Documente Academic
Documente Profesional
Documente Cultură
and if so is it productive?*
LAURIE BAUER
Abstract
Introduction
This paper has its origins in two points of unease with current linguistic
theory. From the second of these we can derive two further questions.
The first of these is the theoretical point made by many Dutch morphol-
ogists from Schultink onward, and inducting Van Marie (1985), that
productivity deals only with the coinage of new words in a subconscious
fashion. In particular, Van Marie interprets this as meaning that no
words that are coined exclusively on a foreign basis can be coined
productively: "Only those morphological processes may rank as 'pro-
ductive' which (i) can be fully characterized in terms of 'major lexical
categories,' and which (ii) are not restricted to the 'normative' strata of
the lexicon" (Van Marie 1985: 60).
Neoclassical compounds
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 405
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
406 L. Bauer
I stated at the outset that part of the motivation for this paper lay in the
observation that there were many words that failed to fit this fairly neat
description of neoclassical compounds. I shall now justify this claim,
illustrating with words that can be found in dictionaries of established
words and dictionaries of neologisms (Ayto 1989, 1990; Barnhart et al.
1990; Butler 1990; Green 1991; Tulloch 1991, etc.).
The first major type to provide such descriptive problems is the type
made up of an English word combined with one of these neoclassical
stems, words like sociof linguistics]. Specifically, we need to ask whether
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 407
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
408 L. Bauer
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 409
A descriptive metaphor
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
410 L. Bauer
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 411
neo-classical cpd:
atypical geology
butchers
transgenic
hydro-slide
management
word-manufacture: native compound:
skag redress house-boat
This description, however, fails to take into account the words that
were listed earlier in which at least one of the elements was clipped.
Where compounds are concerned, the genre is well known for English,
with examples such as heli-tele TV or video camera mounted on a
helicopter'. The examples discussed earlier show that this type is also
found with words that include a combining form. Prefixes and suffixes
are also sometimes added to shortened bases (NB: this should be distin-
guished from instances where an affixed form is subsequently subject to
clipping): heritize (< heritage). At the simplex end of this scale are words
that look like simplices but are etymologically acronyms, though the
original may not be known to speakers: laser, radar, etc. A third dimen-
sion, therefore, is provided by a cline from abbreviated to nonabbreviated,
with intermediate steps being provided by degrees of shortening; see
Figure 2.
Discussion
The suggestion being made here, then, is that new words can be envisaged
as being constructed within a three-dimensional space, and that the three
dimensions are simplex-derivative-compound, native-foreign, and non-
abbreviated (or "full")-abbreviated. A large part of the motivation for
this notion is the number of compromise words that are found. It is
possible to consider this three-dimensional space in two ways: the first is
that it can be seen as a series of 12 cells in a matrix, each representing
the intersection of three of the marked points on the three dimensions;
the second is that compromise words can be sought — forms that sit
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
412 L. Bauer
Perestroika KGB
,
glitznost
heli-tele
op-art
redness pre-AIDS ag
nonabbreviated abbreviated
between the extreme positions. I shall consider the notion from both
points of view.
First of all, consider the 12 cells that arise from the intersection of the
three dimensions. These are laid out in Table 1.
I would like to draw particular attention to two things in Table 1. The
first is the undifferentiated use of the label "loan," which makes no
reference to the structure of the borrowed word. It might be argued that
the structure of the borrowed word is irrelevant in English — after all,
we do not need to know that perestroika is a derivative to use it properly.
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 413
There are at least two problems with such a view. The first is that exactly
the same statement could be made about words whose structure is
English: we do not have to recognise dearth as a derivative to use it
properly. This is not, however, generally accepted as a reason for ignoring
this structure. For the people who introduce loan words, the structure in
the lending language is probably transparent. The second problem is that
there are cases where the structure CAN be deduced by the monolingual
speaker because of parallel structures in other loan words (consider
arbitrageur, justice, sputnik, etc.). It may be that nothing hangs on this;
but unless we are sure, we should not ignore such information. The
second point about Table 1 that I wish to draw attention to is the number
of types of word formation that are not mentioned in the table, includ-
ing, specifically, neoclassical compounds. This is because neoclassical
compounds (and other types such as blends, clippings, etc.) are in fact
compromise types.
I shall now turn to consider the compromises between the types set
out in Table 1. It was suggested in Figure 1 that a compromise between
native and foreign at the simplex end might be loans from other varieties
of English. We might also take the point of view that acronyms that do
not comply with native phonotactic rules are a compromise here: I know
of no such examples in English, but fnac in French is such an example.
We might also point out that some loan words do not comply with
phonotactic requirements of English, though most are made to do so,
and they are more foreign than those that do comply with such rules.
Compromises between native and foreign derivatives are those "less
foreign" foreign derivatives that come from French, and words with
foreign bases but English affixes. Forms like sociolinguistic, discussed
earlier, provide such compromises at the compound end of the scale.
And neoclassical compounds themselves, being formed from foreign
elements but as words of English, are also a compromise set here.
Blends with recurrent splinters are compromises between derivatives
and compounds, as are forms such as sociolinguistic, which are a bit like
derivatives and a bit like compounds. Between derivatives and simplices
we find conversion, which though parallel to derivation in function shows
no formal marking of that function. Another type of compromise here
is derivation by ablaut, where there is some formal marking, but it is not
additive.
At the compound end of the scale we can distinguish between many
degrees of abbreviation, with the examples in Figure 3 showing a cline
of abbreviation. I am in two minds as to whether initialisms that are not
acronyms should count as abbreviated compounds or as new simplices
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
414 L. Bauer
photograph opportunity
hazardous chemical
I
houseboat guestimate photo-op e-mail Hazchem PC
Increasing abbreviation ^
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 415
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
416 L. Bauer
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 417
hard to use on some occasions and easy on others. While I agree that
the matter has not yet been resolved, it seems to me at the moment that
the evidence favors the notion that there is a gradual cline from the most
productive to the least productive type of process rather than an abrupt
division.
