes the first female she-professor
‘Luise Welskopf-Henrich ... first female professor at the Alma Mater of Berlin in
In E11.a, the female sex of the referent is expressed twice, by the adjective attribute and by
the female derivational suffix. In E11.b, it is only expressed by the attribute. Moreover, the
NP has feminine gender in E11.a and masculine gender in E11.b. A search on the web
(Google, 12.02.2004) turns up 57 examples of weibliche Professorin and 4 examples of weib-
The motivation of this kind of pleonasm is intricate. In a diachronic perspective, one can
be sure that the numerical ratio would have been inverse if this web search had been executed
50 years ago. At that time, weibliche Professorin would either have seemed unnecessarily
redundant or else it would have meant ‘feminine [i.e. womanly] professor’. In contemporary
German, female human beings are mostly designated by nouns of feminine gender and, if
possible, derived with the female suffix. In an NP containing the adjective weiblich as a modi-
fier, this rule is almost obligatory, as the numerical ratio shows.
Thus, the use of the adjec-
tive weiblich in sentences like E11 is contrastive, while the use of the female suffix -in is due
to concord pleonasm.
In the varieties of pleonasm analyzed before, the focal component is expressed separately
by the modifier, but is just a semantic component of the head. Pleonasm in such constructions
is thus a purely semantic, not a morphological phenomenon. In morphological concord pleo-
nasm, the focal component receives separate expression by a bound morpheme on the head,
How strong this automatism is may depend on the particular language and a variety of other factors. A text
count of combinations of malen’kij ‘small’ with diminutive nouns in Russian reported on in Rusakova 2004
finds the following numbers of tokens: a) no such adjective + diminutive noun: 200; b) malen’kij + non-
diminutive noun: 58; c) malen’kij + diminutive noun: 14. In that corpus, the concord tendency is thus relatively
weak. Moreover, in contrast with E8 – E10, there are non-pleonastic combinations of ‘small + diminutive’, viz.
whenever the diminutive does not mean ‘small’.
Another example, just to show that professors enjoy no preferential treatment: daß er eine Beziehung zu einer
weiblichen Managerin des Konzerns unterhielt ‘that he entertained a relationship with a female she-manager of
the company’ (Der Standard 06.03.05, p. 3).
Christian Lehmann, Pleonasm and hypercharacterization 13
fulfilling thus the condition for hypercharacterization. In other words, the focal component is
expressed twice separately, so that its two occurrences are near-synonymous. Given that one
of them is a word, while the other is a derivational morpheme, they can hardly be totally syn-
onymous. However, morphological concord pleonasm as illustrated here constitutes a transi-
tion from a purely semantic to a morphological phenomenon.
Both the syntactic modifier and the bound morpheme are optional, but to the extent that
the focal component is more explicitly coded at the syntactic level, its morphological coding
is perceived as a contingent phenomenon. From a processing perspective, the asymmetry pos-
tulated for pleonastic constructions in section 2.3 may be reversed here: In the constructions
of sections 2.4.2 – 2.4.5, it appears that the speaker first selects the head and then expands it
into a pleonastic construction, succumbing to one of the motivations discussed there. Con-
trariwise in morphological concord pleonasm, it appears that the speaker first selects the syn-
tactic modifier together with an unmarked version of the head and only then pleonastically
marks the latter for the focal component. The morphological marking of a feature of one word
on another member of its construction is like agreement. However, concord pleonasm differs
from agreement not only in being largely optional, but also in its direction: inside the noun
phrase, agreement works from the head towards the modifier,
while in the noun phrases of
the present section, concord works from the modifier towards the head.
2.4.7. Summary of motivations
The kinds of motivation for pleonasm that we have distinguished differ in their strength and
may accordingly be arranged on a scale, as in S3 (which embodies S2 at its start):
S3. Strength of motivation of pleonasm
(intensive >) emphatic > rhematic > safety > phatic > concord pleonasm
There are several criteria for identifying the motivation of a given pleonastic construction:
• Entailment: At the end of S3, the meaning of the modifier is entailed by the meaning of
the head. At the start of the scale, the former merely pins down a possibility provided by
• Usualness: At the start of S3, pleonasm is marked; at the end, it becomes the normal way
• Contrast: Contrastive stress on the modifier is normal at the start of S3 and then recedes
down the scale until it becomes outright impossible at the end.
Although the phenomena analyzed so far abide at the lexical-syntactic and derivational
levels, it may be seen that these three criteria are reminiscent of the criteria that define gram-
maticalization (cf. Lehmann 2002, ch. 4). That is, increasing entailment is an early stage of
desemanticization, usualness is a milder form of obligatoriness, and loss of the ability to con-
trast is the prerequisite for cliticization. It is as if S3 were a pre-stage of a grammaticalization
scale. And, of course, a given expression or construction may move down S3 from left to
right. We will come back to this in section 3.1.1.
As we have seen in section 2.4.1, there is a kind of intensification which we called polar
extreme enforcement, illustrated by expressions like boiling hot, which insist that the extreme
pole of a scale is being referred to and which are not yet pleonastic. At the opposite pole,
pleonasm becomes similar to syntactic agreement. At the beginning of S3, semantic and
With some simplification; cf. Lehmann 1982 for more accuracy.
Christian Lehmann, Pleonasm and hypercharacterization 14
pragmatic considerations condition the choices; at the end, usage and grammar start to dictate
them. The scalar representation of pleonasm as in S3 makes us see how it ties in with related
The classification of the five kinds of pleonasm is, of course, not clear-cut. Little baby, for
instance, need not be rhematic in every case, often it will be merely a phatic pleonasm. Again,
the various motivations do not exclude each other. Fall down and repeat again, which I clas-
sified as phatic pleonasms, are almost obligatory collocations, a feature that they share with
Finally, many pleonastic expressions share with the tautologies reviewed in E1 the prop-
erty of being phraseologisms. In such cases, pleonasm is not a collateral result of a constella-
tion at the syntactic or the discourse level, but something built into the lexicon of a language.
