?tekauer Pavol, Lieber Rochelle. Handbook of Word-Formation
Подождите немного. Документ загружается.
LEXICALIZATION AND INSTIT
U
TIONALIZATION
361
generally shared vocabulary,
1
2
along various parameters, on a scale of
idiolect –
d
ialect – languag
e
,
cf.. Lipka (2002: 22), where ‘dialect’ is understood in the
broadest sense of the language of any subclass of a speech community.
The smallest settin
g
of a speech communit
y
,
the subclass
j
ust above the idiolect,
i
s that of a couple. Here, intimac
y
can foster extreme idios
y
ncrasies – however, due
t
o that ver
y
intimac
y
of such a settin
g
, robust empirical data are hard to obtain. Onl
y
v
ery occasionally do such examples surface outside their intimate domain (but
r
eaders are invited to use intros
p
ection her
e
)
. A few anecdotal cases are described in
H
ohenhaus (2005b), e.g. the highly idiosyncratic ‘back-formation’ of a singula
r
*
s
hoo
p
f
r
om
s
heep
–
originally a deliberate jocular deviation, which did howeve
r
become established in the couple’s micro-dialect.
The next larger ‘community’ will be that of the family or other such more or less
stable small
g
roup (close work collea
g
ues, band members, small teams of explorers
o
n an expedition, etc., etc.). Herin
ge
r(
1984: 9
)
mentions the
p
henomenon of
e
p
isodic com
p
ounds for such small
g
roups – a potential example he constructs is
G
erman
Mäusebibel
‘mice bible’, which is useable by family members who all
k
now about a past incident in which a bible showing teeth marks of mice (who had
apparently nibbled at it) was found by the family in a barn. It is thus only on the
basis of the common episodic knowledge that the compound can be institutionalized
i
n that meaning within this family’s small-group dialect.
A
t the next hi
g
her level lie the special vocabularies of technical
j
ar
g
on, slan
g
etc.
– our own context here provides plenty of
examples. Acronyms usually make this
f
particularl
y
clear: ‘la
y
’ people outside lin
g
uistics would hardl
y
be able to decode
NP
,
LF
G
,
GB
,
HP
SG
, or
OT
r
(more than one meaning here!). Likewise, acronyms
T
suc
h a
s
AD
SR
,
VCA
o
r LF
O
(from synthesizer technolog
y
) will probably leave most
o
f the readers of this book baffled. Many of the emerging conventions, including
m
ore or less institutionalized acronyms, in
t
he new ‘e-language varieties’ such as
I
nternet chat pose similar decoding difficulties to outsiders or ‘newbies’, e.g.
lol
,
o4u
,
cul8
r
,
cfv
,
imho
,
etc.
(
cf. section 4.4
)
.
Of course, if we want to approach an
y
de
g
ree of lin
g
uistic
g
eneralit
y
we have to
l
ook at lar
g
er settin
g
s, ultimatel
y
a
t
the
speech communit
y
of a lan
g
ua
ge
at lar
g
e,
i
.e. all its speakers. As the ed
g
es of th
i
s concept are necessa
r
il
y
fuzz
y
, this is
n
aturally an idealized entity itself.
A
t any of these levels, however,
g
enericness – based on shared knowledge
–
s
eems to be of crucial im
p
ortance
(
cf
.
Heringer 1984, Hohenhaus 1998). The more
generic the meaning-form pairing in a word-formation is, the wider its
i
nstitutionalization potential. The less generic, more context-dependent o
r
i
ndividually episodic, the narrower the institutionalization range (cf. section 4.2.).
I
n short
,
what institutionalization actua
l
l
y
amounts to hin
g
es on what sort of
‘in
s
tit
u
ti
o
n’
we
ar
e
talki
ng
about, ran
g
in
g
fro
m
c
ouples and micro-
g
roup settin
g
s via
i
ntermediately-sized groups of jargon speakers to ‘the’ speech community at large.
B
ut such is the sociolinguistic nature of the concept. It is precisely this which makes
12
Cf. McWorter (2001) who repeatedly stresses that th
e
same principle applies generally to languages in
a
w
i
de
r
d
ia
c
hr
o
ni
cse
n
se:
th
es
mall
e
r an
d
m
o
r
e
i
s
o
lated a speech communit
y
is
,
the more ‘exotic’
features its lan
g
ua
g
e tends to develop.
36
2 PETER H
O
HENHA
US
t
he term a useful addition to our other terminology. It also means that ideally we
should always use the term together w
i
th its s
p
ecification ‘institutionalized i
n
X
,
XX
Y
or
Y
Z
’.
3.3
D
e
-I
nstitutionali
z
ation
: the end o
f
a word’s li
f
e
W
e also have to mention briefl
y
that the results of lexicalization and
i
nstitutionalization are not necessaril
y
pe
r
manent states. Obviousl
y
enou
g
h, words
d
o not
j
ust enter a lan
g
ua
g
e, the
y
can also ‘exit’, become obsolete, die out. Lipka
(
2002: 53) predicts such a fate for e.g.
m
illennium bug
(
correctly, it can be assumed,
esp. given that it caused so much less damag
e
than
w
a
s
f
e
ar
ed be
f
o
re
th
e
t
u
rn
o
f th
e
m
illennium
)
. However, because words, o
n
c
e institutionalized in written varieties
,
l
eave a permanent record, they may not completely vanish into oblivion. This may
h
ave a distorting effect especially for corpus-based studies, unless frequency is
o
vertly linked with specific time-spans.
