SOME THOUGHTS ON A COGNITIVE ONOMASIOLOGIC AL APPROACH TO WORD ...

1
Onomasiology Online 3 (2002)
JOACHIM GRZEGA
SOME THOUGHTS ON A COGNITIVE ONOMASIOLOGICAL APPROACH TO WORD-FORMATION
1
WITH SPECIAL REFERENCE TO ENGLISH
Abstract
Word-formation is seldom seen from a cognitive and onomasiological angle. Exceptions are the works by
Pavol Štekauer and articles by Andreas Blank and Peter Koch. This paper evaluates these contributions and
their most relevant points and suggests some further additions to the respective theories. As in Štekauer’s
theory, the approach presented here assumes that a speaker’s mind passes five levels in the name-giving, or
word-finding, process: (i) the conceptual level (analysis of the concept), (ii) the semantic level (structuring of
the semantic markers), (iii) the onomasiological level (“naming in an abstract sense”, i.e. selecting the
iconyms), (iv) the onomatological level (“naming in a concrete sense”, i.e. selecting the morphemes), (v) the
morphonological level (concrete realization respecting a word’s inherent morphonological rules). At the
onomasiological and onomatological levels, speakers can select from 16 different word-formation types
(Štekauer’s 5 types have been supplemented here): conversion (syntactical recategorization), simplex
composites (e.g. lion-hearted), complex composites (e.g. truck driver), mark-absence composites 1 (e.g.
driver) and 2 (e.g. hatter), base-absence composites (e.g. redskin), copulative composites (e.g. deaf-mute),
ellipsis, clipping, acronym, contamination (e.g. brunch), back-derivation, reduplication, morphological
recategorization, word-formation in connection with borrowing (pseudo-loans like telephone, loan-translation
like Fr. gratte-ciel from E. skyscraper or loan-renditions like G. Wolkenkratzer, literally “cloud-scraper”,
from E. skyscraper), clarifying (or post-classifying) composites (like hound dog), and folk-etymology. With
some types formal-aesthetic aspects seem more relevant than salient conceptual aspects.
1. Introduction
At the beginning of each onomasiological approach is a concept that you want to name.
You either (a) choose an already existing name for the concept or (b) you choose to create a
new synonym or (c) it may also be that the concept is so new that it has not even been
given a name yet. As to (a) and (b) two conversational principles that have been felt to be
relevant for linguistic change have been playing an important role for a score of years now:
the so-called efficiency principle and the so-called expressivity principle (cf., e.g.,
Geeraerts [1983] or the summarizing work by Blank [1997a]). At any rate, in cases (b) and
(c) the speakers need find a suitable motivation, an iconym as Alinei (e.g. 1995, 1997) has
called it, for the new coinage. This means that you have to analyze the concept (into salient
aspects): you may see the elements it consists of (partiality), you may see what it looks like
compared to other things (similarity), you may see what it does not look like compared to
other things (contrast), you may see other concepts that the concept to be named is related
with (contiguity) or you may see the relation to other words in the same conceptual field
(taxonomic relations).2 Koch (2001) further subdivides these principles into seven
1
I wish to express my gratitude to Pavol Štekauer for commenting on an earlier draft of this paper. I would
also like to thank my colleague Miller Jones for his linguistic and stylistic comments.
2
Some of these principles remind us of the terms synecdoche/pars pro toto, metaphor, contrast and
metonymy, which, however, have to be placed into the realm of semantic changes only. The associative
principles of “similarity” and “contiguity” in connection with semantic shifts were first investigated by
Roudet (1921), whose assumptions are the basis for Blank’s (1997a) model, in which the principle of
“contrast” has been added. In recent literature (cf. Blank 1997a), synecdoche/pars pro toto has no longer
been separated from metonymy, since the delimitation seems fraught with extreme difficulties. Koch
(1999b), e.g., sees both as relations within a frame (on frame theory cf. Fillmore 1975, 1985). However, in

