Tom Priestly 
University of Alberta 
On Derivational Productivity in Slovene with 
Notes on Lexical Frequency and Awareness of the Norm 
Raziskana sta dva izpridevnigka glagolska besedotvorna vzorca: ~inhoativno<< obrazilo -eti in 
~fakti t ivno- obrazilo i t i .  Poglavitni cilj raziskave je ugotavljanje moinosti za merjenje produktiv- 
nosti besedotvornih vzorcev. Analiza je bila izpeljana na osnovi psiholobkega testiranja, izvedene- 
ga v Ljubljani v letih 1993-94. Glede produktivnosti je avtor priSel do treh zakljutkov: prvit, pro- 
duktivnost se da opisati kot pobevni potek, vendar v dveh razlitnih pomenih izraza; drugit, obsta- 
ja velika razlika med vrstno pogostnostjo in produktivnostjo; tretjit, posebna formula omogota 
merjenje relativne produktivnosti razlitnih vzorcev, in sicer je obrazilo -iti po pritakovanju bolj 
produktivno kot e t i .  Testiran je bil tudi pojav .blokiranja-, vendar brez dokontnih rezultatov. Av- 
tor je prigel tudi do dveh stranskih zakljutkov: nekateri kvalifikatorji za glagole v SSKJ se zdijo 
glede na rezultate raziskave nepravilni in jih bo treba ponovno oceniti in po moinosti tudi empi- 
ritno utemeljiti; moino je dolotiti, kateri govorci slovengtine se najbolj zavedajo norme. Ti rezultati 
imajo pomen za Solske utitelje slovenbtine. 
Two deadjectival verbal derivative patterns - the "inchoative" -eti and the "factitive" -iti - 
were investigated. The major aim was to discover a way of measuring the productivity of derivat- 
ive patterns; the method involved psycholinguistic tests, adminstered in four schools and four uni- 
versity departments in Ljubljana in 1993-94. With regard to productivity, there are three conclus- 
ions. First, it is clear that, at least for these derivative patterns, productivity is a "cline," but in 
two distinct meanings of the term. Second, there is an enormous difference between type fre- 
quency and productivity. Third, one particular formula appears to provide a measure of the relat- 
ive productivity of different patterns, and specifically shows that - as hypothesized - i t i  is more 
productive than -eti. A test of the phenomenon known as blocking was also adminstered, with in- 
conclusive results. Incidentally, two other conclusions were reached. Several of the stylistic labels 
(kval(fikatol:ji) in the SSKJ that are provided for verbs used in the tests appear, on this evidence, 
to be incorrect; this finding may only apply to to subjects in the tests, who were mostly aged be- 
tween 16 and 24; but it is suggested that the kval(fikatol:ji deserve re-examination and, if possible, 
empirical justification. Also, it is shown that it is possible to determine which speakers of Slovene 
have a greater, and which ones have a lesser, awareness of the standard norm; this has implica- 
tions for teachers of the Slovene language in schools. 
1 Introduction 
In his Slovenska slovnica ToporiSiC (1984: 123) writes that "besedotvorne vrste 
in sredstva niso vsi enako iivi i n  pogostni." There are two different, if intercon- 
nected, phenomena involved here: productivity and frequency. The relevance of the 
latter is discussed below. First, let us define derivational productivity, using the 
words of Schultinck (1961, see Lieber 1992): "The possibility for language users to 
coin, unintentionally, a number of formations which are in principle uncountable." 
It is true that Slovene-speakers can freely coin many new derivations; and it is also 
true, as ToporiSiC points out, that there is variation in the productivity of derivation- 
4 Slovenski iezik - Slovene Linguistic Studies 2 (1999) 
a1 processes or patterns in a language like Slovene. Unfortunately, little has been 
done to analyze, or even tentatively explore, the dimensions of this variation in Slo- 
vene; most recent advances in the study of productivity have been based on evidence 
from Dutch, English and French (see, respectively, Baayen 1992, Lieber 1992, Corbin 
1987). Descriptions of Slovene derivation written for Slovenes, such as the section 
"Besedotvorje" from which the above citation comes (ToporiSiE 1984: 109-174), are 
generally uninformative about productivity, and what is imparted is usually only 
implicit: if one derivation is exemplified with more words than a second, the reader 
may infer that the first is more productive. In, e.g., the section "Izpeljava pridev- 
nikov iz predloinih zvez," we find examples of four adjectives with brez- and only 
one with vz-; this ratio presumably reflects the lesser productivity of derivations 
incorporating the second prefix. But this lack of explicit information hardly matters, 
inasmuch as the productivity of the different components is obvious to native speak- 
ers. In specialized studies of Slovene derivation there is a similar dearth of informa- 
tion on productivity: such studies have generally concentrated either on the semantic 
mechanisms involved (thus, many analyses by VidoviE-Muha, e.g., 1981, 1983), or 
single derivative affixes (e.g., Gjurin 1985); remarks about comparative productivity 
have not been appropriate to studies such as these (see Priestly 1987: 195).' I under- 
took a preliminary investigation of productivity in Slovene derivation in the early 
1990s to begin to remedy this lack of informatiom2 Two derivational processes or 
patterns [henceforward, DPs3] were selected for the investigation: the verbalizing af- 
fixes -e- and -i-, when suffixed to adjectival stems and denoting inchoativity and 
factitivity respectively, e.g.: rjav 'brown', rjaveti 'to become brown', rjaviti 'to make 
[someone or something] brown.' A glance at any dictionary of Slovene will show 
that factitive derivations from adjectives outnumber inchoative ones; the hypothesis 
' Occasionally, as in VidoviE-Muha (1983: 371), productivity is mentioned: "Skladenjske, obra- 
zilne, naglasne pa tudi predmetnopomenske lastnosti obravnavanih zloienk so torej takSne, da 
nam omogoEajo natanEno doloEitev obsega njihove tvornosti." This is however an opportunity of 
precisely specifying the 'scope of creativity' not of any one particular process, but rather of the 
components of its 'speech base' (govorna podstava). Moreover, it is the relative frequency of the 
base components that is provided, which may not be the same as their productivity, see below. 