C
onsider another pair of examples of acron
y
ms:
13
WM
D
vs.
IDS
.
Th
e
f
o
rm
e
r i
s
probabl
y
one of the hi
g
hest frequenc
y
acron
y
ms in politics of recent
y
ears
14
(although one hopes it will eventually drop out of
use together with its referents, but
f
for the foreseeable future this is firmly institutionalized
)
. The latter, on the other
h
and
,
was used in a British context
1
5
for a few years with considerable frequency
(
esp. in the tabloids) to refer to the leader of the Conservative Party
Ian
D
uncan
S
mith
.
Since his resignation, the acronym has also notably decreased in frequency
and is likely to slowly fall out of use altogether
.
1
6
Th
e
fa
c
t that
wo
r
ds beco
m
e obso
l
e
t
e
i
so
f
cou
r
se
n
o
t a n
ew obse
r
v
ati
o
n
.
H
owever, it is often assumed that the (En
g
lish) vocabular
y
is constantl
y
en
j
o
y
in
g
‘
g
rowth’ (cf. e.
g
. Hu
g
hes 2000). While this is certainl
y
true for the
reco
r
ds
th
a
t
l
exicographers keep accumulating, it is u
n
l
ikely that the mental lexicon, capacious
as it may be, can enjoy similarly endless ‘growth’.
De
-
lexicali
z
ation
, shrinking o
r
d
eleting of lexical entries in the mind must therefore also be an important
psycholinguistic factor.
13
Lipka (2002: xviii,110, 146) ri
g
htl
y
stresses the importance of ac
r
o
n
y
m
y
as a powerfull
y
productive
m
eans of creating new lexemes, even though it poses problems for many word-formation theories in
t
hat its morphological status is unclear, since the process is at best only partly rule-governed.
1
4
Standing fo
r
Weapons of Mass Destruction
–
a Google Internet search on 28 June 2004 returned
768,000 hits in English, a search restricted to sites under a year old still returned 705,000 hits,
c
orroborating the assumption of highly increased c
u
r
rent frequency. Sampling and refined searches
w
ithin results confirm that the ma
j
orit
y
did ind
e
ed include this ver
y
acron
y
m – althou
g
h there were
also a few in which it stood for something else, i
n
c
luding, curiously, the name of the organization
World Movement for Democrac
y
.
1
5
Of course, for Germanist linguists, the
s
ame acronym still stands for the renowned
I
nstitut für D
II
eutsche
S
prache.
S
S
p
1
6
It is still to be found
,
of course. For instance, it was dug out in June 2004 in a BBC Television
pro
g
ramme comparin
g
the results of the Tories under
t
heir respective leaders i
n
the last few European
elections. Empiricall
y
assessin
g
the frequenc
y
of this word is difficult, a) because the hei
g
ht of its
c
urrenc
y
and its subsequent decline are too recent to be covered b
y
the available standard corpora, and
b) web-searches reveal countless other readin
g
s for the letters ID
S
,
so that even a time-s
p
an-refined
s
earch quer
y
would require disp
r
oportionate filterin
g
efforts.
L
EXICALIZATION AND INSTIT
U
TIONALIZATION
363
4
. PR
O
BLEM
S
As we have alread
y
seen, there is considerable disa
g
reement about the
t
erminolo
gy
and the concepts revolvin
g
a
r
ou
n
d
th
e
i
ssues u
n
de
r
co
n
s
i
de
rati
o
n h
e
r
e.
Partl
y
, these derive from different theoretical priorities and decisions, but the
y
can
also be related to
p
articular em
p
irical and further conce
p
tual
p
roblems, whose
h
andling must affect such decisions.
4.1 Nonce-formations and neologisms
The terminological confusion already noted above is perhaps at its worst here.
Frequentl
y
, the two terms in the title of th
i
s section are used almost s
y
non
y
mousl
y
.
1
7
Sometimes ‘nonce-formation’ is restricted to lin
g
uisticall
y
irrelevant, quirk
y
st
y
listic
‘novelties’
,
18
sometimes it is seen as full
y
representative of the s
y
stem of word-
formation defining ‘possible words’.
A
long the scale of (tacitly or ov
e
r
tly) suggested meanings fo
r
n
once-formation
I
prefer a compromise, similar to Bauer’s (1983: 45), covering both perfectly regula
r
o
utputs of productive rules as well as stylistically (or otherwise) more marked,
c
reative, even deviant ‘playful’ formations.
A
ccordingly, in Hohenhaus (1996) I propo
s
ed a scalar definition along a set of
c
riteria, includin
g
context-dependenc
y
and various t
y
pes of deviation. This is
not
to
t
sa
y
that I consider the most deviant, hi
g
hl
y
context-dependent nonce-formations to
be the most t
y
pical ones in a quantitative sense (as ‘the t
y
pe most frequentl
y
encountered’
)
,
nor
that they should be treated in the same manner as well-formed
r
formations in a theory of word-formation
.