As anticipated in section 2.1, one can conceive of repetition as a particular kind of tautology
where the synonymous elements are identical. And repetition does have some of the functions
of pleonastic expressions that we saw in section 2.4. It may have the same function as tautol-
ogy – insistence on the focal component –, as illustrated by the German adverbials in E12.
E12.a is synonymous with E12.b (although their use is slightly different). E12.a is an example
of tautology as the phrases of E1. E12.b differs from those cases only by the formal identity of
the synonymous items.
Repetition may have a purely phatic function, as in E13.
Thus, the scale of S3 may be applied to repetition as to non-identical redundancy. As has been
indicated above and as will be argued further in section 3.1.1, pleonasm has a grammatical-
ized manifestation, which is hypercharacterization. In the same way, reduplication may be
seen as grammaticalized repetition (cf. Marantz & Wiltshire 2000:558). We will therefore
consider repetition as a limiting case of the redundancy phenomena analyzed here. A couple
of examples involving repetition will come up; however, reduplication and iteration will not
be treated per se.
A peculiar kind of repetition may be seen in abbreviation elaboration, which is, at the
same time, a kind of hyponym compounding and therefore treated in section 3.4.
In hypercharacterization, the focal component is expressed by an inflectional or derivational
morpheme (cf. section 2.2.2). It should be born in mind that the concept of hypercharacteriza-
tion imposes no conditions on the expression of the second occurrence of the focal component
constitutive of any pleonasm. Thus, English more easier is hypercharacterized by the adverb
more combining with a morphological comparative form; but so is German (der) einzigste
‘the most only’, where the superlative suffix repeats the idea of singling out one individual
Christian Lehmann, Pleonasm and hypercharacterization 15
fulfilling a relevant condition, which is also part of the concept of einzig. We already saw
some relevant cases of hypercharacterization in morphological concord pleonasm (section
2.4.6). Other typical examples include, in the domain of inflection, the English children and
brethren adduced in T1, and in the domain of word-formation, the German examples given in
T3. Hypercharacterization in German word-formation
hypercharacterized basic meaning
aufoktroyieren oktroyieren impose, force upon
retract, play down
Given that hypercharacterization is a kind of pleonasm, it may be motivated in the same ways
seen before. The German preverbs are added to their bases in order to make explicit a mean-
ing component commonly expressed by these preverbs, as in the near synonyms aufzwingen
and wegschaffen, respectively. Analogy is clearly at work here. Since the bases are French
loans of whose meaning one cannot be entirely sure, the motivation of these formations com-
bines rhematicity with safety.
At the grammatical level, pleonasm concerns linguistic theory in a much more vital way.
All of the expressions analyzed in section 2 are syntactically and semantically well-formed, so
that they do not constitute a problem for either syntax or formal semantics.
may thus safely be relegated to stylistics. At the level of morphology, however, we deal with
specific operators combining specific structural features with their operands in a rule-
governed way. Now how can OE brether and childer take a plural affix if they are already
marked for plural? Any theory of grammar that constructs complex forms in a compositional
fashion by combining an operand of a certain category with an operator that transforms it into
a resulting expression of another category has a serious problem here.
We shall come back
to these problems in section 4 and first review a couple of examples of hypercharacterization
in order to familiarize ourselves with the phenomenon.
3.1. Hypercharacterization in syntax
In doing this, we can take up where we left off in section 2.4.6, viz. at the level of syntactic
3.1.1. Personal agreement
In Latin just as in the written norm of several Romance languages, the personal ending con-
tained in the finite verb form is sufficient reference to the subject; thus neither grammar nor
semantics require an overt subject. In several spoken varieties, and in French even in the writ-
ten standard, the subject pronoun is obligatory. T4 visualizes the situation in two Romance
languages in a simplified way.
There may be exceptions to this. Taken literally, a predicate like more than unique is self-contradictory. How-
ever, a literal interpretation is out of place, because then more would have to be the syntactic head of the phrase,
while in fact more than is a modifier to unique.
Ortmann 1999 quotes a number of morphological theories – and defends one himself – that exclude hyper-
characterization because of its redundancy. Such theories declare the non-existence of facts like those adduced
here and in the literature.
Christian Lehmann, Pleonasm and hypercharacterization 16
T4. Pronominal subjects in Romance ‘we live’
without with pronominal
example value example value
viviamo normal noi viviamo emphatic
vivons ungrammatical nous vivons normal
In the right-hand column of T4, we have hypercharacterization of the subject reference. In
Italian, its motivation is emphasis, while in French, it is sheer concord. As is well-known, the
French construction evolves by grammaticalization of a Proto-Romance construction that is
reflected in Italian. This shows that once we concentrate on grammatical pleonasm, the scale
S3 becomes a manifestation of a general grammaticalization scale.
At the start of the development, the verbal clause is expanded by an emphatic subject
pronoun. The Italian line of T4, read from left to right, illustrates this process. This kind of
pleonasm comes under the concept of reinforcement (cf. Lehmann 2002, ch. 2.5). At this
stage, the subject pronoun is clearly the surplus element of the pleonastic construction. In the
further course, the reinforcement of the subject reference no longer works at the communica-
tive, but merely at the syntactic level. I.e. the pronoun is needed to ensure the subject refer-
ence in the first place. This shows that the concept of pro-drop occasionally used to describe
the Italian situation is misconceived with regard both to the diachrony and to the function of
the construction. It is only from an anglocentric perspective that Italian drops some element
that should be there. On the contrary Italian optionally and French obligatorily add a subject
As concord hypercharacterization is grammaticalized to mere agreement, redundancy
seems to be introduced into the grammar. However, in this course it looses its pleonastic func-
tion at the communicative level and gets a new function at the structural level, in the marking
of syntactic relations.