1 express my thanks, above all, for funding this research, to the Social Sciences and Human- 
ities Research Council of Canada; and also to the following: for his untiring practical assistance, 
to Zoran StarEeviC; for their advice, to Ada VidoviE-Muha and Velemir Gjurin; for their interven- 
tions, to Minister Slavko Gaber and other personnel in the Ministrstvo za golstvo in Sport; for 
their graciously allowing me into the@ schools and classes, to principals SreEko Boinar, Janja 
Kamnikar, Mirjena Kovai: and Miha Sengalaja and the many teachers in their schools, and to 
university colleagues Miran Hladnik, Milica KaE, Stanko Klinar, Cveto Klofutar, Ales KrbavEiE, 
and France iagar; and to the hundreds of students who participated in the tests, the vast majority 
of whom did so good-humouredly and conscientiously. The students were at the following high 
schools: Srednja gola za gostinstvo in turizem; Srednja gola za farmacijo in zdravstvo; Srednja 
upravno administrativna Sola; and Gimnazija Moste, and in classes in the following departments 
and faculties of Ljubljana University: the Oddelek za germanske jezike Filozofske fakultete; Od- 
delek za farmacijo Fakultete za naravoslovje in tehnologijo; BiotehniSka fakulteta; and the Peda- 
gogka akademija. 
No assumption is made here in favour of or contrary to treating derivation as a process, or 
as a pattern, or as both; thus Schultinck's definition, which refers to "ability to coin new words," 
can be interpreted as involving mentally-stored processes, or as relying on psychological capacities 
to draw analogies from mentally-stored patterns, or as utilizing both. All my comments, descrip- 
tions and conclusions will, I suggest, be relevant for all three kinds of theoretical stance. 
T. Priestlv. On Derivational Productivitv in Slovene ... 5 
to be tested was that this preponderance reflects a relative difference in productivity 
- a possibility that may or may not be true, given the unreliability of dictionary 
evidence. 
The first decision was to not rely purely on data available in dictionaries. These 
data may appear useful, but discussions in the literature have emphasized how un- 
reliable dictionary data are in this context. There are many reasons for this unreli- 
ability; especially, (1) dictionaries vary, and what is included in and excluded from 
them often depends on previous lexicographic work andlor the preferences of the 
compilers; (2) dictionaries usually fail to list every word written in the language, 
but even if (like the Oxford English Dictionary) they approach this ideal, they can 
never list every "actually existing" word in the spoken language, however this is 
normally defined; and (3) - most important in this context, see 3.2 - although type 
frequency can (laboriously) be extracted from them, dictionaries normally omit in- 
formation about the relative token frequency of words.4 Additionally, (4) the rele- 
vance of possible but non-occurrent words for a study of productivity is difficult to 
deny, yet, obviously, no such words appear in dictionaries (see Corbin 1987: 18, Bau- 
er 1988: 62-63, Baayen 1991). 
It was therefore decided to investigate productivity by means of psycholinguistic 
testing: for if, following Schultinck, by this term we mean the ability of speakers to 
unintentionally create new lexemes, a study of Slovene productivity requires attempts 
to investigate the thought processes of speakers of Slovene. There were few models 
upon which such research could be based: and apparently none using Slavic lin- 
guistic material; hence, the choice of DPs was in part arbitrary: the selection of 
cues and of psycholinguistic testing procedures was cautious, and the methodology 
was developed in the field. The results presented below thus derive from a first foray 
into a little-developed subject-area, and should be considered tentative. 
Type frequency is the number of different lexemes in which a given item occurs; token fre- 
quency is the frequency with which that item occurs in a corpus. So, e.g., there being 64 words 
with the adverbial suffix -omal-emu listed among the approx. 40,000 words in Mader (1981), we 
may say that this suffix has a type frequency 0.16. An analysis of the frequency of occurrence of 
all of these 64 words in a corpus of Modern Slovene may be expected to yield a higher percent- 
age, given the frequency with which the word oziroma is used. 
The most suggestive study was Aronoff and Schvaneveldt (1978); these authors' technique - 
which involved the three questions, "Is this item in your vocabulary?", "Is the item an English 
word?" and "Is the item a meaningful word?" - was used as a basis for the methodology chosen 
here, but one serious shortcoming in their experiment was not copied, namely, the fact that they 
did not use the same cues for the different components of the experiment, with the result that the 
results for these different components could not be meaningfully contrasted. Another example, 
but using only one question, is Anshen and Aronoff (1981). Greenberg and Quirk (1970) emphas- 
ize the informativeness of these three questions, cf. section 3.3 below. Psycholinguistic testing has 
been uncommon in Slavic linguistics generally; the only example known to me, which relies on 
grammaticality judgments, is Vakareliyska (1994). Asking subjects for assessments of potential 
(i.e., semantically plausible) words is just one of a large number of ways of investigating product- 
ivity; another, e.g., is asking for judgments of what are clearly nonsense words, see Derwing 
(1976). 
See further in 2.1. The choice was not arbitrary in that the inchoative and factitive dead- 
ival derivations form a contrasting pair; it was arbitrary in that I could easily have chosen another 
(contrasting or non-contrasting) pair of derivational processeslpatterns. 
6 Slovenski jezik - Slovene Linguistic Studies 2 (1999) 
2 Cues and Tests 
We administered an extensive battery of psycholinguistic tests over several 
weeks in Ljubljana in 1993. One result was very suggestive (see 3.3) and I returned 
to Ljubljana for a limited follow-up test in 1994. 
2.1 Cues 
Words to be used as cues were taken from the SSKJ: 27 randomly-selected 
verbs in -eti and 27 in -iti, all being semantically inchoative and factitive, respect- 
ively, and having an adjectival base (in most instances the semantic relationship be- 
tween adjective and verb was transparent, see 5.1) To each list of "real" verbs six 
made-up verbs were added - verbs which could be derived from real Slovene adject- 
ives, but which were not glossed in the SSKJ (see however 3.2). The complete lists 
are set out in Appendix I. 