On the contrary: those nonce-formations
t
hat display the most feature
s
are in fact the rarest
;
and deliberate deviation of course
h
as to be separated from regular outputs of ru
l
es. The latter distinction is these days
o
ft
e
n a
ssoc
iat
ed w
ith a
d
iff
e
r
e
n
ce be
t
wee
n productivit
y
vs.
creativit
y
–
cf.
Ba
uer
(2001: 62ff). An example (fro
m
T
ime
Ma
g
azine 08/10/1990, p.90) appears in the
followin
g
context:
(
2) It’s an oid-y world out there. Tabloids run factoids about humanoids on
steroids. In a world gone synthetic, why should movies offe
r
something as organic as a hero? W
e
l
come, then, to the age of the
he
r
o
i
d.
1
7
As far as the sheer number of competing terms equiva
l
ent to Engl. ‘nonce-formation’ is concerned, the
situation is even worse in German linguistics. But
A
d-hoc-Bildung
,
Augenblicksbildung
an
d
g
O
kkasionalismus seem to be the most frequent
c
hoices
(
cf. Hohenhaus 1996: 17-20
)
18
E.
g
. Lieber (1988: 206) seems to have such a n
o
ti
o
n
o
f ‘n
o
n
ce
’
w
h
e
n
s
h
e
r
e
mark
s
that
so
m
e
productively formed items “do not sound like nonce fo
r
ms” – this narrow conce
p
t of ‘nonce’ would
e
quate it with ‘creative coinings’ that are perceiv
e
d
a
s
‘
odd
’
.
E
ve
n m
o
r
ee
xtr
e
m
euseso
f ‘n
o
n
ce wo
r
d
’
c
an be encountered in psycholinguistic pap
e
r
s (cf. e.g. <www.speech.psychol.ucl.ac.uk
/
sr.savagelieven.final.doc>) where it is equivalent to ‘n
o
n-
wo
r
d
’
o
r ‘n
o
n
se
n
se wo
r
d
’
.
Thi
s
i
so
f
cou
r
se
m
ost unfortunate from our perspecti
v
e
, but we can onl
y
note that we have to remain aware of the
g
reat
v
ariati
o
n in n
o
tati
o
n
a
l
terminology here.
364
PETER H
O
HENHA
US
U
nlike the formation
heroid
, which the text is commenting on (overtly as an
i
nnovation), the jocula
r
oid-
y
lacks a proper base: suffix + suffix is not a regularly
available morphological pattern in English. Of course, the formation makes
sense
–
t
h
ese
manti
c
f
u
n
c
ti
o
n
o
f -
y
-
suffixation does apply here: ‘characterized by/full of
X(es)’, here in a meta-communicative variety of that function. Morphologically,
h
owever
,
the formation is rather the result of so-called
r
ule-changing creativit
y
.
(
Heroid
(
(
, in contrast,
can
be taken to represent re
g
ular nonce word-formation.)
Similarl
y
, Bauer (2001: 206) mentions the formation of
g
reenth b
y
Walpole
s
ome 150
y
ears a
f
te
r
-
th
-su
ffixati
o
n
ce
a
sed
t
obe
a productive (available) pattern,
and he comments that such individual, irregular “innovations are viewed as creating
t
heir effect precisely because they are not standardly r
e
g
ular morphology.”
However, such creatively deviant formation
s
are comparatively rare (cf. the table
i
n Hohenhaus 1996, appendix II). Far mor
e
o
ft
e
n
e
n
cou
nt
e
r
ed
ar
e
n
o
n-
dev
iant
bu
t
c
ontext-dependent nonce-formations, such as the famous example apple-juice sea
t
used by Downing (1977) – cf. section 4.2.
The one feature that applies to all nonce
-
formations, i.e. the necessar
y
(but not
n
ecessaril
y
sufficient) condition for ‘nonce-n
e
s
s’ as such
,
is that the formation is
‘
new’ – more precisel
y
: ‘new’ in a ps
y
cholin
g
uistic sense, i.e. formed activel
y
(b
y
whatever means) by a speaker – as opposed to retrieved ready-made from his/he
r
s
torage of already existing listemes in the lexicon.
This sets nonce-formations apart from neologisms. Neologisms are not new in
th
e
a
bso
l
u
t
ese
n
se
that n
o
n
ce
-f
o
rmati
o
n
s
ar
e.
Rather, the status of neologism is the
next
stage in the life of a word, namely when it begins to be recognized as item-
t
familiar and catches on in the usa
g
e of other speakers. Neolo
g
isms are thus onl
y
new
i
n a relative sense, diachronicall
y
, from the point of view of the lexicon. The
y
s
hould therefore rather be described as ‘
y
oun
g
listemes’.
19
The problem posed by nonce-formations and neologisms for the concepts of
l
exicalization and institutionalization are thus linked to the ones they pose for the
c
once
p
t of ‘
p
ossible word’. Nonce-for
m
ati
o
n
s
ar
eso
m
ew
hat ‘in
be
t
wee
n’ a
c
t
u
al
words and possible words: once attested, i.e. having (had) physical reality, they are
c
learly not (or no longer) merely possible, but nor do they ‘exist’ in the sense of
being part of the lexicon – which is the usual understanding of the notion of ‘actual
word’. In fact, their existence is t
y
picall
y
maximall
y
short-lived: limited to a sin
g
le
1
9
Again, this terminological distinction is ultimately of a notational nature.