---

2
cognitive-associative relations: contiguity (i.e. relations within a conceptual frame; he also
calls these conceptual hierarchies engynomies in order to distinguish them from
taxonomies), metaphorical similarity, taxonomic similarity, taxonimic superordination,
taxonomic superordination, cotaxonomic contrast, and conceptual contrast. When trying to
find a name for a given concept the speaker not only has to select from cognitive
possibilities, but s/he also has to select from formal possibilities to transfer these
associations into actual sound: basically s/he may either (a) take an already existing word
and give it a new meaning (i.e. semantic change), (b) borrow an already existing word with
the same meaning from another dialect or language (loan-word), (c) coin an entirely new
lexical item, or (d) form a new word from already existing material (word-formation); the
speech community may also use a combination of these possibilities.3 For illustration I will
take Alinei’s (1995, 1997) example of the terms for GLASSES in various languages and
Dirven/Verspoor’s4 (1998: 54f.) example on the terms for the CELLULAR PHONE. For GLASSES
we find the terms E. glasses (associative principle: partiality; formal type: semantic
change), Fr. lunettes (literally “little moons”; similarity; word-formation), It. occhiali
(literally “things belonging to the eyes”; contiguity; word-formation), G. Brille (from Fr.
briller ‘shine’; partiality; borrowing). For CELLULAR PHONE we find AmE cellular (phone)
(partiality; word-formation); BrE mobile phone (partiality; word-formation) or carphone
(contiguity; word-formation), Fr. portable (partiality; word-formation/semantic change?5),
G. Handy (meaning “[portable in the] hand”; partiality; (pseudo-)loanword)6. While the
topic of semantic change has been seeing a cognitive and onomasiological revival in recent
years (cf. especially Blank 1997a), it is astonishing, though, that hardly any theoretical,
general attempt has been made to view word-formation as a forming process, as an active
process, in other words: as an onomasiologically and cognitively relevant phenomenon.7
Word-formation did not start to be considered a separate branch in English linguistics until
the pace-setting work from the pen of Hans Marchand (1960, 2nd ed. 1969).8 However,
Marchand’s book as well as other frequently cited basic works such as the ones by Lees
(1960), Adams (1973), Halle (1973), Lieber (1981, 1992), Kastovsky (1982), Hansen et al.
(1982), Bauer (1983) and Anderson (1992) share the feature of focussing primarily on the
analysis aspect and neglect or exclude the synthesis aspect, i.e. the active process of
forming proper. Exceptions are Jackendoff (1975) and Aronoff (1976). But in these
(sometimes mathematics-laden) works from the realm of generative linguistics the
extralinguistic concept is more or less ignored. All these theories and approaches9 have in
some cases two concepts within a frame are mingled and in some cases the “frame heading”, as it were,
and a concept within this frame are mingled. I will see the first as contiguity/metonymy and the second as
partiality/synecdoche, which is similar to Bredin’s (1984) nomenclature that synecdoches have to do with
structural relations, while metonymy is based on extrensic relations; but a more detailed discussion of this
issue must be reserved for another occasion.
3
For a more detailed survey on these various formal possibilites cf. Zgusta (1990). The variety of name-
giving possibilites is already remarkably presented by Whitney (1867, Chapter 3, and 1875, Chapter 8).
4
Dirven/Verspoor’s book is a good introduction to linguistics from a cognitive and onomasiological
viewpoint (cf. Grzega [forthcoming]).
5
We will come back to this problem later.
6
The comparison of such possibilities is not only relevant as to single new objects, but sometimes also as to
the development of an entire lexicon, as can currently be demonstrated with the establishment of a
standard variety for the five Dolomitic Ladin dialects in South Tyrol (cf., e.g., Grzega 2000b with a study
of concrete problems).
7
The onomasiological importance of word-formation within a specific word-field, namely trees and fruits,
has been dealt with by Koch (1999a).
8
despite the already very valuable early work by Koziol (1937)
9
Except for Hansen et al. (1982) the theories of the authors mentioned are summarized and evaluated in
the comprehensive survey by Štekauer (2000). A rich bibliography of works on English word-formation
until 1972 is offered by Stein (1973).