3.1.2. Spatial relations
Another area where hypercharacterization is very frequent in the languages of the world is
spatial deixis and spatial relations. E14 illustrates four variants of a sentence containing the
deictic da ‘there’ in Standard German (a), Northern Colloquial German (b), Bavarian (c) and
Alemannic German (d).
‘I know nothing of it.’
All of these variants are attempts to solve the problem of the topicalization of the pronominal
complement of the preposition. Both Standard German pied piping and Northern Colloquial
German preposition stranding solve it without redundancy. Southern dialects avoid preposi-
tion stranding while feeling that mere pied piping is communicatively insufficient and the
topical pronoun must be present separately.
Christian Lehmann, Pleonasm and hypercharacterization 17
E15 illustrates three different uses of the spatial relator ex ‘out of’, all of which are from
Classical Latin. In E15.a, the relator appears only as a preposition; in b, it appears only as a
preverb; in c, it is used pleonastically both as a preposition and a preverb.
out.of town:ABL.SG flee:INF
‘to flee out of town’
German constructions of the kind illustrated in E16 appear to be structurally similar to E15.c.
‘to adjoin to something’
‘to watch over somebody’
They differ from the Latin construction in several respects. One that we can forgo relates to
the fact that the German compound verb is separable. What is of more importance is that the
compound verb governs the preposition of its complement.
This pleonasm is therefore com-
pletely grammaticalized or lexicalized.
Both redundant demonstrative topicalization and preverb-preposition concord are cases of
concord hypercharacterization; but they are special in that they involve repetition of the same
element. Since hypercharacterization plays at the level of grammar, the choice of synonymous
morphs decreases, so that pleonasm often takes the form of identical repetition [sic!].
3.1.3. Other cases
In German, subjunctive II, which like a Romance conditional marks unreal propositions, is
obligatory after certain modal adverbs such as beinahe ‘almost’, as in E17.
‘I almost fell’
The adverb is syntactically optional, but if it is omitted, the meaning changes. The irreal sub-
junctive here is redundant, because the adverb by itself says that the situation was not real-
ized. Other languages, e.g. English and Latin, have the indicative in such sentences. Since the
subjunctive here is predictable, we deal with a case of concord pleonasm.
Another kind of construction, known from Latin, involves what has been called the ‘pleo-
nastic reflexive’ for a long time. Since it is analyzed at length in Cennamo 1999, an example
may suffice here:
what(ACC) then RFL:DAT wants father(NOM.SG)
‘What then does my father want?’ (Ter. Andr. 375 ap. Cennamo 1999:117)
It could, in principle, govern any preposition. For instance, in mit x aufhören ‘to stop doing x’, the verb parti-
cle and the preposition governed are distinct.
Christian Lehmann, Pleonasm and hypercharacterization 18
The reflexive pronoun in E18 is omissible with no change in meaning. It is pleonastic insofar
as it underlines the subjective component inherent in the notion of volition. It also combines
with other verbs of inactive meaning, focusing on the fact that the process abides in the sphere
of the subject. In Vulgar and Late Latin, the construction loses its marked character and
evinces some symptoms of grammaticalization.
3.2. Hypercharacterization in inflection
Hypercharacterization in inflection has been a topic in linguistics at least since Paul
1920:162f, where it is treated as ‘pleonasm of formative elements’. This term allows for the
possibility that an inflectional category may be hypercharacterized by different morphological
processes. Since the phenomenon is well documented (see also Haspelmath 1993, section 5f
and Dressler 2004), we can limit ourselves here to a couple of examples.
In Middle High German, the suppletive comparative of the adjective guot ‘good’ was bass
‘better’, as in E19 (from ~1200):
MHG ‘von Veldeke, the wise man! He could have praised her better.’
Secondarily, the form gets the comparative suffix -er, which triggers metaphony, so that the
modern form besser results.
Redundant comparative and superlative marking is common in Indo-European languages.
In colloquial English, we find more easier, in French and Spanish, we find E20f:
‘the best country in the world’
‘KEV ... the most handsome, the best one!’
The examples from the three languages have it in common that the surplus element is an ana-
lytic marker attached to a synthetic form of grading. They differ in that the synthetic compara-
tive has a morpheme of its own in the English example, while E20f evince a suppletive super-
lative. A pleonastic superlative is, of course, motivated by emphasis. In addition, it may be
relevant that the pleonastic comparative and superlative in the Romance languages is re-
stricted to adjectives with suppletive grading. Insofar, it is safety pleonasm. On the other
hand, no emphasis and no safety is discernible in more easier; this is just phatic pleonasm.
The examples of E22 are similar both functionally and structurally:
‘the most only / extreme / optimal’
I will assume without further discussion that comparison is an inflectional category in the languages at hand.
Christian Lehmann, Pleonasm and hypercharacterization 19
The underlying Latin forms extremus ‘outermost’ and optimus ‘best’ have the position of su-
perlatives in their paradigm and are even marked as such by an – admittedly irregular – super-
lative allomorph. Naturally, this does not matter for German grammar.
Here, the examples
in E22.a are on a par: The focal component – the function of the superlative suffix – is some-
thing like ‘the relevant domain (identified by the argument of the adjective) is restricted to
that subset (or individual) that occupies the positive pole of the scale designated by its host
(the adjective stem)’. It is represented by a dedicated inflectional or derivational morpheme
(which assigns these cases to hypercharacterization), but otherwise just entailed by the mean-
ing of the latter’s host. The application of the superlative suffix to kein (E22.b) works simi-
larly insofar as it pretends kein to mean something like ‘occupying the positive pole on a scale
A related phenomenon occurs in German adjectival compounds whose determinans is a
superlative form of some adjective and whose determinatum is another adjective or participle,
as those in E23.a.
‘best possible / shortest possible / most sold’
‘the best possible result / the shortest possible way / the best selling book’
Hypercharacterized forms as those in E23.b are very frequent. In this case, we clearly have
concord pleonasm. In addition, pleonasm is here motivated by the principle that inflection
should be at the word margin.