Given the importance of token frequency (see 3.2), it was judged desirable to 
select verbs that could be categorized in two groups - those occuring with greater 
and lesser token frequency as normally used by Slovene-speakers. Since no frequency 
dictionary of the language is yet available, the only clues of token frequency to be 
found in the SSKJ - indications of individual words' range of utilization - were 
used; viz., the kvalifikatorji or stylistic tags, here referred to as indicators, which 
are abbreviated and defined in the SSKJ as follows: knjii.: "beseda, ki se rabi zlasti 
v leposlovnem ali znanstvenem jeziku"; raba peia: "beseda, ki se umika iz (splo9- 
ne) rabe"; redko: "beseda, ki je sicer v splogni rabi, pa je v razmerju do sinonima 
precej redka"; star.: "beseda, ki je bila nekoE (v kaki funkciji) sploSno rabljena; da- 
nes ima arhaiEno patino"; and zastar.: "nekoE rabljena beseda; danes je v knjiinem 
jeziku mrtva." One problem was posed by dictionary entries with multiple defini- 
tions, some with and some without an indicator. As a solution, the procedure adopt- 
ed was as follows: if the first definition in the lemma for a given verb was furn- 
ished with one of these indicators, the verb was considered less frequently-occurring 
than one with no such indicator in its first definition. The inclusion of knjii .  as an 
indicator of infrequent use also requires comment, for a word may be used only in 
literarylscientific styles but yet occur frequently therein; it was considered, however, 
that mere restriction to this stylistic function indicates that a word is not normally 
used in other functions, and is thus relatively rare. It might have been possible to 
rank-order the different indicators - thus, zastar. clearly is meant to indicate ex- 
tremely low frequency - but the experiment design precluded including enough ex- 
amples for such details to be comparable. 
Here the presentation must be interrupted briefly to discuss what may be con- 
sidered a potentially awkward procedural step, namely, reliance upon a comparison 
of two different DPs. At first glance it may appear that such a comparison is inval- 
id, because the two DPs produce verbs with obviously different kinds of meaning. 
However, most of the meanings of the two kinds of Slovene verbs involved in these 
experiments differ minimally: e.g., 'to become brown' and 'to make [someone or 
something] brown' only differ in that the first is intransitive and the second transi- 
tive: in other respects, they both express a change of state from non-brownness to 
brownness. It is therefore assumed that comparisons of results for the two different 
T. Priestlv. On Derivational Productivitv in Slovene ... 7 
DPs are, generally speaking, valid. Nevertheless, it is still possible that the compari- 
sons made below are specifically invalid, because such tests depend, for their accur- 
acy, on the sample of words being representative of the whole set. As one example, 
it could have happened - all the SSKJ indicators to the contrary - that, in my se- 
lection of actually-existing verbs, I picked 27 -iti verbs which are very common, and 
27 -eti verbs which are very rare. To guard against this as far as possible, I made a 
random selection of verbs, within two clearly-defined groups - those with no indi- 
cators in the SSKJ,  and those with the specific indicators listed above. For the 1994 
tests the two sets of cues, each numbering 30, comprised verbs similar to those used 
in 1993, but many more "made-up" words were substituted for "real" words, with the 
result that 16 "made-up" -eti and 14 "made-up" -iti words were included in the two 
sets of cues; they are listed in Appendix 11. 
2.2 Tests 
The tests whose results are described here consisted of four  question^:^ 
(1) Ali se po vagem beseda nahaja v knjiini slovengCini? 
(2) Ali vi to besedo kdaj uporabljate? 
(3) Ali je ta beseda po vaSem moina in razumljiva? 
(4) Kako pogosto sami uporabljate to besedo? 
For brevity, these are referred to in the text and on the tables with the follow- 
ing labels: (1) "Standard?", (2) "Use?", (3) "Possibility?" and (4) "Frequency?". In all 
the tests described here, subjects selected and indicated a response on a five-point 
Likert scale. For questions (I), (2), and (3) the scale was "da/lahko/ne vemlmenda 
ne / sploh ne"; for question (4) the scale was "zelo pogosto / pogosto / povpreCno / red- 
ko/zelo redko." 
The subjects for the tests were secondary and post-secondary students in Ljub- 
ljana, totalling 186 at high school and 180 at uni~ersity;~ each subject provided per- 
sonal details, see 4.2. For the 1994 follow-up study I limited the tests to three 
groups of respondents, totalling 116 in all? 
In the 1993 tests, all of questions (1) - (4) were asked. In the 1994 follow-up 
study, only questions (1) and (3) were asked (for rationale, see 3.3). Apart from an- 
swering questions (1) - (4) with reference to both sets of 33 verbs selected as cues, 
the 1993 subjects performed other tasks connected both with these verbs and (as 
"distractor tasks") with completely different cues; in 1994, they performed no other 
tasks. In each successive test, in both years, the order of cues was varied to minim- 
ize any "order effect"; and before the first test with -eti verbs and in the first test 
with -iti verbs, a short explanation was provided to ensure that the subjects would 
See footnote 4 for previous use of the first three questions by linguists investigating pro- 
ductivity. 
For details, see footnote 2. 
The three groups were in the Oddelek za farmacijo Fakultete za naravoslovje in tehnologijo; 
the BiotehniSka fakulteta; and the PedagoSka akademija. 
8 Slovenski jezik - Slovene Linguistic Studies 2 (1999) 
know, from the start, that they were to make judgments about inchoative and facti- 
tive verbs." One specific sub-routine to tests involving question (2) was incorporated 
to try to investigate what is known as "blocking." This sub-routine, and its analysis, 
are described in 3.4. 
2.3 Analysis 
All the responses were scored, respectively, 5, 4, 3, 2 and 1;" this implies that 
the mean scores displayed in the tables and discussed below may be interpreted as 
follows for questions (1) through (4): 
Question I :  the higher the mean score for a verb, the more did the 
students consider that the verb belonged to Standard Slovene; 
Question 2: the higher the mean score for a verb, the more did the 
students believe that they personally used that verb; 
Question 3: the higher the mean score for a verb, the more did the 
students assess that verb as a possible and understandable verb; 
Question 4: the higher the mean score for a verb, the more fre- 
quently did the students estimate that they themselves used that 
verb. 