---

3
common that diachronic facts, i.e. historical processes, are not taken into account where
this seems valuable. The same defaults can be observed within other philologies. It was
only in 1998 that Pavol Štekauer rang in the cognitive, “onomasiological turn” in word-
formation,10 even though Andreas Blank (1997b) had lectured on word-formation from an
onomasiological viewpoint on the occasion of the International Congress of Linguists one
year earlier—with particular focus on Romance examples. These two linguists as well as a
few thoughts of Dirven/Verspoor (1998) and Koch’s (2001) three-dimensional grid for
lexical diachrony shall be discussed in the following sections. Their ideas will be evaluated
and, if need be, also be complemented in order to enable the integration of word-formation
into a larger project of historical onomasiology that I am carrying out at present.
2. Approach by Pavol Štekauer
2.1. The Elements of Štekauer’s Theory
For Štekauer11 word-formation is about “productive, regular, and predictable
onomasiological and word-formation types producing motivated naming units in response
to the naming needs of a speech-community, by making use of word-formation bases of
bilateral namings units and affixes stored in the lexicon” (Štekauer 1998: 33, his
emphasis; similarly stated already in 1996: 113). These naming units, according to
Štekauer, have a purely lexical function; in contrast to the generative grammatical claim,
there is no link between word-formation and syntax.12
According to Štekauer a word-forming, or word-finding, process consists of five levels: (1)
the conceptual level, where the concept to be named is analyzed and conceptually
categorized in the most general way (i.e. “SUBSTANCE, ACTION (with internal
subdivision into ACTION PROPER, PROCESS, and STATE), QUALITY, and
CONCOMITANT CIRCUMSTANCE (for example, that of Place, Time, Manner, etc.)
” [Štekauer 2001: 11]), (2) the semantic level, where the semantic markers or semantic
components are structured,13 (3) the onomasiological level, where one of the semantic
components is selected as the onomasiological basis (representing a class like agent, object,
instrument etc.14) and another as the so-called onomasiological mark of this basis (the mark
can further be divided into a determining constituent—sometimes distinguishing between a
specifying and a specified element—and a determined constituent),15 (4) the so-called
10
Cf. also the preliminary works by Štekauer (1992, 1996). A concise illustration of his onomasiological
theory is presented in Štekauer (2001) and Štekauer (2000: 1-28).
11
I will mostly quote from Štekauer (2001), since this article as a publication in an internet journal can be
accessed very easily. The passages cited can also be found —partly in the same wording—in other
contributions by Štekauer (cf. bibliography).
12
Problematic cases such as sit-around-and-do-nothing-ish or leave-it-where-it-is-er are solved as follows:
“the Lexical Component cannot fulfil its typical function of feeding the required word-formation bases to
the WF Component for the simple reason of not having them in stock. Therefore, the Lexical Component
mediates the required material from Syntax” (e.g. Štekauer 2001: 26). For a counter-view cf. Hansen
(2000: 173f.).
13
The structuring of semantic markers from an onomasiological point of view is also in the center of a
recent article by Horecky! (1999).
14
Cf. also Beard’s (1995) assumption that there exists a series of universal supralinguistic cognitive
categories (such as “Subjective/Agent”, “Objective”, “Instrumental”, “Locational”, “Diminuition”,
“Augmentation” etc.). These categories, as Štekauer (e.g. 2001: 3) rightly underlines, must not be mixed
up with the distinction between nouns, verbs etc. The category “action”, for instance, can be expressed by
nouns as well as verbs, the category “quality” by nouns, adjectives, or verbs, etc.
15
Štekauer (e.g. 2001: 4) paraphrases this as “naming in a more abstract sense”.