B
auer
(
2001: 38f
)
, however,
sees a ‘more fundamental’ problem in it, namel
y
that it is “not possible to tell at the point when a
wo
r
d
i
sco
in
ed w
h
e
th
er
i
t
w
ill t
u
rn
ou
t t
obe
a n
o
n
ce wo
r
do
r a neolo
g
ism” – so that “a term is
r
equired which is neutral with re
g
ard to the diach
r
o
nic im
p
lications that these terms have,” and he
proposes we use ‘coina
g
e’ as such a term. I choose to differ here
,
on two counts: a) ‘coina
g
e’ to me is
n
ot free from diachronic im
p
lications either, due to its connotations of intended
p
ermanence
(
a coin,
o
nce ‘minted’, doesn’t suddenly drop out of existe
n
c
e again, whereas nonce-formations typically do –
see below); and b) I’d say that at
t
he
p
oint when a word is formed it
is
a n
o
n
ce
-f
o
rmati
o
n
per
d
efinitionem, the question is only what happens next. If w
e
need a neutral cover term, why not simply
speak of ‘new formation’ for both (with systematic ambiguity): absolutely new (nonce) and relatively
new diachronically (neologism). The latter
is
,
after all, a fuzzy concept. Admittedly, though, Baue
r
(2001) only rejects the nonce vs. neologism distinction as one irrelevant to the notions of productivity
and morpholo
g
ical structure – which is probabl
yju
s
tifi
ed.
F
o
r th
e
n
o
ti
o
n
so
f l
e
xi
c
alizati
o
n an
d
i
nstitutionalization
,
however
,
the distinction is crucial.
L
EXICALIZATION AND INSTIT
U
TIONALIZATION
365
o
ccurrence only. ‘Nonce’
can
be the first stage in a longer life-span of a word but
n
eed not be – and mostly it is also the last stage.
Neologisms are also awkward because they
h
a
ve
t
obeco
n
s
i
de
r
ed
a tran
s
iti
o
nal
phenomenon – no lon
g
er a nonc
e
-
formation, but not
y
et a full
y
institutionalized
me
m
be
r
o
f th
e
l
e
xi
co
n
e
ith
e
r
.
Similar distinctions of perspective have
t
o be observed here to the ones re
g
ardin
g
‘established’ words in section 3.2 (type-familiar vs. item-familiar, institutionalized,
l
exicalized). A formation may be instituti
o
nalized in the language of one speaker’s
(
subset of a) speech community, but may be perceived by an ‘out
s
i
de
r’ li
s
t
e
n
e
r a
s
a
n
once-formation, even though it wasn’t one from the speaker’s point of view – and
v
ice versa, i.e. a speaker may form a no
n
c
e-formation
(
from his/h
e
r
perspective) but
i
t may already be part of a listener’s lexicon.
Such uncertainties ma
y
appl
y
mostl
y
to nonce-formations which have onl
y
begun the transition to the status of
neologism. A large proportion of nonce-
f
formations
,
however
,
never even make it this far.
4.2 (Non-)Lexicalizability
Quite frequently the assumption can be encountered that any new word-
formation is potentially lexicalizable, e.g. Kastovsky (1993: 6): “The output of
word-formation processes consist of lexical items. These are […] potentially listable
an
d
ar
e
in fa
c
t m
o
r
eo
ft
e
n than n
o
t int
e
g
rated into the per
m
anent vocabular
y
of a
speech communit
y
[…].”
H
owever, this common assumption is not really supported by the evidence
available. While all words that are in the lexicon are positive evidence of words
becoming listed (they must have started out as new words at some point), evidence
o
f words that may have been formed at some point but never were listed is hardly
q
uantifiable. We simply haven’t got sufficient records of all the words that neve
r
m
ade it into the lexicon. To test Kastovsky’s assumption that new words “more
o
ften than not” end up in permanent vocabular
y
, however, we ma
y
, as a first
a
pp
roximation, attem
p
t to kee
p
track of ‘
n
ew wo
r
ds
’
co
ll
ec
t
ed
in
d
i
c
ti
o
nari
es o
f
n
eolo
g
isms b
y
checkin
g
how man
y
of these end up in standard dictionaries some
years later. Kjellmer (2000: 226) summarizes the outcome of such studies as
follows: “It appears that neologisms due to semantic change have in general a much
better chance of survival than other neologisms, but that in other cases only half o
r
l
ess than half of them stay on in the language.”
H
alf or less is not ‘more often than not’ – not even for (also formally) new words
t
hat were at least temporarily already institutionalized to a degree as neologisms. As
r
e
g
ards wholl
y
new formations that didn’t even make it to the neolo
g
ism status, i.e.
proper nonce-formations, these are b
y
their v
e
ry
nature even more elusive. Specific
studies of this area of word-formation like Hohenhaus
(
1996
)
,
(
2000
)
,
(
2004
)
suggest that it is far more typical of such new formations not to become lexicalized,
n
ot even temporarily as neologisms. Again, hindsight corroborates this: out of the
c
a. 600 nonce-formations collected as the empirical basis of Hohenhaus (1996),
v
irtually none have been adopted as current permanent vocabulary items.