The Old English forms children and brethren illustrate hypercharacterized nominal plural.
Other examples of this kind are Dutch kinderen ‘children’ and German Jungens ‘boys’, all
with two different allomorphs of the plural morpheme. Pleonastic plural marking is particu-
larly common in loans. Thus the Italian plural form spaghetti ends up as spaghettis, with a
plural -s, in English, Spanish and optionally in German. In contemporary German, the plural -
s is sufficiently productive to yield such hypercharacterized forms as Praktikas = Praktika
‘practical courses’, Visas = Visa ‘visas’, Lexikas = Lexika ‘lexicons’.
All of these examples clearly involve analogical transfer of a marker from a context in
which it is the only operator to fulfill the function in question to a context where it pleonasti-
cally duplicates an operator already applied. We may generalize that hypercharacterization in
morphology is based on analogy.
Moreover, in a diachronic perspective, the two concurrent
markers are not on the same level. There is an inner marker which for some reason does not
quite do the job, and an outer marker which is currently productive and which speakers feel
should appropriately appear on such a word form (cf. Dressler & Dziubalska-Kolaczyk 2001,
It may be comforting for German speakers that already the Romans did not shun extremissimus.
See Haspelmath 1993, section 2.4 for discussion of cases of this kind and Dressler 2004 for the sequence of
hypercharacterized and hypercharacterizing affix.
Cf. Booij 2005[g]:259 for similar examples in Dutch.
In view of Haspelmath’s (1993, section 5.2) objections against an analogical account, it should be stressed that
an analogical model need not be perfect in motivating each and every feature of the transformed item; it suffices
that it share some features with the latter.
Christian Lehmann, Pleonasm and hypercharacterization 20
section 5, Dressler 2004). A more precise formulation of the analogical account might there-
fore say that hypercharacterization is a kind of adaptation of a stem or word form based on
paradigmatic pressure (Koefoed & Marle 2004:1581, Booij 2005[P]).
A special case of inflectional hypercharacterization may be seen in word-internal
agreement. E24 provides an example from Lithuanian.
white-NOM.PL.F-DEF-NOM.PL.F night-NOM.PL.F I:ACC seem-PRS.3
like mysterious-NOM.PL.M witchcraft-NOM.PL.M
‘the white nights seem like mysterious witchcraft to me’ (Stolz 2004:17)
Synchronically, the Lithuanian definite adjective consists of the adjective stem, inflected for
case, number and gender, and a suffixal definiteness morpheme that is again marked for the
same categories, often with the same declensional allomorph (Stolz 2004). Such cases arise by
grammaticalization, where an erstwhile syntagma consisting of two words showing syntactic
agreement is univerbated. At the level of syntax, agreement, although pleonastic, fulfills a
function in marking syntactic relations (cf. section 3.1.1). At the level of morphology, it loses
any kind of motivation.
3.3. Hypercharacterization in derivation
3.3.1. German action nouns
Consider the derivational relationship between noun and verb. Since we have both deverbal
nouns and denominal verbs, this relationship is not per se directional. From the root of the
German verb konzipieren ‘conceive’, we form the action noun Konzeption ‘act of conceiving’,
and on the basis of the noun Analyse we form the verb analysieren ‘make an analysis’. In both
cases, an iconic interpretation of the derivational process would make one believe that the
derived stem is semantically more complex than the base; but since the two processes are mir-
ror images of each other, this would lead into a contradiction. We have to conclude that a
stem does not, in general, become semantically more complex by mere derivational transferral
into a different category. As a matter of fact, we simply get the same concept in two different
Deverbal nouns in -ion (with its allomorphs) such as condition and relation have been
polyfunctional since Latin times. They are primarily action nouns (nomina actionis), as rela-
tion originally signifies the act of referring. Secondarily, they are act nouns (nomina acti), as
relation signifies the result of referring something to something else. Moreover, such a verbal
noun from a transitive base may develop a nomen patientis reading, as in derivation (= deri-
vatum) ‘derived word’, which shares its non-dynamic character with the nomen acti. Once the
noun has acquired the secondary meaning, it may seem too weak to serve as an action noun; it
may seem to lack in “dynamic force”. A clear example is the English noun position, which no
longer signifies the act of putting, but only its result. The act must now be expressed by posi-
tioning, which itself is on the way of loosing its dynamic character.
The semantic passage of nomina actionis into nomina acti and nomina patientis and the
corresponding functional shift in the derivational morpheme forming such deverbal nouns is
probably widespread. The German derivational suffix -ung is subject to the same process.
Thus Glättung ‘smoothing’ is a nomen actionis, Bewerbung ‘(job) application’ is a nomen
Christian Lehmann, Pleonasm and hypercharacterization 21
actionis and acti, Spannung ‘tension, voltage’ is only a nomen acti, Packung ‘package’ only a
nomen patientis. There seems to be a drift towards stativization and reification.
Sometimes the speaker wants to make sure that an action is being designated. His prob-
lem is then to signal that whatever nominalizing process is applied to the verbal base is not
subject to the semantic shift just observed. Abstract nouns that are not overtly derived, like
those of the first column of T5, are the first to become suspect of stativity. To ensure their
dynamic character, they are first verbalized by the suffix -ier- (second column), which forms
verb stems chiefly from non-German bases. In a second step, these verb stems are nominal-
ized by -ung, which, one hopes, conserves the action meaning (third column).
T5. Action noun renewal in German
action noun meaning
Reform reformieren Reformierung reform(ation)
Typologie typologisieren Typologisierung typologization
Metapher metaphorisieren Metaphorisierung (application of) metaphor
Hypostase hypostasieren Hypostasierung hypostasis
Although the nouns of the first column do conserve a nomen actionis reading besides the fre-
quent nomen acti use, they are now mostly replaced – chiefly in the former function – by the
nouns of the third column.