---

4
onomatological level (with the Form-to-Meaning Assignment Principle [FMAP]), where
the concrete morphemes are selected,16 (5) the phonological level, where the forms are
actually combined, respecting morphological and suprasegmental rules.
Štekauer distinguishes five types of word-formation processes: (a) the “Complete Complex
Structure (CCS)”, which formally shows all three constituents—onomasiological base,
determining constituent, determined constituent—, e.g. [[truck] [drive]]-[er]; (b) the
“Incomplete Complex Structure R (ICSR)” (with R standing for ‘right’), where the
determining constituent is not represented in the form, e.g. [lock] [pin], [drive]-[er]; (c)
the “Incomplete Complex Structure L (ISCL)” (with L standing for ‘left’), where the
determined (actional) constituent is not represented in the form, e.g. [hat(t)][er]; (d) the
“Simplex Structure (SS)”, where the onomasiological mark cannot be split into a
determining and a determined part, e.g. [lion-heart][ed] (lion and heart are the specifying
and the specified element of the onomasiological mark, but not the determining and the
determined constituent; cf. Štekauer [1998: 89]); (e) the “Onomasiological
Recategorization (OR)”, which is called conversion or zero-derivation in the traditional
terminology.
Since the terms ICSL[eft] and ICSR[ight] are very Anglocentric (and probably
Slavocentric), I suggest speaking of “ICS2” (“Incomplete Complex Structure 2”) and
“ICS1) “Incomplete Complex Structure 1”. It may be added that not even in English is the
“determinant” always in first position as shown by the type pickpocket (which may be
influenced by French, e.g. coupe-gorge; cf. Marchand 1969: 381) or by a form like center
of attraction (vs., e.g., detention center) with a formative element of17.
2.2. “Conversion”/“Onomasiological Recategorization”
The last type that was mentioned here, “Onomasiological Recategorization”, is especially
important to Štekauer; he even dedicated an entire book to it (Štekauer 1996). Štekauer (cf.
especially 1996: 23-43) views the process traditionally called conversion as a pure
restructuring on the conceptual level and pronounces himself clearly against the theory of a
zero-suffix, a theory that is often found in traditional literature (cf., e.g., Marchand 1969
and Bauer 1983). Cases like e-mail→to e-mail can thus not be explained on the formal
level. The theory of a zero-suffix only makes sense, according to Štekauer (1996: 29, 38),
when there are “true” suffixes with the same function. Otherwise we would also have to
postulate a zero-suffix as a singular morpheme, and cases like sheeppl. would have to be
interpreted as cases with a double zero-suffix or as cases where a singular zero-suffix is
replaced by a plural zero-suffix. However, only with a minority of so-called conversions do
we find variation with “true” suffixes; a good example is cheat (sb.), where a formation
cheater is also imaginable (cf. the pattern write→writer). Other examples are less
supportive of the zero-suffix theory. Thus, Štekauer writes that when we compare clean -
clean and legal - legalize that a form *cleanize is impossible, because -ize can only be
attached to Latinate elements. But then, one could also reply that in- never precedes stems
of Germanic origin (un- can be attached to both inherited and borrowed word-stems).
Nevertheless, his argument must not be ignored in general. Štekauer (1996: 40) still adds
further arguments against the zero-suffix theory: “derivational morphemes can occur in
16
Štekauer (e.g. 2001: 4) paraphrases this as “naming in a more concrete sense”. It means a selection from
the possibilities of expressing, for example, “Agent”; in English this can be expressed by man, -er, -ist,
-ant etc. This also means that synonymy, which can be explained through a diachronical approach, is also
natural in word-formation.
17
On this cf. also Section 6.1.

---

5
word-formation either as allomorphs (e.g. -er, -or, -ar for agent nouns), or as homonymous
morphemes whose word-formation meaning differs (-er1 meaning ‘Agent’, -er2 meaning
‘Instrument’)”. But Štekauer (1996: 40) continues: “In the case of zero word-formation
morpheme, the first, above mentioned, possibility must be rejected. A zero morpheme
cannot be an allomorph of, e.g.[,] the suffix -er because it—if conceded—functions as a
parallel meaningful unit to a number of other suffixes. Moreover, it lacks any formal
relations to the would-be allomorphs”. To me, the similarity does not seem a pre-condition
for allomorphic relationship (cf. more and -er as allomorphs of the comparative). As to the
equivocal nature of a postulated zero-suffix one could object that there are simply several
homonymous zero-suffixes. But Štekauer (1996: 40) writes:
“this yields scores of homonymous zero morphemes because one and the same zero cannot cover all,
semantically very different functions, e.g. Agent (cheatN-V), Quality as a result of Action (cleanA-V)[,]
Time of Action (timeN-V), Object of Action (insertV-N), Objectification of Action (experimentN-V),
Directional nature of the Object of Action (contourN-V), Instrument of Action (switchN-V), and dozens
of others.”
Here we could reply, though, that some of the functions could surely be subsumed in a
more general way. Nevertheless, we must not underestimate the polysemy of some suffixes
(including their metonymical and metaphorical functions)—cf., e.g., the very different
functions of -er in teacher, villager, drawer, toaster, best-seller. However, Štekauer’s
arguments cannot be totally invalidated and all include aspects that, in sum, do indeed
support his objection against the zero-suffix theory to a certain degree.
To Štekauer, the process of conversion is the following. The first basic feature is the
conceptual recategorization: “Thus, for example, databank represents a SUBSTANCE.
When, however, conceptually recategorized, it becomes an ACTION; experiment expresses
a PROCESS—after recategorization it refers to an ACTION PROPER”. With to dance and
dancer we could equally well speak of a recategorization (on the basis of the associative
principle of contiguity) from ACTION to AGENT OF ACTION, of course in combination
with a formal change. It seems as if Štekauer focusses too much on the word instead of the
concept. Therefore, the basic feature of conceptual recategorization doesn’t suffice to
characterize conversion. Štekauer’s second feature is the non-analyzable onomasiological
level, which Štekauer (e.g. 2001: 17) explains as follows: “the onomasiological connective,
as an expression of logical-semantic relations, does not relate the base and the mark; rather,
it relates the motivating and the motivated conceptual categories” (similarly Štekauer 1996:
48). This is convincing and, once more, shows the similarity of this process with semantic
changes, which also take place without formal changes. The third feature is the change of
word-class, which, for Štekauer (e.g. 2001: 18) is a strong argument against the assumption
of a zero-suffix: “While suffixation can be divided into class-changing and class-
maintaining, all new coverted coinages—irrespective of considerable semantic differences
—behave equally in this respect: all types of conversion are class-changing” (similarly
Štekauer 1996: 47). Here, one could argue that the zero-suffix simply belongs to those
suffixes that change the word-class (just like synonymous “true” suffixes). Plus, we may
ask whether the problem of a change of the word-class is not only a problem of languages
that have word-classes. And we may then ask whether “conversion” should be
distinguished from semantic change at all18. If Hockett’s (1976: 23) observation is true that
all languages have at least a “major form-class distinction reminiscent of ‘noun’ versus
‘verb’ [...], though not always at the same size-level”, then we may keep the distinction
18
Tournier (1985: 48) also groups conversion and semantic change (which he calls “métasémie”) under the
same category of “semantic neologisms”.