There is an apparent counter-argument to this, which the approach
outlined above appears to ignore to its detriment, and that is that
although actual word types that get listed in dictionaries of neologisms
have many intermediate patterns, as we discussed earlier, the really pro-
ductive patterns — the patterns that are so productive that the words
formed according to them do not make it into dictionaries at all and yet
can be found in large corpora — are not formed according to these
intermediate patterns, but are formed on relatively restricted patterns
that are much closer to the prototypes (Baayen and Renouf 1996). The
most productive patterns in English, for instance, include the formation
of adverbials with -ly suffixation, the formation of nouns by -ness suffix-
ation, and so on. If words are formed without recourse to formal rules,
why should this general pattern be observed? If the answer is simply one
about rule-governed or non-rule-governed, then the new model is of no
value. If there is an alternative answer, the new model may be useful.
A first approximation to an alternative view may be found in the view
of analogy put forward by Becker (1990). According to Becker all word
formation is by analogy, but some analogies are more likely than others.
One of the factors affecting the likelihood of analogies is the type fre-
quency of potential models. Because -ness and -ly provide many potential
models, they are more likely to form the input to new analogies than the
-ter of laughter and the -ric of bishopric. Of course, this is not sufficient.
There is no precise equivalence between the type frequency of words
using a particular morphological pattern and the productivity of that
pattern. Productivity can vary apparently independently of such matters.
Consider Figure 4, which shows the diachronic productivity of -ment in
English according to the OED. If the productivity of -ment were deter-
mined purely by the number of appropriate analogies, we would expect
a gradual and consistent increase in the use of the suffix. If the number
of available bases were the only other consideration, we might expect a
reduction at the end of the period. In neither case would we expect the
two peaks of productivity that the OED suggests. (The problems of
dealing with a source like the OED apply here, as elsewhere, of course;
the real productivity in terms of new coinages must have been higher at
most periods than is shown by this data; nevertheless I take it that the
pattern shown here is at least indicative.)3
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
§
Years
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 419
This is still not enough: truth, reality, and correctness are all formed in
the densely populated part of the space, but only one of these patteras
shows any great productivity, so more is required. In this connection, it
would be interesting to know whether all languages choose the same part
of the three-dimensional space to have their preferred word-formation
patterns or not. If they do, this may be to do with general cognitive
principles for the formation of new words.
One such general cognitive principle, helping to answer the question
as to why the prototypical cases are the most productive ones, may
have to do with transparency. Classic cases of prefixation, suffixation,
and native compounding are exhaustively analyzable into recurrent
units of form that have a fixed meaning (the classical morpheme, in
fact). This makes such words not only easy to interpret, but easy to
form (see, e.g., Clark 1993: 116). Along with other principles, such as
the preference for formations that do not involve morphophonemic
changes (Clark 1993: 120), such considerations restrict the patterns
that can be maximally productive. There must be other such prin-
ciples, and there would be value in finding out what they are. An
anonymous referee suggests the existence of a transpositional function
as well as a lexical-enrichment function for some processes and the
ability of some processes but not others to create textual coherence.
These seem like excellent candidates, and I should like to see the list
developed.
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
420 L. Bauer
Notes
* Versions of this paper were presented at the University of Essex and at the DGfS
conference in Düsseldorf in February, 1997. I should like to thank members of both
those audiences for their comments, and the Deutsche Forschungsgemeinschaft for
financial support to allow me to attend the conference. I should also like to thank
Winifred Bauer for her considerable help with this paper, and two anonymous referees
for their comments. Correspondence address: Victoria University of Wellington,
P.O. Box 600, Wellington, New Zealand. E-mail: laurie.bauer@vuw.ac.nz.
1. There are some exceptions in Classical Greek, such as hippopotamos 'horse-river'. Some
grammarians seem to suggest random head-modifier order when they comment that
laimo-ponos 'throat pain' is synonymous with pono-laimos 'pain throat' ('sore throat') —
I suspect this is not so, although the two compounds must have very similar freedom of
occurrence.
2. The term is not a general one but can be seen, to some extent, as the converse of Bybee's
(1985:16f.) notation of generality: the more general an affix, the less specific its meaning;
neoclassical elements have a level of specificity more nearly equivalent to lexemes that
to most affixes. This notion, under various headings, goes back at least to Sapir (Bybee
1985: 7). It is not, however, a measurable concept, but an observation of difference
based on what I take to be a fairly common intuition.
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
Neoclassical compounds 421
3. An anonymous referee points out that the figures from a source such as the ED have
no implication for the ability of speakers to construct new words. This is correct, but
not strictly relevant. In terms used by Corbin (1987: 177), the figures from the OED tell
us that the suffix -ment was disponible 'available' in all of the periods shown; but they do
tell us quite a lot about the rentabilite 'profitability' of the suffix at various periods. Since
the discussion here is in terms of how productive things are, it is the latter that is in
question. There is another point of view, that profitability is something that is related to
speakers rather than to grammars (Langacker 1987: 71-2). If that is the case, then all
the compromise types discussed in the early part of this paper are productive on a par
with the common types like suflfrxation and native compounding. Moreover, under this
assumption, the counter-argument being discussed here becomes vacuous, and the view
that there is no difference between automatic and limited types of productivity becomes
easier to uphold.
References
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum
422 L. Bauer
- 10.1515/ling.1998.36.3.403
Downloaded from De Gruyter Online at 09/19/2016 11:20:53AM
via Ruhr-Universität Bochum