Nouns derived in -tion are not exempt from this remodeling. There are two variants of
applying to them the combination of operations observed in T5. The first is illustrated by T6.
Here, the base underlying the derivation in -tion is derived, by means of -ier-, into a verb. The
latter is then nominalized by -ung. Thus we find, instead of the age-old action/act nouns in the
left column of T6, alternate action nouns newly derived in -ierung, as in the middle column.
T6. Alternate action noun derivation in German
Latinate germanized action reading
Integration Integrierung integrating
Qualifikation Qualifizierung qualifying (oneself)
Klassifikation Klassifizierung classifying
Konversion Konvertierung converting
Konzeption Konzipierung conceiving/planning
Revision Revidierung revising
Tradition Tradierung transmitting
The second solution to the expression problem – this one involving hypercharacterization – is
to derive a verb from the act noun itself and nominalize this again. For instance, Konzeption
‘conception’, both an action and an act noun, can be verbalized by the suffix -ier-, yielding
konzeptionieren, and this can be nominalized again by the suffix -ung, yielding Konzep-
This is visualized in S4, together with the parallel Revisionierung ‘revision’.
Some of the nouns in the left-hand column have actually lost their action meaning. Thus: Unser aller Pflicht
ist die Tradierung /
Tradition von Werten. ‘Everybody among us has as his duty the tradition of values.’
Konzeptionierung is absolutely fashionable; a Google search (02.11.04) turns up 29,200 examples. One can
also hear Konzeptionalisierung (Google: 2,280 examples). Many of the examples of both nouns exhibit a stative
Christian Lehmann, Pleonasm and hypercharacterization 22
S4. Recursive nominal and verbal derivation in German
revid(ieren) Revision revisionier(en) Revisionierung
Konzeptionieren is the same as konzipieren, and Konzeptionierung is the same as Konzeption
(or Konzipierung, for that matter). As the examples show, the processes of action noun deriva-
tion and denominal verb derivation may be applied recursively, either one undoing the result
of the application of the other. Hypercharacterization here requires the execution of two deri-
vational operations in tandem, since if I am not content with Konzeption, I must first verbalize
and then nominalize it again in order to arrive at Konzeptionierung.
The motivation behind this trend is not easy to pin down. -ung by itself displays the same
polysemy as -tion, which makes it hard to believe that speakers trust in its dynamicity. One
might hypothesize that it is the component -ier in -ierung which guarantees the action noun
reading because -ier reflects the underlying verbality. (For speakers’ motivations, it would not
matter that no theoretically sound argument could be made along these lines, since verbs de-
rived in -ier are also at the basis of categorically stative deverbal nouns, e.g. nomina agentis in
-ierer like Kopierer ‘copying machine’.) We would then be faced with a new suffix -ierung,
exclusively dedicated to the formation of nomina actionis. Some documented cases do pre-
suppose that if there is a pair of nouns one of which is derived in -tion and the other in
-ierung, then the second is dynamic. Clear witness of such reasoning is a publication title such
‘conceptualization of motivation and motivating in the context of situated learning’
Here Motivation refers to the pupils’ disposition, while Motivierung refers to the teachers’
action. Klassifikation vs. Klassifizierung is a stock example of the distinction intended here.
However, a web search turns up a host of examples like Deutsche Hotelklassifizierung ‘Ger-
man hotel classification’,
which clearly refer to the result of the action. Equally in E26f, the
nouns derived in -ierung are clearly nomina acti.
‘The spatio-temporal area concerns positioning of the referent in space and mainte-
nance of the latter [i.e. the position] in the course of time.’
‘Ostasien’ or ‘East Asia’ – a German conceptualization’
Stark, Robin & Mandl, Heinz (1998), Konzeptualisierung von Motivation und Motivierung im Kontext situier-
ten Lernens (Forschungsbericht Nr. 091). München: Ludwig-Maximilians-Universität, Lehrstuhl für Empirische
Pädagogik und Pädagogische Psychologie, Internet.
It is adduced, e.g., in Fleischer 1971:156f and Knobloch 2002:336. In other cases, an investigation into the
differential function of rival forms of derivation does yield positive results; cf., e.g. Kaunisto 1999 on the Eng-
lish suffixes -ic and -ical.
the main title of the website http://www.hotelsterne.de/
Christian Lehmann, Pleonasm and hypercharacterization 23
An alternative hypothesis is that – apart from a couple of specific cases like E25 – no seman-
tic issue is involved here, and what matters instead is only the replacement or reinforcement
of an unproductive derivation mechanism (-tion) by a productive one (-ierung). It is true that
many formations in -ierung are just now replacing older formations in -tion. Examples in-
clude Demonstrierung ‘demonstration’ and Variierung ‘variation’, each found several hun-
dred times on the web (01.11.04), but absent from Duden Wörterbuch, 2001 edition. On the
other hand, there are also recent well-established neologisms in –tion like Animation, Emis-
sion, Präsentation (and even more scientific terms such as Extraposition, Kollokation, Par-
tizipation) which show no tendency to get ousted by counterparts in -ierung.
Thus, every attempt to come up with a specific semantic motivation of the redundant ap-
plication of the derivation in -ierung fails. It seems that the general motivation of phatic pleo-
nasm must suffice: the sheer desire to make words sound more impressive. The hypercharac-
terization resulting from this in cases such as S4 is not specifically intended, but does not
bother most speakers either.
The analysis shows that hypercharacterization in derivational morphology must be seen in
the context of the renewal of inherited derivata by productive means.
This renewal itself is
not hypercharacterization; but sometimes the renewal does not go back to the roots, but sim-
ply works on some available base, which may or may not already be marked for the category
3.3.2. Other cases of derivational hypercharacterization
Derivational processes which come under intensification in the broadest sense, including
diminution, augmentation, iteration etc., are particularly prone to hypercharacterization.