---

6
between “conversion” and semantic change. It is then the only criterion so far. Another
important feature according to Štekauer is the phonological/orthographical identity
between the original form and the converted form (which, again, yields no basis for
differentiationg between “conversion” and semantic change). Štekauer (e.g. 2001: 20)
criticizes Marchand for his alleged natural definitions: “Marchand’s definitions of whistleV-
N ‘forcing the breath through the teeth or compressed lips’ vs. ‘instrument used for
whistling’ do not appear to be more natural or obvious than the following pair: ‘to use a
whistle’ vs. ‘an instrument operated by air expelled from lungs’.19 Well, it seems logical,
and therefore indeed natural, to suggest that ‘forcing the breath through the teeth or
compressed lips’ must be the primary sense, whereas ‘wind instrument’ is secondary and
‘to use a whistle’ must be tertiary (no use of the instrument without the existence of the
instrument). But I would argue that for an onomasiological approach diachronic facts must
be regarded as decisive20. Štekauer’s (2001: 20) second point of criticism that “[c]ontrary to
Marchand’s assumption (1955: 172) it is possible to ‘saw without a saw’ just as it is
possible to hammer without a hammer” can be refuted by the help of prototype theory.
Sawing and hammering without a saw and a hammer seem just peripheral, or metaphorical,
members of the respective categories. After all, even Štekauer (e.g. 2001: 21) admits:
“in the vast majority of cases, this way of determining the ‘derivational’ relations resembles the
‘familiar’ chicken-or-egg problem [....]. Therefore, the only way out seems to consist in the
complementary effect of a multiplicity of criteria, including the criterion of extralinguistic
subsequence, diachronic data, formal criteria (like stress pattern), morphosyntactic effects [...],
structural relations (combinability with affixes), etc.”
Nevertheless, diachrony is far too often neglected, and this seems to me the most
vulnerable aspect in Štekauer’s theory. This is plainly visible in his own example of milk,
the evolution of which he sees as milk ‘liquid substance given by a cow’→milk ‘to obtain
milk from a female mammal’. A look at the historical facts shows that we are not dealing
with a case of conversion, but with one of derivation; from the noun milc (according to the
OED recorded for the first time around 900) speakers derived a typical denominal weak
verb of class 1, milcian, (according to the OED recorded for the first time around 1000).
Besides, we may wonder whether today we would coin, for a still unnamed concept ‘to get
milk out of a cow’s udder’, a form to milk or whether a new form to milk would not rather
serve to denote ‘to give milk’, ‘to use milk’, or ‘to add milk’; many conversions—at least
those between nouns and verbs—seem to express ‘making’, ‘using’, ‘providing’ or
‘directional/locational’ relations. Thus, we have “true” conversions of milk in to milk the
tea, to milk one’s lamb [of a cow], and to milk the bottles.21 There are even cases of re-
conversions, e.g. handbag [object]→to handbag [action]→handbag [process].
In Štekauer’s theory a few cases are problematic, because they do not show total
phonological identity, e.g. ábstract (sb.) vs. abstráct (adj.). Tournier (1985: 174) speaks of
“quasi-conversions” here. In these instances Štekauer (1996) takes historical facts into
account and comes to the following result:
19
Similar criticism was already raised in Štekauer (1996: 130).
20
As a matter of fact, according to the OED, ‘instrument’ is already recorded for ca. 950, ‘breathing’ only
for ‘1050’ (by accident?). The sense ‘using a whistle’ is not attested before 1530.
21
It seems as if all of Štekauer’s (1996: 104ff.) examples can equally be subsumed under these few major
relations. Štekauer himself, however, refrains from such a narrow limitation and says: “The number of
possible meanings of new converted meaning units is limited by the number of actual meanings of a
potentially polysemantic motivating naming unit, and the number of potential onomasiological
connectives (logical and semantic relations) between the motivating and the motivated neaming unit”
(Štekauer 1996: 106).