The results are of different kinds and provide evidence not only about 
productivity but also about other phenomena. Results with a bearing on pro- 
ductivity are described in section 3., and other results in section 4. Much more 
could be written about individual results than what follows. For example, dre- 
veneti and dreveniti received generally similar ratings for questions (I), (2) and 
(3), but very different ratings for questions (4). Why should this have happen- 
ed? 
First, however, the overall results may be inspected, i.e., the means of all the 
means. These are set out on Table I, where it can be observed - for both -eti and 
-iti derivations - that, overall, the subjects had higher average rankings on question 
(3) than on (I), and higher on (1) than on (2). This is as expected: namely, that 
speakers of the language will accept more words as "possible and understandable" 
(3) than those they believe to be "standard" (I), while they will believe that they 
themselves actually use fewer (2) words than those they believe to be "standard" (1). 
For all three questions, also, subjects ranked derivations in -iti higher than they 
ranked -eti derivations; this may reflect the higher productivity of -iti, see further 
below. 
lo Viz., respectively: "Vse sledete besede so glagoli na -eti, ki pomenijo 'postajati nekaj'. Na 
primer, beleti pomeni 'postajati bel', in bogateti pomeni 'postajati bogat'," and "Vse sledete besede 
so glagoli na -iti, ki pomenijo 'narediti, da nekaj postane drugate'. Na primer, obeliti pomeni 'na- 
rediti nekaj belo', in bogatiti pomeni 'narediti koga bogatega'." 
l1 Analysis was performed with the SPSS statistical package. I am indebted to Darrell Wil- 
liams for help in the final stages of this analysis. 
T. Priestlv. On Derivational Productivitv in Slovene ... 9 
Table I. Means of Means 
Questions: (1) "Standard?" (2) "Use?" (3) "Possible?" (4) "Frequent?" 
-eti cues 3.12 2.63 3.46 3.79 
-iti cues 3.20 2.91 3.52 3.46 
3 Results: Productivity 
3.1 Productivity as a cline 
It is frequently emphasized that productivity is a cline, i.e., is a matter not of 
"eitherlor" but of "more or less." As is not always made clear, this must be under- 
stood in two senses. First and most obviously (see Bauer 1988: 57-60), one process 
may be synchronically more productive than another; so, English -th is a much less 
productive nominalizing suffix than -ity," and in Slovene -oba is a much less pro- 
ductive nominalizing suffix than -ost. Second, what may be called the productive 
strength of any one process also varies. This variation is well illustrated in Tables 
11-111; to take just two examples from the mean responses to question (3) "Possible?" 
we see that godneti was assessed, on average, as extremely "possible and under- 
standable," while at the other end of the scale plaieti was assessed, on average, as 
well nigh impossible and incomprehensible; and the other thirty-one cues are strung 
out along the cline in between the two. The same is true for the reponses to the 
other three questions. It is clear that the subjects, in general, considered the verbal 
cues, variously, to belong more or less to Standard Slovene; to be generally used 
more or less; to be more or less possible and understandable; and they thought that 
they personally used them more or less frequently. This is, of course, what would be 
expected intuitively, but has not been demonstrated before for Slovene. 
It is even more interesting to inspect the ratings given in each set of answers 
for individual verbs. These can be seen on Tables IV-V. What is strikinn is the " 
amount of inconsistency among the four lists. No verb is ranked very high on all 
four lists; no verb (with the possible exception of pomnogoteriti] is ranked very low 
on all four. Most verbs, indeed, receive a thorough mixture of high and low rank- 
ings; it may be asked - to use the same example - why the subjects should have 
considered that godneti, while receiving a high rating for belonging to Standard 
Slovene and for comprehensibility, should be assessed as being used infrequently not 
only by the subjects themselves, but by Slovene speakers in general. Inconsistencies 
like this are, indeed, the rule. They may appear anomalous, but given the large 
number of subjects who participated in the experiments they must be accepted as 
valid, at least for speakers of Slovene in their late teens and early twentiesJ3 The 
differences in results can perhaps be explained by the semantics of the individual 
verbs and their function in the l ang~age?~  
l2 Bauer refers to -th as "unproductive." In the 1960s, however, the neologism coolth was con- 
sidered acceptable in some styles of English, and it may be that this suffix is still not totally un- 
productive. 
l3 The subjects' average ages were 18.06 (high school) and 21.33 (university classes). 
l4 The wide differences in the responses to these four questions are of interest for psycholin- 
guistic testing in general, for questions of "acceptability" and "grammaticality" are frequently 
featured in such tests, but with little attention to what the questions actually imply. 
10 Slovenski jezik - Slovene Linguistic Studies 2 (1999) 
Table 11. Mean Responses to Questions (1) through (4), -eti Verbs Ranked 
in Numerical Order 
Questions: 
(1) "Standard?" (2) "Use?" (3) "Possible?" (4) "Frequent?" 