Diminution provides some well-known examples. Sometimes different allomorphs of the di-
minutive marker are stacked, as in Italian Bertinetto ‘little Bertie’, librettino ‘little booklet’.
Sometimes the most productive diminutive suffix can be iterated, as in Spanish chiquitito
‘tiny little’. There is also a derivational counterpart to the pleonastic superlative in such Italian
forms as ultrabellissimo ‘most hyperbeautiful’, typical of the language of publicity. The Latin
intensive-iterative suffix -(i)t- is reapplied in verbs such as dic-t-it-o (say-
dic-t-o ‘say repeatedly’, iac-t-it-o (throw-
) = iac-t-o ‘throw repeatedly’.
However, derivational hypercharacterization occurs in other functional contexts as well.
The German suffix -lich may derive adjectives like freundlich ‘kind’, but also adverbs like
schwerlich ‘hardly’, gröblich ‘in a gross way’, fälschlich ‘wrongly’. In the latter function, the
suffix is barely productive today. Since most adjectives can be used in adverbial function
without morphological change, there are many words derived in -lich that function both as
adjectives and as adverbs, like wissentlich ‘knowing(ly)’. Furthermore, there is a more recent
and productive adverbializing suffix -weise, which has an analogous diachronic origin as Ro-
mance -mente, viz. its basis are circumlocutions such as in freundlicher Weise ‘in a friendly
The derivational suffix German -ität = English -ity forms abstract nouns on the basis of adjectives chiefly of
Greco-Latin origin, as in Publizität ‘publicity’. This is currently one of the most productive means towards this
end. Other suffixes like -ie ‘-y’, as in Monotonie ‘monotony’, are losing ground. For some time, old derivata in
-ie/-y have been replaced by more modern (and longer!) ones in -ity. Thus, Anonymie/anonymy have been all but
ousted by Anonymität/anonymity; Synonymie/synonymy and Homonymie/homonymy still go strong, but Syn-
onymität/synonymity and Homonymität/homonymity are on the advance.
German allows this to a much more limited extent, as in Kinderleinchen ‘children-
’ or Schatzileinchen
Christian Lehmann, Pleonasm and hypercharacterization 24
way’, which get univerbated to derived adverbs like freundlicherweise ‘kindly’. Now this
suffix is also sometimes added to adverbs derived in -lich. Thus we find E28.
‘... a turmaline that was long regarded as a ruby in a wrongly way.’
Besides such hypercharacterizations, there are also constructions like E29 which go back to
the periphrasis, but use an adverb as the attribute to Weise:
‘A member may be excluded by the executive board if he has infringed the associa-
tion’s interests in a seriously way.’
It seems that the authors of E28 and E29 regarded fälschlich as synonymous with falsch, and
gröblich as synonymous with grob; i.e. they did not feel that fälschlich and gröblich are char-
acterized as adverbs. This is then safety pleonasm.
Transitivization of transitive verbs also belongs here. The German applicative prefix be-
generally transitivizes verbs, as in singen ‘sing’ – besingen ‘sing to the honor of’. It applies
redundantly in examples like befüllen = füllen ‘fill’, bejagen = jagen ‘hunt’,
füttern ‘feed’. Such examples appear to be due to phatic pleonasm.
If, however, a loan is provided with a marker specifying its grammatical class irrespective
of the fact that, in the donor language, it already belongs to that class, it is rather a case of
safety pleonasm. Thus Spanish alcanzar ‘reach’ is a transitive verb, but as a loan in Yucatec
Maya, it is provided with the transitivizing suffix -t-, as shown in E30 and a myriad of similar
YM IMPFV-SBJ.3 achieve-TRR-INCMPL
‘he achieves it’
3.4. Hypercharacterization in compounding
In the endocentric nominal compound, the determinans forms a more specific concept on the
basis of the determinatum. There are at least two pleonastic varieties of this compounding
type. The first is illustrated in T7.
T7. German hyponym compounds
expression composition meaning
Sturmwind storm:wind storm
Enkelkind grandchild:child grandchild
Eichbaum oak:tree oak
The typo in E29 is telling: the text is evidently an emendation of an earlier version that contained gröblicher-
Bejagen also means ‘hunt in (a hunting-ground)’ and then is a regular, non-pleonastic applicative derivation.
IMPFV imperfective, SBJ subject, TRR transitivizer, INCMPL incompletive
Christian Lehmann, Pleonasm and hypercharacterization 25
Grammatikalisierungsprozeß grammaticalization:process grammaticalization
Here the determinans is a hyponym of the determinatum; the construction may therefore be
called hyponym compounding.
Although these compounds satisfy the semantic characteri-
zation of pleonasm given in section 2.1, they differ in their structure from the phenomena con-
sidered so far because it is the modifier alone, not the structural head, that is synonymous to
the complex. Unlike all the other cases of pleonasm, it is, thus, the head of the construction
that is redundant. It is known that the strategy of hyponym compounding can give rise,
through grammaticalization, to a system of nominal classification. In this, hyponym com-
pounding is functionally similar to concord pleonasm, which, as we saw, may evolve into
A subvariety of hyponym compounding may be seen in the left-hand column of T8:
T8. Abbreviation elaboration
elaborated abbreviation resolution of abbreviation
ABS system Anti-lock Braking System
HIV virus Human Immunodeficiency Virus
LCD display Liquid-Crystal Display
PIN number Private Identification Number
ABM-Maßnahme Arbeitsbeschaffungsmaßnahme ‘labor provision measure’
This construction may be called abbreviation elaboration. It is very common both in English
and in German.
As in T7, the added noun repeats a semantic component already contained
in the abbreviation. It also has the same kind of motivation as other pleonasms: In abbrevia-
tion elaboration, safety pleonasm concurs with phatic pleonasm. The peculiar feature of ab-
breviation elaboration is that the focal component is identically present in the base. It is not a
matter of synonymy, but of repetition, although the component is not spelt out in the base.