---

7
“The employment of a diachronic method resulted in the division of examined material into two
groups: genuine conversion pairs, on the one hand, and etymologically excluded pairs, on the other.
[....] It is only the first of them which results from a word-formation process (conversion), while the
identical orthography of the pairs of the latter group resulted from a historical convergence of two,
originally independent, forms. [....] conclusion: there is basically no difference in the phonological
behaviour, or properties, between the two groups in question. From this it follows that the
phonological differences between the converting and the converted words of any conversion pair have
not been predetermined by any specific word-formation (i.e. conversion-specific) rules. On the
contrary, all these differences follow the general tendencies rooted in the word-class of the particular
members of a conversion pair [....]: they are not meaning-constituting devices, but only devices that
may function as meaning-distinctive ones” (Štekauer 1996: 93f.; his emphasis).22
This view, however, appears a little simplistic to me and seems to be thought of as an
auxiliary contrivance to be able to defend the thesis of a hundred-percent regularity and
predictability of word-formations. In general, I accept this thesis, but I don’t consider
Štekauer’s wording very efficacious, since the consequence is that many word-formation
processes are not viewed as such or are—as in this case—misinterpreted. I will delve into
this problem in more detail below. First, a few more fundamental thoughts on conversion
shall be added here. In an onomasiological approach, the starting-point should always be
the concept to be named. The concept gets analyzed, and salient features and associations
(similarity, contrast, contiguity, partiality, taxonomic relations) are activated in the mind.
Then the speaker, or the speech community, selects from the repository of productive
word-formation possibilities and discovers that, particularly with contigual associations,
there is also the possibility of selecting, without any formal modification, a word that is
used in a different syntactical position, but typically in a frequent paraphrase for the
concept to be named. From the paraphrase to write an e-mail or to use e-mail the speaker
“takes out” the rhematic, salient part and gets to e-mail. Out of the instrument for whistling
the speaker makes a whistle and from to use a whistle s/he forms a new to whistle. The
occasional shift in the stress pattern is explanable through the synchronically different
model patterns (which, in return, are themselves explanable by a diachronic study, e.g.
through the loss of inflectional suffixes with the borrowing of Gallicisms).
Besides, I do not want to ignore the fact that some words are certainly converted rather
subconsciously, e.g. fun. The starting-point is the choice of saying That’s funsb! and That’s
funnyadj! without a difference in meaning. The noun and the adjective take the same
syntactical position here. Therefore it can happen that word-class boundaries are blurred
and that in the formation of a comparative fun is treated like an adjective. At least in the
US, That’s even funner! or That’s a fun thing to do! can be heard (at least in some regions),
so that future lexicologists may add a new sub-entry funadj. to their dictionaries.
Štekauer (1996: 115ff.) also deals with the typically English feature of converting proper
names. In Clark/Clark’s (1979) standard sentence My sister Houdini’d her way out of the
locked closet, for instance, the verb to Houdini has to be understood as ‘to escape by way
of a trick’. A salient feature of the name-giving person serves to denote the same feature of
other persons. In contrast to other denominal verbs, the hearer can only decode such
sentences and forms when provided with the relevant encyclopaedic knowledge.
One particularity hasn’t been mentioned so far. It may be that a word is obviously not fully
conversed, i.e. that it doesn’t adopt all features of its new word-class, e.g. the poor (instead
of *the poors). Tournier (1985: 174) speaks of “partial conversion” here. I, on the contrary,
22
Tournier (1985: 180), too, points out that there may be “pseudo-conversion” because of double
borrowing.