otolsteti 
nemeti 
godneti 
iolteti 
rjaveti 
ogrobeti 
apneneti 
plaSeti 
globeti 
kasneti 
porevneti 
plitveti 
kosmateti 
prestareti 
bakreneti 
debeleti 
srepeti 
dreveneti 
mrtveti 
kruteti 
samotneti 
leneti 
jekleneti 
osmeleti 
suroveti 
pogrdeti 
goleti 
pobedasteti 
mrzleti 
medleti 
tvrsteti 
hrometi 
trdeti 
rjaveti 
medleti 
hrometi 
trdeti 
dreveneti 
tvrsteti 
debeleti 
nemeti 
jekleneti 
kosmateti 
suroveti 
kasneti 
mrtveti 
bakreneti 
srepeti 
apneneti 
pogrdeti 
plitveti 
leneti 
prestareti 
samotneti 
ogrobeti 
otolsteti 
kruteti 
goleti 
pobedasteti 
mrzleti 
iolteti 
porevneti 
globeti 
plaSeti 
osmeleti 
godneti 
godneti 
leneti 
kosmateti 
prestareti 
tvrsteti 
pogrdeti 
apneneti 
mrtveti 
osmeleti 
kruteti 
goleti 
samotneti 
pobedasteti 
suroveti 
globeti 
plitveti 
nemeti 
mrzleti 
porevneti 
debeleti 
ogrobeti 
medleti 
bakreneti 
dreveneti 
srepeti 
kasneti 
iolteti 
jekleneti 
hrometi 
otolsteti 
trdeti 
rjaveti 
plaSeti 
debeleti 4.74 
prestareti 4.68 
nemeti 4.61 
osmeleti 4.58 
apneneti 4.55 
jekleneti 4.54 
hrometi 4.46 
globeti 4.44 
srepeti 4.44 
suroveti 4.36 
rjaveti 4.34 
iolteti 4.29 
bakreneti 4.27 
kruteti 4.28 
tvrsteti 4.25 
medleti 4.25 
ogrobeti 4.17 
porevneti 4.04 
pobedasteti 3.87 
pogrdeti 3.69 
goleti 3.66 
kosmateti 3.64 
plitveti 3.59 
otolsteti 3.56 
kasneti 3.55 
mrtveti 3.34 
leneti 3.09 
trdeti 2.91 
samotneti 2.46 
plaSeti 2.44 
mrzleti 2.27 
godneti 2.08 
dreveneti 1.59 
T. Priestlv. On Derivational Productivitv in Slovene ... 11 
Table 111. Mean Responses to Questions (1) through (4), -iti Verbs Ranked 
in Numerical Order 
Questions: 
(1) "Standard?" (2) "Use?" (3) "Possible?" (4) "Frequent?" 
izljubiti 4.72 
porevniti 4.69 
tesniti 4.67 
slabotniti 4.66 
ledeniti 4.20 
mrtviti 4.15 
pootititi 4.10 
jekleniti 4.01 
sladiti 4.01 
tujiti 4.00 
onetistiti 3.98 
rjaviti 3.97 
huditi 3.92 
osmeliti 3.89 
oiiti 3.80 
srepiti 3.78 
zanoviti 3.62 
jasniti 3.55 
blazniti 3.53 
gluSiti 3.44 
ogrditi 3.29 
blaiiti 3.21 
siviti 3.20 
pomnogoteriti 3.01 
grobiti 2.95 
bahatiti 2.92 
postekleniti 2.85 
besniti 2.60 
oleseniti 2.57 
kisliti 2.47 
drobniti 2.30 
dreveniti 2.18 
leniti 2.02 
ledeniti 4.73 
slabotniti 4.72 
sladiti 4.66 
blaiiti 4.60 
tujiti 4.17 
onetistiti 4.08 
besniti 3.85 
rjaviti 3.76 
gluSiti 3.72 
srepiti 3.51 
blazniti 3.48 
pootititi 3.43 
kisliti 3.33 
oiiti 3.32 
zanoviti 3.28 
dreveniti 3.25 
jekleniti 3.25 
grobiti 3.14 
porevniti 3.14 
leniti 2.98 
ogrditi 2.86 
tesniti 2.83 
bahatiti 2.79 
drobniti 2.73 
jasniti 2.71 
mrtviti 2.49 
postekleniti 2.44 
huditi 2.37 
pomnogoteriti 2.34 
oleseniti 2.00 
osmeliti 1.95 
izljubiti 1.91 
siviti 1.85 
blaiiti 4.60 
jasniti 4.41 
sladiti 4.32 
tesniti 4.31 
onetistiti 3.84 
oiiti 3.55 
ledeniti 3.53 
oleseniti 3.35 
rjaviti 3.33 
besniti 3.28 
postekleniti 3.24 
leniti 3.24 
siviti 3.13 
tujiti 2.92 
izljubiti 2.90 
dreveniti 2.89 
gluSiti 2.85 
blazniti 2.84 
kisliti 2.84 
jekleniti 2.76 
mrtviti 2.74 
slabotniti 2.50 
drobniti 2.47 
porevniti 2.41 
ogrditi 2.33 
srepiti 2.32 
huditi 2.24 
pomnogoteriti 2.21 
grobiti 2.20 
bahatiti 2.05 
pootititi 1.80 
osmeliti 1.32 
zanoviti 1.31 
rjaviti 4.91 
ogrditi 4.75 
grobiti 4.69 
izljubiti 4.69 
gluSiti 4.61 
blaiiti 4.53 
blazniti 4.35 
bahatiti 4.27 
postekleniti 4.27 
dreveniti 4.19 
drobniti 4.15 
siviti 4.00 
sladiti 3.77 
slabotniti 3.75 
tesniti 3.73 
zanoviti 3.69 
srepiti 3.68 
kisliti 3.61 
huditi 3.55 
porevniti 3.48 
jasniti 3.31 
jekleniti 3.31 
tujiti 3.23 
leniti 2.91 
osmeliti 2.91 
ledeniti 2.79 
mrtviti 2.72 
besniti 2.63 
oleseniti 1.72 
pootititi 1.63 
pomnogoteriti 1.51 
oiiti 1.51 
onetistiti 1.29 
12 Slovenski iezik - Slovene Linguistic Studies 2 (1999) 
Table IV. Mean Responses to Questions (1) through (4), -eti Verbs Ranked 
Alphabetically According to Dictionary Entry 
Questions: 
(1) "Standard?" (2) "Use?" (3) "Possible?" (4) "Frequent?" 