366
PETER H
O
HENHA
US
I
t is thus worthwhile and sensible to ask whether there are any generalizations
p
ossible about formations that can
(
and do
)
enter the lexicon and those which
apparently don’t, and whether there may be systematic reasons for this, rather than
assuming that all words are intended for the lexicon (like e.g. Motsch 1977 holds)
and ascribing it to histo
r
i
cal accident alone if they don’t make it there.
D
owning (1977) provided an early challenge to this view, exploring
c
haracteristics of new nominal compounds, includin
g
ones that have a s
y
stematic
i
mpact on lexicalizabilit
y
– in particular: permanence or fortuitousness of the
u
nderl
y
in
g
relationship between the constituents. Her well-known example of a
c
ompound lacking a suitable lexicalizable underlying relationship is apple-juice sea
t
–used
t
o
r
e
f
e
r t
o
a
se
at in fr
o
nt
o
f
w
h
i
ch a glass of apple juice was placed at a
p
articular social event
(
‘refer’ in the sense
o
f single out, not designate). This is why
D
owning (1977: 819) speaks of ‘deictic compounds’ in such cases. And it seems to
be the context-dependency of such forms that make them non-lexicalizable
.
2
0
This was further explored in Hohenh
a
us (1996), (1998), (2005b) and a general
a
a
t
heor
y
of (non-)lexicalizabilit
y
proposed. It entails an im
p
ortant conce
p
tual
d
i
s
tin
c
ti
o
n
be
t
wee
n
p
otential listem
e
(i.e. a possible word purel
y
from the point of
v
iew of the lexicon
)
and
possible word
in th
ese
n
se o
f a
wo
r
d
d
t
hat
c
an
be
f
o
rm
ed
an
d
u
sed in
p
erformance
(
as an
X
0
,
in generative terminology) irrespective of whether o
r
n
ot it could also be listed. Genericness of meaning vs. individual (local) context-
d
ependency seems to be a crucial factor, but there are also whole productive patterns
o
f formation that only yield non-lexicalizable, though perfectly regular, formations,
i
.e. ones that have no counterpa
r
t in th
e
l
e
xi
co
n at all
:
F
or instance, Bauer
(
1983: 90
)
remarks on so-calle
d
expletive in
f
ixation
(
such as
abso-blood
y
-lutel
y
) that “words produced b
y
such infixin
g
never seem to become
established.” Other such t
y
pes include what I have dubbed
d
umm
y
compounds an
d
i
dentical constituent com
p
ounds. The former are a type of highly productive text-
d
eictic compounds employing
thing
or
g
business
as virtually empty pro-forms in head
p
osition – such as
v
acation thing
,
g
reengages business
,
used to refer back to
stretches of previous co(n)text in whic
h
something was said about vacations and
g
reengages, respectively (for contexts and extensive discussion see Hohenhaus
2
0
Note that in such cases we are talking about pa
r
t
icular pairings of form and meaning. So when we
s
peak about the non-lexicalization of a
g
iven fo
r
m, this is reall
y
shorthand for sa
y
in
g
: non-
l
exicalization of this particular for
m
with this particular meaning. It
m
i
s not necessaril
y
to sa
y
that the
form as such
(p
aired with
p
erha
p
s
q
uite a different
m
eanin
g
) is in principle non-lexicalizable. We also
h
a
ve
t
o obse
r
ve
th
ed
iff
e
r
e
n
ce be
t
wee
n nami
ng
and referrin
g
here. The fact that normall
y
apple-
j
uice
seat
i
s non-existent in extra-lin
g
uistic realit
y
(see Štekauer 2002: 110
)
is not so crucial alone. In its
s
pecific context such an object
did
‘exist’ (temporarily!) – the point is rather that it didn't exist as a
d
g
enerally ‘name-worthy
C
ATE
GO
RY’ (Downing 1977: 823, emphasis in the original). While the
permanent lexicon contains permanent names, it is
no
t th
ec
a
se
that
wo
r
ds
a
r
e formed exclusively to
fill namin
g
needs. Nonce word-forma
t
ion in particular can fulfil a ran
g
e of functions quite different
from namin
g
, includin
g
purel
y
fortuitousl
y
deictic and meta-communicative ones that have nothin
g
to
d
o with permanent cate
g
ories and are thus much mo
r
e s
y
ntactic in function (cf. also Kastovsk
y
’s
1982b distinction between namin
g
and
‘
s
y
ntactic recate
g
orization’). Such functional considerations
,
h
owever, are be
y
ond the scope of this chap
t
er
(
but see Hohenhaus 1996, cha
p
ter 5
)
.
LEXICALIZATION AND INSTIT
U
TIONALIZATION
36
7
2000, 1996: 281ff
).
2
1
The latter consist of repeated identical constituents, e.g.
instant-instant
or
t
pain-pain
,
as in the following contexts:
(
3) Felix: Tim! I'd be careful. That’s instant
g
lue
y
ou’re usin
g
.
Tim
:
It’
s
n
o
t lik
e
in
s
tant-in
s
tant
.