The other main variety of pleonastic compounding is synonym compounding, as in
‘excessive, extra’ (Chao 1968:374f)
While this type does not appear to constitute a productive pattern in German, it has been very
important in Mandarin Chinese, apparently as a form of safety pleonasm to disambiguate ho-
4. Theoretical implications
There is pleonasm at all the grammatical levels from the sentence down to the stem. As in any
movement downwards the grammatical levels, paradigmaticity increases. At the highest level,
Fleischer (1971:93f) speaks of ‘clarifying compounds’. See Bloomer 1996 for a detailed study.
According to anecdotal evidence provided by G. Banti (p.c.), abbreviation elaboration occurs in Italian, too,
although there it has a different structure, e.g. virus HIV.
Christian Lehmann, Pleonasm and hypercharacterization 26
pleonasm starts out as a strategy that lends emphasis to an expression. At the lowest level, it is
a strategy that fits an expression in a paradigm or in a structural class. This has become clear-
est in the two sections (3.2f) devoted to hypercharacterizing inflectional and derivational mor-
phemes. Discourse motivations here give way to system-internal pressure.
Theoretically, the pleonastic comparative of English could involve a repetition of the op-
erator more, just as the pleonastic action noun in German could involve a repetition of the
nominalizer -tion. However, this is not what happens. As we have seen, repetition is only a
limiting case of redundancy. Hypercharacterization is therefore not just a process of copying.
It shares properties with contamination (blending), where the speaker cannot make a choice
between two synonymous expressions.
Hypercharacterization provides some important lessons for linguistic theory. In the fields
of syntax, inflection and word-formation, formal descriptive models describe the formation of
complex units of a certain category by the combination of an operator with an operand of a
certain category. An extreme form of this approach has been known as the ‘unitary base hy-
pothesis’. As we have seen (and as has been shown repeatedly in the literature), this purely
analytic description runs into problems. There is one category in syntactic and morphological
constructions that is of prime importance, and that is the category of the resulting syntagma.
The speaker uses an operator to create an expression with certain grammatical properties. Op-
erators are often sensitive to the properties of the operand. But these are of secondary impor-
tance, and often the speaker simply does not care.
This is clearest in the treatment of loans.
Here one might want to argue that the grammatical properties of the donor language cannot
possibly play a role in the recipient language. However, things are not so simple. Borrowing
an item presupposes some degree of knowledge of the donor language, and the item is bor-
rowed precisely for its properties. The most one can say is that the speaker wants to make
sure, with the means of the code he is currently using, that the item has the properties needed
in the discourse. The most transparent way of guaranteeing this is the application of a produc-
tive operation whose operator confers just the desired property. This speaks in favor of a goal-
directed theory of language and of a holistic approach to grammatical description, to com-
plement the otherwise needed analytic approach.
Grammaticalization has often been described as a transition from universal iconic dis-
course strategies to language-specific system-dependent grammatical rules. As far as that
goes, grammaticalization involves a loss in motivation. As we have seen, pleonastic phenom-
ena can be arranged on a scale that starts from full motivation in terms of intensification and
emphasis and where motivation weakens gradually. As soon as we get to a stage where ele-
ments in a sentence start being in concord with each other, we enter the domain of rules of
syntax. From here on, the phenomenon goes by the name of hypercharacterization. Further
grammaticalization leads to grammatical agreement, first at the syntactic level, finally even
inside a word-form.
Some have proposed a principle of derivational blocking which says that an otherwise
productive derivational process is blocked for a particular base if there is already a derivatum
– formed from this base by another process – that occupies the target position. The facts ad-
duced in section 3.3 falsify a simple general version of such a principle. Sometimes the oppo-
This point is stressed in Dressler et al. 2001. Cf. Haspelmath 1993, section 6.2 for some discussion.
Plag 2005 argues emphatically that categorical properties of the bases of word-formation processes are in
general irrelevant. Earlier proponents of a holistic, semantic rather than structural approach to word-formation
include Plank 1981.
Christian Lehmann, Pleonasm and hypercharacterization 27
site principle seems to be active: concepts that continue to be needed deserve to be expressed
by currently productive means, which may lead to a renewal of their expression. This phe-
nomenon is well-known from grammaticalization research. It suffices to mention a stock ex-
ample like the renewal of various verbal categories of Ancient Indo-European languages in
their modern descendants (see Lehmann 2002, ch. 2.4 for details). The marking of inflectional
categories is thus far from being constrained by a blocking principle. Quite on the contrary, if
the system of grammatical meanings includes a certain category, then that category will be
marked by such structural means which correspond to the type the language is currently fol-
lowing; and at the same time, their marking by means that belonged to a previous type will
fossilize. Research into hypercharacterization may shed new light onto the corresponding is-
sue in derivational morphology. Blocking is counteracted by renewal there, too.
Safety pleonasm evinces a basic insecurity in the control of the code. Since none of us is
the master of the norm, we do not have full certainty of the meaning of a word and the service
it can do in our speech. Therefore we prefer to play it safe and to combine it with another sign
which should also contribute the desired meaning and of which we may feel a little more sure.
Pleonasm and hypercharacterization thus provide evidence of a peculiar kind that lan-
guage is not a stable system. Older textbooks teach that language changes because we have to
adapt it to new needs. Younger textbooks teach that it changes because the language acquisi-
tion device comes up with an original analysis of the input. Pleonasm and hypercharacteriza-
tion confirm what Coseriu (1958, ch. III) said long ago (cf. also Booij 2005[g]): Language
changes because we create it every day. We have to do so to the extent there is no ready-made
language that we could rely upon.
M(iddle) H(igh) G(erman)
Bloomer, Robert K. 1996, „Die pleonastischen Zusammensetzungen der deutschen Gegen-
wartssprache.“ American Journal of Germanic Linguistics and Literatures 8: 69-90.
Booij, Geert 2005, The grammar of words. An introduction to linguistic morphology. Oxford:
Oxford University Press.