---

8
would prefer to categorize these formations as ellipses (e.g. from the poor [people]).
In sum, we may still wonder whether semantic change and conversion should be kept apart.
Cognitive-associative differences are absent, the formal differences are minimal and only
become visible within the surroundings of a text. However, conversion allows stress shift,
which semantic change does not (unless we newly define it that way). It is for these two
differences that the distinction between conversion (or “syntactical recategorization”, as we
may henceforth call it) and semantic change will be kept here.
2.3. “Exocentric Compounds,” “Back-Derivation,” and “Bracketing Paradoxes”
Štekauer also casts light on three other traditional “problems”, namely the problem of
exocentric compounds (cf., e.g, Štekauer 1998: 147-154), that of back-derivation (cf., e.g,
Štekauer 1998: 154-162) and the problem called “bracketing paradoxes” (cf., e.g., Štekauer
1998: 127-142).
As an example Štekauer mentions the form unhappier, which would have to be analyzed as
[un]-[[happy][er]] from a morphological point of view, since the comparative suffix -er is
only added to monosyllabic and some disyllabic words. However, from a semantic point of
view, as Štekauer convincingly states, unhappier has to be interpreted as ‘more unhappy’
rather than ‘not happier’. Štekauer (e.g. 2001: 29) demonstrates how the problem can be
solved with his approach:
“Since the onomasiological theory with its FMAP [i.e. Form-to-Meaning-Assignment Principle] does
not rely on a binary word-formation structure, the problem of bracketing paradoxes is meaningless.
Moreover, the proposed approach is based on the principle that the relations in question are not
hierarchical. The members of the onomasiological structure (the base, the determining and determined
constituents of the mark, and the specifying and specified elements of the determining constituent)
function at the same level of description.”
Although the comparative form unhappier is actually a problem of morphology, not of
word-formation, the Form-to-Meaning-Assignment Principle can nevertheless solve such
problems due to the assumption that people simply select from the number of semantic
markers given.
As regards the compounds that are traditionally called “exocentric”, “bahuvrihi” or simply
“pseudo”-compounds Štekauer writes (e.g. 2001: 3; his emphasis):
“I propose to explain ‘exocentric compounds’ by a two-step process in which only the first has word-
formation relevance. The first step consists in the formation of an auxiliary, onomasiologically
complete (i.e. with both the base and the mark included), naming unit. The second step is based on
mere elliptical shortening. [...]. Therefore, this type of naming units can be analysed on a par with the
underlying ‘full’, auxiliary, version, although the latter has not come to be used (institutionalised).”
Štekauer substantiates his theory by claiming that the plural of sabertooth is not
*saberteeth, but sabertooths; therefore, we would have to depart from a shortened
onomasiological base (e.g. animal or tiger). But the plurals of the plant-name horsefoot and
of tenderfoot ‘newly arrived immigrant’ would have to be *horsefoots and *tenderfoots
then, but this is not the case (in both instances we have -feet). Therefore, it seems more
suitable to assume a combination of metonymy/pars pro toto and composition (or to say
that not both elements of the contigual relation have to be expressed in a word-formation
unit). Štekauer (2001: 32) says that his explanation “is more ‘natural’ in terms of word-

---

9
formation principles and corresponding to the psychological reality of coining new naming
units,” which includes the theory of the traditional identification-specification scheme. To
me, it appears equally natural to say that, at first, a specific salient feature of the concept to
be named is selected and then formally realized by way of compounding. Štekauer could
solve the problem with his own approach if he added a sixth word-formation structure,
which could be termed “Incomplete Complex Structure B (ICSB)”, where B stands for
base and where the base is not represented in the form.23 Then the type killjoy, wagtail,
catchfly would easily fit into this category, too, even though with a reverse determination
structure. This structure seems especially popular when the possible base is semantically
very vague and general, a passepartout word such as man, thing, or animal. As regards the
cognitive process, though, catchfly and redskin do not quite fall together: in the first case
the object is a catching thing, whereas in the second case the object has a skin.24
Finally, there is the problem of back-derivation, e.g. stage-manager→to stage-manage.
Štekauer (e.g. 2001: 32) writes:
“The conceptual fallacy in traditional accounts of back-formation is that they explain the origin of a
‘shorter’ naming unit (e.g., stage-manage) without accounting for the way in which a ‘longer’ (stage-
manager) naming unit came into existence. ‘Longer’ naming units must have been somehow coined,
they could not merely have appeared ‘out of the blue’. Moreover, the suffixes included in ‘longer’
naming units have all the features of ‘normal’ suffixes. Therefore, I believe that both members of the
‘pairs’ related by the notion of ‘back-formation’ are generated separately.”
This, however, is not only against intuition, but also against the historical facts, which are,
once again, excluded. Of course it is correct that the speaker first goes through the
conceptual, the semantic and the onomasiological level. On the onomatological level,
though, the “longer” form comes into play as a formal model and onomatological lure. It
seems inept to assume totally separated formation filiations.
2.4. Morphemes and Morphs
A few more thoughts shall be added to Štekauer’s approach. Štekauer writes (e.g. 2001: 2):
“While Beard ‘evicted’ affixes from the ‘community’ of majors classes (N, V, A) by
claiming that—like articles, adpositions, conjunctions, and some pronouns—they ‘bear no
semantic content but reflect grammatical function [...]’ I find affixes to be on a par with
lexemes (both are form-meaning units)”. Here it could be replied that there are simply two
types of affixes: one with semantic function (e.g. ModE un-), the other with grammatical
function (e.g. ModE -ness). It also seems not right to say “that no naming unit can be
generated from units smaller than the morpheme, with the morpheme being defined
traditionally as the minimum bilateral sign, having its own specific form and specific
meaning”. Certain expressive or onomatopoetic words are surely based on morphs, not
morphemes. A word like clash, for instance, is on the one hand formally based on words
like cl-ather, cl-ack, cl-ap etc., on the other hand on words like d-ash, l-ash, cr-ash etc. (in
clash cl- could be regarded as the determinant and -ash as the determinatum [cf. Hansen et
al. 1982: 141ff.]).
2.5. Blends and Acronyms
Since for Štekauer word-formation patterns are a hundred percent productive (and thus
23
I think that this is more apt than postulating an intermediate type ICSR (cf. also, e.g., Štekauer 2001: 34).
24
In Blank’s (1997b) approach these two types are separated, as will be illustrated below.