22
A
dam: [referring to his partn
e
r
who is undergoing fertility
t
reatment] … that she’s going through all this pain. I mean not just
emotional pain, but pain-pain
.
2
3
A
gain, these serve a meta-communicative function by regularly prompting a
‘prototype-type’ interpretation: ‘(an) XX = (a) proper/real X’ (for nouns, fo
r
ad
j
ectives and verbs: ‘XX = properly/absolu
t
ely/completely X’). In a corpus-based
stud
y
(Hohenhaus 2004), it was confirmed that, while the t
y
pe is productive, its
products, the individual compounds themselves, are hapaxes. If none are established,
t
he pattern has to be re
g
arded as bein
g
exclusivel
y
a nonce t
y
pe
.
24
W
hat is im
p
ortant to stress here overall is: some new formations, even if
productively formed and used in an
X
0
position in performance, are not necessarily
p
otential listemes at the same time. What is
p
ossible ‘for kee
p
s’ in the lexicon is not
i
dentical to what is possible as a formation.
The concept of non-lexicalizability is obvious
l
y at odds with models in which all
o
utputs of word-formation first have to e
n
t
er the lexicon before being usable in
s
y
ntax (as in Halle 1973 or Štekauer 2002). In m
y
model (Hohenhaus 1996: 249)
w
ord-formation can feed directl
y
into s
y
ntax, b
y
passin
g
the lexicon alto
g
ether,
w
hich is more in line with e.
g
. Aronoff & Anshen’s (1998: 237) emphasizin
g
that,
i
nterdependent as morphology and the lexi
c
on may be, they constitute two different
sources of words. A theory of word-formation should reflect this.
4.3 What is in the (mental) lexicon and how does it
g
et there?
As we have seen above, the question of
what ends up in the lexicon can be
f
approached b
y
‘ne
g
ative elimination’, as it were, b
y
tr
y
in
g
to identif
y
what cannot
enter the lexicon. While this is still an advancement over a
pp
roaches that leave this
basically to historical chance alone, it still does not describe how items that
a
r
e
potential listemes do become actual listemes.
H
erbermann
(
1981: 325ff
)
addresses this question more directly, by proposing
specific procedures by which a newly c
r
e
ated complex lexical item can then be
promoted to the status of ‘lexeme’ via successive stages of introduction, reiteration,
2
1
Context-dependency is clearly at work here as well: e.g.
vacation thing
‘means’ in its context: ‘what
g
y
ou told me at a
g
iven time/place about (a particula
r
) vacation’. The deictic specificit
y
of reference of
d
ifferent sorts
(p
ersonal, tem
p
oral, s
p
atial
)
makes this a non-generic form-meaning pairing, and thus a
n
on-lexicalizable one in any case.
22
Fr
o
m th
e
t
e
l
ev
i
s
i
o
n
se
ri
es
H
ome Im
p
rovemen
t
,
e
p
isode 69; a transcri
p
t can be found at: <htt
p
:
//www.
hiar
c
hi
ve.co.u
k
/
s
cript.php?s=3&e=20>
23
Fr
o
m th
e
t
e
l
ev
i
s
i
o
n
se
ri
es
C
old Fee
t
, third series, episode 3, ITV, aired on 26 November 2000.
2
4
See, however, Hohenhaus
(
2005b
)
for some caveat
s
regarding the equival
e
nt phenomenon in German.
368
PETER H
O
HENHA
US
transposition to other ‘texts’ etc.; but while Herbermann’s considerations may be
plausible enough in themselves, they do not go very far in explaining what happens
to an individual speaker’s mental lexicon. They are rather speculations on how a
n
ew formation may
spread
as a new lexeme from the orig
d
i
nal speaker to a smaller o
r
l
arger speech community, i.e.
t
hey concern institutionaliza
t
ion of neologisms. But if
words enter a collective lexicon, they have to enter the individuals’ lexicons. How is
t
his achieved? Is there a specific ‘operation’
t
hat
c
r
e
at
es
n
ew
l
e
xi
c
al
e
ntri
es?
Thi
s
i
s
s
ometimes implied (cf. e.
g
. Toman 1983: 4f, 38f, Lieber 1992: 159). However,
ps
y
cholin
g
uisticall
y
more flexible approaches ma
y
be preferable; cf. for instance
Meys’ (1985: 77) speculations about “ment
a
l
traces left by productive/interpretative
occurrences” which would gradually f
acilitate listing according to frequency.
f
f
Especially if we conceive of the mental lexicon as a (neural) network (cf. Aitchison
1994: 228), this seems quite plausible, a
s
higher frequency could be understood as
s
trengthening the links between nodes
.
2
5
A
n
d
if th
e
r
e
i
s
gradual listing
,
we would of course expect to find degrees of
(
semi-
)
listedness. This would also be in line with Jackendoff
(
1997: 231, note 11
)
:
“[
There needn’t be
]
a strict cutoff in frequenc
y
between stored forms and forms
g
enerated on-line. Rather I would expect a cline in accessibilit
y
for stored forms
,
i
ncluding the usual differential between
r
ecognition (better) and recall (worse).”