1. verbs with no kvalifikatorji 
dreveneti 
hrometi 
medleti 
mrtveti 
nemeti 
prestareti 
rjaveti 
trdeti 
2. verbs with kvalifikator (k) 
apneneti 3.99 
bakreneti 3.17 
kasneti 3.67 
plitveti 3.26 
pobedasteti 2.37 
pogrdeti 2.41 
srepeti 2.91 
suroveti 2.43 
3. verbs with other kvalifikatorji 
Evrsteti (r) 
debeleti (r) 
globeti (z) 
godneti (r) 
goleti (r) 
jekleneti (r) 
kosmateti (r) 
mrzleti (k,r) 
ogrobeti (k,r) 
otolsteti (k,r) 
iolteti (s) 
4. "made-up" verbs 
kruteti 2.62 2.12 
leneti 2.57 2.30 
osmeleti 2.49 1.44 
plaSeti 3.85 1.50 
porevneti 3.42 1.82 
samotneti 2.58 2.24 
T. Priestlv. On Derivational Productivitv in Slovene ... 13 
Table V. Mean Responses to Questions (1) through (4), -iti Verbs Ranked 
Alphabetically According to Dictionary Entry 
Questions: 
(1) "Standard?" (2) "Use?" (3) "Possible?" (4) "Frequent?" 
1. verbs with no kvalifikatorji 
dreveniti 
gluSiti 
jasniti 
mrtviti 
ogrditi 
onetistiti 
oiiti 
rjaviti 
sladiti 
tesniti 
zanoviti 
2. verbs with kvalifikator (k) 
blazniti 3.53 3.48 2.84 4.35 
jekleniti 4.01 3.25 2.76 3.31 
kisliti 2.47 3.33 2.84 3.61 
leniti 2.02 2.98 3.24 2.91 
siviti 3.20 1.85 3.13 4.00 
3. verbs with other kvalifikatorji 
bahatiti (r) 2.95 
besniti (r) 2.60 
blaiiti (r) 3.21 
huditi (z) 3.92 
izljubiti (k,r) 4.72 
ledeniti (k,r) 4.20 
osmeliti (z) 3.89 
pomnogoteriti (s) 3.01 
pootititi (z) 4.10 
postekleniti (k,r) 2.85 
tujiti (k,r) 4.00 
4. "made-up" verbs 
drobniti 2.30 2.73 2.47 4.15 
grobiti 2.95 3.14 2.20 4.69 
oleseniti 2.57 2.00 3.35 1.72 
porevniti 4.69 3.14 2.41 3.48 
slabotniti 4.66 4.72 2.50 3.75 
srepiti 3.78 3.51 2.31 3.68 
14 Slovenski jezik - Slovene Linguistic Studies 2 (1999) 
3.2 The relevance of frequency 
The importance of the relative frequency of DPs in speakers' lexicons is unde- 
niable: if a derivative pattern is productive, it will probably have been productive for 
some time, and this will be mirrored in dictionaries and in speakers' own lexica. 
Moreover, the existence of larger and smaller stocks of words that utilize different 
derivative patterns must in turn have some influence on these patterns' productivity. 
Over two decades ago Aronoff (1976: 35) emphasized that type frequency of words 
based on a given DP did not accurately reflect the productivity of that DP. The sit- 
uation in Slovene, where the number of factitively derivated verbs appears to be 
greater than that of inchoatively derived verbs (see above), may merely be an acci- 
dent of history that does not reflect synchronic psycholinguistic processes. It may 
rather (or, also) be due to constraints (see 3.4) on the applicability of the DPs, 
whereby (and I speculate!) -e- may be fully productive within the limitations set by 
these constraints, whereas -i- is only partly productive while yet being limited by 
fewer constraints; which could perhaps be why -iti verbs occur more frequently in 
dictionaries. More recently, it has been very convincingly argued (see Baayen and 
Lieber 1991, Baayen 1993) that token frequency can provide a more accurate meas- 
ure: if a process is very productive, it will be used that more frequently to "coin 
new formations," and so there will be more instances of new words being thus 
coined on the spur of the moment and used relatively seldom. Baayen and Lieber go 
so far, indeed, as to develop a measure which requires a count of hapax legomena 
(derivations that are used only once) in a language corpus. - A large corpus of the 
kind that they use1' is, however, not available for Slovene; hence my recourse to 
classifying the verbs separately, depending on the provision of indicators in the 
SSKJ, see above. On Tables IV-V the mean scores are displayed in four groups: 
verbs without indicators in the SSKJ, as defined above; verbs labelled as "bookish"; 
verbs with other indicators; and made-up verbs. Let us inspect the mean responses 
in the fourth column of each of these tables, the results for responses to question 
(4). When we contrast the first three groups on the one hand with the fourth on the 
other, i.e., the results for the 27 "real" verbs with those for the 6 "made-up" verbs, 
it is clear that the subjects were of the opinion, in general, that they used most of 
the "made-up" verbs with considerable frequency. 
Now it must be admitted that just because these verbs are not in the SSKJ it 
does not mean that they have not been used before and are not thus "occurrent" in 
Slovene; indeed, although I did literally devise the verbs, there is evidence of previ- 
ous use for some of them: of the six "made-up" verbs in -eti, one is in Pletergnik 
(1894), namely leneti 'faulenzen'; and possible related derivatives for two more, viz. 
osmeljevati and plaienje, are in Besediite 1987; while of the six in -iti, three are 
in Besediite 1987, namely grobiti, oleseniti, srepiti. (Since the Besediite does not 
gloss its forms, these comments are speculative, for it is not clear whether these 
words have inchoativelfactitive, or other meanings.) Still, the ratings for other 
"made-up" verbs are high - especially those for kruteti and also porevneti (Table 
IV) and drobniti (Table V), and I have found no published evidence of these verbs' 
existence. 
l5 Their corpus is the 'Celex database' of texts comprising over 18 million English words 
(Baayen and Lieber 1991: 803). 
T. Priestlv. On Derivational Productivitv in Slovene ... 15 
Let us now reconsider questions (1) and (3). A response to "Ali se po vagem 
beseda nahaja v knjiini slovengEini?" shows whether the subject thinks the verb 
belongs to Standard Literary Slovene - i.e., that it will be used on formal occasions 
and in the official media, and that it will be listed in official dictionaries. 