---

10
regular and predictive), he excludes blends and acronyms from word-formation.25 My view
is different. I see word-formations as neologisms out of material in one’s own
dialect/language. Thus, blending and acronyming, although not traditional and central
word-formation processes, fall perfectly well into this category. In any case, I do not really
understand Štekauer’s view that acronyming cannot be seen as a word-formation process
on the ground that acronyms have the same meaning as their long forms. At least, I cannot
agree with this view—or at least not with the wording. When an American calls a black co-
citizen not Black any longer, but Afro-American or African American, then a new meaning
hasn’t been created either; nevertheless everybody would regard the two new terms as a
result of word-formation. However, I do agree with Štekauer when he states (personal
communication) that the two latter examples represent the result of a fully new and
independent word-formation process passing all word-formation levels, whereas acronyms
are formed on a formal level only. Another aspect that is a little unfortunate in my view is
that Štekauer pursues only Modern English situations. For him Monday and cranberry are
uninteresting for word-formation, because Mon- and cran- are not morphemes, but rather
similar to phonemes (since they don’t carry, but only distinguish meaning). However,
when these words were coined they were of course transparent compounds/syntagms;
Mo_nan dæg was absolutely transparent in Old English times. I would like to see the
beginning of a word at the beginning of an onomasiological theory. On the other hand, the
following allegations are fully convincing. Štekauer (2001: 8) answers to the “Chomskian
claim that words which result from derivational processes often depart from their
‘expected’ meaning”—like revolve vs. [French] revolution or construct vs. [genitive]
construction—that this is not part of a word-formation process, but takes place in the
lexical component of the mind. We could also say that this is a case of semantic change, or
even: collocational semantic change.
2.6. Analyzing a Few Problematic Word-Formations
At the end of the evaluation of Štekauer’s approach I want to contemplate a few concrete
problematic cases.
(1) Let us have a look at the word butterfly. According to Štekauer’s model, we would have
to view fly as the onomasiological base. The base is the element “denoting a class, gender,
species, etc. to which the object belongs” (cf., e.g., Štekauer 2001: 11). In the first case we
could at least speak of a metaphorical classification (with butter being the onomasiological
mark), but in the second? It would in my opinion be wrong to put all such cases completely
into the Lexical Component. I shall analyze butterfly as “mark + base” here.
(2) Let us now have a look at the term brimstone butterfly. Here we can’t assume a typical
three-fold distinction brimstone-butter-fly, with brimstone being the determining
constituent and butter the determined constituent. It is rather the case that brimstone
specifies butterfly as a whole. In this case, it only makes sense to assume that butterfly is
the onomasiological base and brimstone the onomasiological mark. This already seems to
be covered by Štekauer’s model, but it seems important to me to show the difference
between “bi-partite” compounds and “pluri-partite” compounds.
(3) We will now ask for the theoretical classification of skyscraper, which will also be
analyzed in Blank’s approach (cf. below). Štekauer (personal communication, 1998: 89s.)
25
More bluntly, blending, to Štekauer, is a two-step process, the first step being identical with compounding,
the second step (“shortening”) falling into the Lexical Component. (cf. also Štekauer 1997).

---