A
ssessing psychological reality empirically, however, has always been a
problem. Word recognition tests and other psycholinguistic experiments may infor
m
us about whether or not, or to what degree, complex words are stored in individuals'
m
ental lexicons and whether they are st
o
r
ed
a
sw
h
o
l
e
-f
o
rm
e
ntri
es o
r
w
ith
m
orpholo
g
ical structure/deco
m
position pla
y
in
g
a role. Ho
w
ever, the experimental
evidence, as McQueen & Cutler (1998) or Bauer (2001: chapter 4) summarize it, is
s
till
so
m
ew
hat mix
ed.
Overall, theoretical models that focus on conceptual simplicity and thus allowing
o
nly for an impoverished-entry lexicon of minimal signs appear least corroborated
by the psycholinguistic findings so far, while full(er)-entry models
,
especially
netwo
r
k
models (cf. also Lipka 2002: 197ff) seem to fare better. Clearly,
k
psycholinguistics still has a long way to go until we get a fuller understanding of the
r
eal nature of the mental lexicon, but there is an emerging thread: both simple and a
s
izeable subset of complex words and word
-
forms, even includin
g
re
g
ular ones (e.
g
.
h
i
g
h-frequenc
y
past tense forms), are ment
a
l
l
y
stored. And whether or not a word-
formation is stored depends on a complex balance of factors other than (or in
addition to) idiosyncrasy, including, apart from meaning and morphological make-
up, indeed frequency (see above), but also aspects such as family size effects
(
word
r
ecognition tends to be better if the word shares a constituent with many other words
i
n the lexicon; cf. De Jong, Schreuder & Baayen 2000).
A
particularly compelling argument for storage rather than rule-governed on-line
c
omposition comes from lan
g
ua
g
e acquis
i
tion, namel
y
from the so-called
c
r
itical-
mass h
y
pothesis
: “acquirers need a large number of
words at their command before
f
t
he
y
can
g
eneralize over them” (Bauer 2001: 114). Thus a word such as oran
g
e
j
uic
e
i
s likely to be learned (stored) as a whole before it is even felt to be transparent (see
25
B
u
t f
u
rth
e
r fa
c
t
o
r
s
ha
ve
t
obeco
n
s
i
de
red – cf. Hohenhaus (1998), K
j
ellmer (2000)
LEXICALIZATION AND INSTIT
U
TIONALIZATION
369
B
auer 2001: 122). That is to say lexicaliz
a
tion (in the synchronic sense) precedes the
emergence of the means for on-line com
position. The question then is whether
m
m
words once stored should later become ‘unlisted’ again. Given that word-memory
h
as to continue throu
g
h life in an
y
case (new acron
y
ms, new loan words, new
terminolo
gy
, new names, etc.), it seems unlikel
y
that ‘unlistin
g
’ and switchin
g
to on-
l
ine (de-)composition should automaticall
y
be preferable or more economical,
although for words learned in later stages the role of morphological rules in lexical
access does seem to increase
(
cf. Bauer 2001: 212
)
.
4.4 Unpredictable & playful formations, analogy, fads, and new developments
As so often, the devil is the details of concrete cases – and
j
ust a few shall be
m
entioned briefl
y
here:
R
ec
all th
ec
a
ses o
f ‘
c
r
e
ati
ve
’
dev
iant f
o
rmati
o
n
ssuc
h a
s
oid-
y
or
g
reenth
.
It
cou
l
dbe
a
ssu
m
ed
that
suc
h
c
a
ses
r
e
main
i
ndividual ‘quirks’ or one-offs. Consider,
h
owever
,
a case such as German unka
p
uttba
r
(literally ‘un-broken-able’), which is
r
c
learly morphologically deviant because
un-
+
A
+ -
bar
is not an available pattern
r
(like for English -
able
, the base for suffixation by -
bar
has to be a transitive verb).
r
Nevertheless, the word sprung into almost instantaneous currency through an
advertising campaign in Germany in 1990 by a big American soft-drinks corporation
(of brown fizzy liquid fame) showing one of the then newly introduced plastic
bottles (of this brand) to
g
ether with the sin
g
le word unkaputtbar
.
26
Apparentl
y
this
w
a
ssu
ffi
ce
nt t
o
in
s
tit
u
ti
o
nalize the formation enou
g
h to become usable outside its
co
nt
e
xt
.
2
7
On the other hand, in wides
p
read u
s
e as it may be, it is not fully accepted
–
in fact it is a regular target for purist criticism,
28
in particular on the grounds that a
perfectly regular word for th
e
concept already existed:
un
z
e
r
b
r
echlich
.
Th
us eve
n
though its validity is contested, the deviant f
o
rmati
o
n ha
s
t
obe
tak
e
n t
obe
established, due to, at least initially, med
i
a power rather than normal spread through
a speech community. Rather it was ‘artificially’ institutionalized rapidly and despite
l
ackin
g
acceptance
.
Another
p
roblematic case for the conce
p
tion of normal institutionalization and
l
exicalization as
g
radual enrichment of vocabular
y
are sudden bursts of
m
orphological ‘fads’. One such case is/was the craze in Germany around the yea
r
2000 of forming more and more complex
w
ords roughly in anal
o
gy to the initial
mode
l
o
f
Warmduscher
(literally ‘warm-shower-er’) such as
r
Auf-dem-Schrank
-
S
taubwischer (‘on-top-of-the-wardrobe-duster’) o
r
Bei-Mami-
W
äscher
(
‘laundry-at-
Mom’s-doer’). They are all intended to mea
n
ultimately the same (i.e. naming needs
play hardly any role here): something between ‘wimp’, 'pathetically conventional o
r
26
C
f.