Responses, on the other hand, to "Ali je ta beseda po vagem moina in razumljiva?" 
constitute a direct testimony to the subjects' appreciation of the verb's availability to 
their own selves. Question (3), then, provides the best first indication of the product- 
ivity of the DPs; for if productivity is (see above) "the possibility for language users 
to coin, unintentionally, a number of formations," these formations must necessarily 
be both "possible" and "understandable." - When we constrast the results for ques- 
tions (3) and (4), we see how very different they are. For example, on Table IV, 
both rjaveti and hrometi are rated as not very possible and understandable, but the 
subjects considered that they used them very frequently. These results emphasize 
how very different are productivity and frequency. 
3.3 Towards a Test for Productivity 
Let us consider again questions (1) through (3). Greenbaum and Quirk (1970) 
argue that answers to these three questions will reflect three different kinds of 
evaluation, namely what they term, respectively, (1) "recognition of a precept," (2) 
"belief about own use," and (3) "willingness to tolerate." Of these, the first reflects 
subjects' assessments of the standard grammatical prescriptions that are recognized 
in the linguistic community; the second reflects their assessments of their own 
active use of the language; and the third reflects their assessments of their passive 
acceptance of linguistic usage. 
The first parameter - "recognition of a precept" - will depend enormously on 
the existence in the linguistic commmunity of recognized precepts, their prestige in 
the community, and the extent to which they are known in that community. I sug- 
gest that, for example, Slovenes' attitudes to Standard Slovene will be different from 
the attitudes of North Americans or of Britons to Standard English, and also that 
these different linguistic communities will not all be equally aware of the respective 
standardized norms. The results described for the Slovene respondents in 4.2 may 
be very different from comparable results that could be obtained in Anglophone 
communities. 
Here, however, we will explore a proposed test for productivity which utilizes 
the means of responses to questions (3) and (2), which are abbreviated here as p 
(for "possible") and u (for "used") respectively. This is based on an analysis which 
has, as its point of departure, the assumption that, normally, p judgements will be 
higher than u judgements: i.e., that in speakers' opinions they understand more 
words than they actually use - that their passive vocabulary is larger than their 
active vocabulary. The approach reflects the finding by Baayen (see 3.2) that more 
productive derivative processes result in more hapax legomena, that is, more words 
glossed only once in a corpus - more types occurring with only one token - and 
less productive derivative processes result in fewer such words; or, more generally, 
that "unproductive affixes show a large proportion of high-frequency types and ... 
productive ones a much lower proportion" (Lieber 1992: 4). 
16 Slovenski iezik - Slovene Linguistic Studies 2 (1999) 
With respect to real versus made-up words, I interpret this as follows. For real 
words, the difference between p-ratings and u-ratings for our two DPs may be sig- 
nificant, but will not be very marked, because we are dealing with words which 
speakers may well have used and most of which will therefore tend to be in their 
active vocabularies. This difference can be expected to be greater for the (presumed- 
ly) more productive suffix -iti, and smaller for the (presumedly) less productive suf- 
fix -eti, because speakers will tend to consider more of their "passive vocabulary" 
words with the former suffix as "possible and understandable" - hapax legomena 
with this suffix occur more often! - but this will not hold so true for the latter. 
For made-up words, on the other hand (to the extent that they are truly 
"made-up," see 3.2 and below), and for both DPs, the u-ratings will be much low- 
er, both in relation to the same scale for real words, and in relation to the p-ratings 
for made-up words. The important point is the following. If a DP is more produc- 
tive, its "use" rating will not be quite as low, because speakers will be more accust- 
omed to words formed with that derivative process, and the made-up words will 
sound more plausible; hence, the "use" rating for made-up words with the more pro- 
ductive derivative process will be closer to the "possibility" rating for the same 
words. The difference between the two ratings will however be greater for a less 
productive (and therefore less familiar) DP. 
This analysis therefore involved calculations of p and u separately for "real 
words" and "made-up words." On Table VI are displayed the calculations according 
to the formula (p  - u )  t u ,  which expresses the amount by which (on average) the 
responses to Question 3 exceed the responses to Question 1 in our experiments in 
Ljubljana in 1993. As is apparent, for real words, the figure for -iti verbs is only 
slightly higher than for -eti verbs. In contrast, the figure for made-up words is 
much higher for -eti verbs than for -iti verbs. 
Table VI: Productivity Formula as Applied to  Results from Tests in  Ljub- 
ljana 1993 
Actual Words Made-up Words 
In other words: where real words are concerned, the difference between p-ratings 
and u-ratings is greater for -iti, which is therefore the more productive suffix. 
Where made-up words are concerned, on the other hand, the difference between 
p-ratings and u-ratings is greater for -eti, which is therefore the less productive suf- 
fix. 
This is a very neat result; perhaps too neat. There are three problems that 
have to be taken into consideration, and which may affect either the analytical 
approach, or its specific applications, or both. First, one must always be aware of the 
co-occurrence of the constraint known as blocking, see 3.4 below; to the extent 
that this constraint operates, it will affect less productive DPs more than more pro- 
ductive ones: it has been suggested, even, that totally productive DPs can never be 
T. Priestlv. On Derivational Productivitv in Slovene ... 17 
blocked. Moreover, blocking will affect the "use" evaluation more than the "possibil- 
ity" evaluation. Second, the "made-up" words numbered only six, which makes the 
result relatively unreliable. Third, I risked "making up" words which do in fact oc- 
cur, even if they are not glossed in the SSKJ; and, as shown above, I may indeed 
have managed to "devise" words that already existed. To the extent that this actually 
happened, real words will have been, by chance, included in the list of (so-called) 
"made-up" ones. The words glossed by Pletergnik and in the Besediite could of 
course have been omitted from analysis, but even if I were sure of their meaning 
(and thus of their status as inchoative or factitive derivations) this would leave too 
small number of "made-up" words for a useful c~mparison!~ 
Nevertheless, the results as displayed in Table VI were encouraging. It was for 
this reason that I decided to replicate the tests with questions (3) and (2) in Ljub- 
ljana in 1994. The results were as on Table VII. They strongly support our hypothes- 
is. I suggest that the numerical imbalances between the figures on Tables VI and 
VII may derive from the different proportions of "made-up" words in the 1993 and 
1994 tests, and perhaps from the different influence of specific "blocking" factors. 