Booij, Geert 2005, “Paradigmatic morphology”. Tranel, Bernard & Dal, Georgette (eds.),
Hommage a Danielle Corbin. Amsterdam & Philadelphia: Benjamins; 000-000.
Cennamo, Michela 1999, "Late Latin pleonastic reflexives and the unaccusative hypothesis."
Transactions of the Philological Society 97(1):103-150.
Chao, Yuen Ren 1968, A grammar of spoken Chinese. Berkeley & Los Angeles: University of
Coseriu, Eugenio 1958, Sincronía, diacronía e historia. El problema del cambio lingüístico.
Montevideo: Universidad de la República de Uruguay, Faculdad de Humanidades y
Ciencias (Repr.: Madrid: Gredos, 1973.)
Christian Lehmann, Pleonasm and hypercharacterization 28
Dressler, Wolfgang U. & Dziubalska-Kolaczyk, Katarzyna & Spina, Rossella 2001, “Sources
of markedness in language structures.” Folia Linguistica Historica 22/1-2: 103-35.
Dressler, Wolfgang U. 2004, "Hypercharacterisation and productivity in inflectional mor-
phology." Krisch, Thomas & Lindner, Thomas & Müller, Ulrich (eds.), Analecta homini
universali dicata. Arbeiten zur Indogermanistik, Linguistik, Philologie, Politik, Musik und
Dichtung. Festschrift für Oswald Panagl zum 65. Geburtstag. 2 Bde. Stuttgart: H.-D.
Heinz (Stuttgarter Arbeiten zur Germanistik, 421); 2:515-524.
Fleischer, Wolfgang 1971, Wortbildung der deutschen Gegenwartssprache. Tübingen: M.
Niemeyer (2. unveränd. Aufl.; 1.: Leipzig: VEB Verlag Enzyklopädie, 1969).
Haspelmath, Martin 1993, "The diachronic externalization of inflection." Linguistics
Haiman, John 1985, Natural syntax. Iconicity and erosion. Cambridge: Cambridge University
Jakobson, Roman 1960, "Closing statement: Linguistics and poetics." Sebeok, Thomas A.
(ed.), Style in language. Cambridge, Mass.: MIT Press; New York & London : J. Wiley &
Kaunisto, Mark 1999, “Electric/electrical and classic/classical: Variation between the suf-
fixes -ic and –ical”. English Studies 80:343-370.
Knobloch, Clemens 2002, "Zwischen Satz-Nominalisierung und Nennderivation: -ung-
Nomina im Deutschen." Sprachwissenschaft 27:333-362.
Koefoed, Geert & Marle, Jaap van 2004, "Fundamental concepts [of morphological change]."
Booij, Geert & Lehmann, Christian & Mugdan, Joachim & Skopeteas, Stavros (eds.),
Morphologie. Ein internationales Handbuch zur Flexion und Wortbildung. 2. Halbband.
Berlin: W. de Gruyter (Handbücher der Sprach- und Kommunikationswissenschaft, 17.2);
Lausberg, Heinrich 1990, Handbuch der literarischen Rhetorik. Eine Grundlegung der Litera-
turwissenschaft. Stuttgart: F. Steiner (3
ed. = repr. of the 2
., enl. ed. 1973).
Lehmann, Christian 1982, "Universal and typological aspects of agreement." Seiler, Hansja-
kob & Stachowiak, Franz Josef (eds.), Apprehension. Das sprachliche Erfassen von Ge-
genständen. Teil II: Die Techniken und ihr Zusammenhang in den Einzelsprachen. Tü-
bingen: G. Narr (LUS, 1, II); 201-267.
Lehmann, Christian 2002, Thoughts on grammaticalization. Erfurt: Seminar für Sprachwis-
senschaft der Universität (ASSidUE, 9) (2. revised edition).
Löfstedt, Einar 1933, "Wortform, Wortumfang und Verwandtes." Löfstedt, Einar, Syntactica.
Studien und Beiträge zur historischen Syntax des Lateins. Zweiter Teil: Syntaktisch-
stilistische Gesichtspunkte und Probleme. Lund etc.: Gleerup (Acta Reg. Societatis Hu-
maniorum Litterarum Lundensis, X:2); 35-62.
Malkiel, Yakov 1957f, "Diachronic hypercharacterization in Romance." Archivum linguisti-
cum 9:79-113; 10:1-36.
Marantz, Alec & Wiltshire, Caroline 2000, "Reduplication." Booij, Geert & Lehmann, Chris-
tian & Mugdan, Joachim (eds.), Morphologie. Ein internationales Handbuch zur Flexion
und Wortbildung. 1. Halbband. Berlin & New York: W. de Gruyter (Handbücher zur
Sprach- und Kommunikationswissenschaft, 17.1); 557-567.
Ortmann, Albert 1999, "Affix repetition and non-redundancy in inflectional morphology."
Zeitschrift für Sprachwissenschaft 18:76-120.
Paul, Hermann 1920, Prinzipien der Sprachgeschichte. Tübingen: M. Niemeyer (5
Christian Lehmann, Pleonasm and hypercharacterization 29
Plag, Ingo 2005, „Syntactic category information and the semantics of derivational morpho-
logical rules.” Folia Linguistica 39:000-000.
Plank, Frans 1981, Morphologische (Ir-)Regularitäten. Aspekte der Wortstrukturtheorie. Tü-
bingen: G. Narr (Studien zur deutschen Grammatik, 13).
Rusakova, Marina V. 2004, “Russian diminutive/hypocoristic adjectives: between inflection
and derivation”. 11
International Morphology Meeting, Vienna, 14-17 February 2004.
Sherer, Tim D. 1994, Prosodic phonotactics. Amherst: University of Massachusetts PhD diss.
Stolz, Thomas 2004, “Pleonastic morphology dies hard. Evidence from Lithuanian and sundry
International Morphology Meeting, Vienna, 14-17 February 2004.