<
http: //www.slogans.de/
>
; Apparentl
y
there was also a televi
s
ion clip in which such a bottle was
s
hown tumbling down a staircase without suffering damage.
2
7
An Internet search engine returned over 14,000 hit
s
(in July 2004); the word is now also used in all
m
ann
e
r
o
f
co
nt
e
xt
s
in th
ese
n
se
o
f ‘unbreakable’ – including the title
o
f a recent CD by a German pop
s
inger!
2
8
Cf. for instance <htt
p
: //uleuschn
e
r.be
i
.
t-
o
nlin
e.de/
r
e
z
e
n
s
i
o
nen/ri9610szymanski.htm>, or the relevant
discussion thread in the chat forum of the purist association VDS
:
<http: //www.vds-ev.de/forum/>.
370
PETER H
O
HENHA
US
pedantic person’ or ‘do-gooder’. Some specimens clearly overstretch the limits of
wellformedness (e.g.
H
ochzeitstagdrandenker ‘anniversary-remember-it-er’
–
c
ontaining the object of the verb simultaneously in lexical and pronominal for
m
i
nternal to the compound!). However, man
y
w
ill ha
ve bee
n in
s
tit
u
ti
o
naliz
ed
f
o
r a
s
izeable speech-community (much large
r
t
han the small-group intimate varieties
r
eferred to in section 3.2 as more likely domains for such phenomena) – there were
c
lubs and popular dedicated websites indul
g
in
g
in this particular exercise in word-
pla
y
. Ultimatel
y
, few speakers will have escaped contact with the phenomenon. A
few
y
ears on, thou
g
h, the fad has lar
g
el
y
ebbed awa
y
and hardl
y
an
y
products of it
s
urvive (perhaps only the original
W
armduscher), i.e. the ‘vocabulary’ has not been
e
nriched in any permanent sense at all. De-institutionalization followed
i
nstitutionalization rapidly
.
29
Finally, consider all those new phenome
n
a in computer-mediated language (cf.
C
rystal 2001, Hohenhaus 2005a), such as
r
eduplications lik
e
nodnod
or
wavewave
used
a
sso
-
c
all
ed
emotes
in certain genres. These seem to be formed productively,
bu
t at l
e
a
s
t a
ce
rtain
subse
t
w
ill ha
ve
conventionalized to the de
g
ree of full
i
nstitutionalization. In chat and text messa
g
in
g
(SMS), where brevit
y
is imperative,
acron
y
ms and other methods of shortenin
g
are thrivin
g
,
g
ivin
g
rise to a mix of
co
n
ve
nti
o
naliz
ed
f
o
rm
ssuc
h a
s
lol
(for ‘laugh out loud’) or
l
CU
l8r
(for ‘see you
r
l
ater’) and generally applicable technique (non-lexicalized, rule-governed) – e.g. the
r
eplacement of any homophone s
y
l
lable by numbers (esp
.
1
,
2
,
4
,
8
an
d
9
).
F
urthermore there is the use of iconic signs, the best known example being the
emoticon
: –) but again there is a fuzzy cline of institutionalization: ;–) for ‘wink’
and : –( for ‘unhapp
y
’ are probabl
y
known
t
o most e-lan
g
ua
g
e users, but rare
r
c
onstructions such as :–
@
for ‘screamin
g
’ are probabl
y
onl
y
familiar to
c
omparativel
y
fewer, seasoned ‘netties’, where the
y
ma
y
even become in-
g
roup
ide
ntifi
e
r
s.
Th
ese
ar
es
till
vo
latil
e
ar
e
a
sof
language where standardization may just
f
be setting in and which need to be observed more to see what influence they may
exert on general word-formation and the lexicon.
4.5 Lexicalization be
y
ond words
So far we have onl
y
considered lexicalization and institutionalization as appl
y
in
g
t
o words – as is a
pp
ro
p
riate in a handbook of word-
f
ormation
.
However
,
a few brief
r
emarks are in
p
lace about the
q
uestion whet
he
r th
e
r
e
ar
eu
nit
so
th
e
r than
wo
r
ds
in
t
he lexicon as well. And indeed there are good arguments for such an assumption.
Early approaches, notably Weinreich (1969), proposed the integration of a
s
eparate list of idioms alongsid
e
the (word-)lexicon proper. Furthermore, the fact
t
hat larger-than-word objects ca
n
also play a role in word-formation, in particular in
phrasal compounds, lead to various approaches of integrating this, either wholesale
29
Admittedly, this is not easy to prove empirically
,
as the
p
roducts remain accessible in written records
o
f the fad. Also, quite a few items in such lis
t
s (e.
g
. <http: //www.ablachen.de/weichei.htm>
)
are
actuall
y
normal words, such as F
ah
r
schüle
r ‘learner driver’, that have lon
g
been established in thei
r
primary, neutral meaning, but were given a connotational twist simply by inclusion in such a list
alon
g
side
g
enuinel
y
new formations.