Table VII. Productivity Formula as Applied to Results from Tests in Ljub- 
ljana in 1994 
Actual Words Made-up Words 
Nevertheless, the reservations listed above still hold. To prove the reliability of 
the (p - u) t u measure, it will be necessary to replicate these tests using very many 
more cues, including (as in the 1994 test) a greater proportion of "made-up" words. 
3.4 "Blocking" 
It is acknowledged that one major obstacle to progress in research into deriva- 
tional productivity is the lack of a proper understanding of the forces that impede 
derivation. Even very productive DPs do not apply to every single form,'7 and 
enough is known of the obstacles to derivation for it to be clear that they are of 
great significance. There are several kinds of obstacles, both formal (e.g. Slovene 
-ski can only be suffixed to stems ending in consonants; Slavic *-ba could not be 
suffixed to stems ending in labials, see Browne (1981)) and semantic. Among sem- 
antic obstacles is what is now usually called bl~cking!~ An often-cited example is 
the English deverbal suffix -er which, though productive, is not normally affixed to, 
e.g., type to produce a derivative *typer meaning "person who types"; it is argued 
l6 Another solution - asking native speakers if these words existed - was an integral part of 
the experiment, and was therefore not resorted to a priori. 
l7 This statement applies to the languages in which DPs have been studied, which are appar- 
ently all Indo-European and typologically similar. 
l8 The idea is not new: Hermann Paul identified it in 1896 (see Reiner 1988: 155). The term 
blocking was proposed by Aronoff (1976); a less used but more precise term is pre-emption. 
18 Slovenski jezik - Slovene Linguistic Studies 2 (1999) 
that the prior existence of the word typist pre-empts, or blocks, this derivation; in 
Slovene, similarly, one could argue that the derivation of "tipkalec is "blocked" by 
the prior existence of tipkar. In the current state of uncertainty about bl~cking,'~ 
one theme keeps recurring: namely, that there should be some link, perhaps even 
some correlation, between blocking and frequency. Thus Aronoff (1976: 43), "there 
is a ... direct connection between lexical listing and ... blocking"; and at the very 
least it is likely that the more productive a DP is, the more hapax legomena will be 
derived, hence the less blocking will occur. For these reasons, a sub-routine was de- 
vised to investigate blocking: 
Some of the students - a total of 123 in all - who completed the questionnaire 
with question (2) "Use?" were asked to do a supplementary task. Each, as with all 
these tests, circled one of the five responses with respect to "Use?" If they had cir- 
cled c ,  d or e - i.e., if it was their judgment that they did not use the word very 
much or at all - then they were asked to write down which word they preferred in- 
stead of that verb.20 The analysis of these "write-in" parts of the test was conducted 
as follows. First, the "written-in" forms were classified as valid replacements, or re- 
jected, using four criteria: (1) Wherever subjects had circled a or b, their 
"written-in" forms were ignored, however interesting they seemed: after all, if a sub- 
ject said they used a particular verb, I could not accept that its use was "blocked". 
(2) What may be called "empty" aspectual counterparts of verbs were ignored. Ex- 
ample: responses from the 4 subjects who said they did not use goleti and who wrote 
in ogoleti as the preferred alternative were ignored, since these two forms must be 
considered as belonging to the same verb: in the SSKJ goleti = postajati gol; ogoleti 
= postati gol. This information may be of interest to aspectologists, but is irrelevant 
to studies of "blocking." (3) When subjects clearly did not understand either the in- 
structions, or the verbal cue, their answers were ignored. Examples: godrnjati 'to 
grumble' for godneti 'to become mature' (here, the instructions were presumably 
misunderstood); and ne imeti vet rad 'not to love any longer' for izljubiti 'to love to 
an extreme' (here the cue was obviously unfamiliar and was misinterpreted). (4) A 
valid replacement had to be semantically similar to the form which it was replacing, 
at least to the extent that it could be considered a potential synonym in some norm- 
al context. This criterion was interpreted liberally; so, e.g., two suggested replace- 
ments for globeti 'to sink' were considered valid: globiti se, whose normal meaning 
is 'to become more sensitive', was accepted as valid because it could be interpreted 
literally as 'to make oneself deeper'; and nizati se 'to lower oneself' was accepted 
because 'lowering oneself' would be synonymous with 'sinking' if one were in water. 
Phrases were included: e.g., dobivati bakreni lesk 'to get a coppery sheen' was con- 
sidered a valid replacement for bakreneti 'to become coppery.' Next, replacements 
were classified in two groups: single words (whereby reflexive verbs were counted as 
one-word units); and replacements of more than one word, i.e., "paraphrases." Ex- 
l9 For this uncertainty, compare Marle (1985: 63) who calls for a systematic investigation of 
"impeding forces"; Di Sciullo and Williams (1987: 13), "it remains a mystery what blocking really 
is"; and Bauer (1988: 67), "it rather looks as if blocking is not strictly relatable to derivational 
morphology." See also Rainer 1988. 
20 The instruction: "Ce ste obkroiili c, d ali e, napigite besedo ali frazo, ki jo bi rajgi uporab- 
ljali." 
T. Priestlv. On Derivational Productivitv in Slovene ... 19 
amples: for debeleti 'to become fat' - one-word synonyms: debeliti se, rediti se, zre- 
diti se; paraphrases: postajati debel 'to become fat' and dobivati meso 'to gain flesh'. 
The results are set out in Table VIII. In the left-hand column are the cues, list- 
ed in numerical order of means as on Table 11. In column A the numeral shows the 
number of times a single synonymous word was entered by the students. In column 
B is a numeral showing the number of times a synonymous paraphrase was suggest- 
ed. Since "blocking", to the extent that it exists and operates, involves the restrictive 
influence of words and not of paraphrases (otherwise, few if any new words would 
ever be derived!), I considered single synonyms as potential examples of "blocking," 
and paraphrases as instances where "blocking" was (for an individual respondent) 
inoperative. 
The table suggests several conclusions, chief among which (unfortunately) is 
that if there is a pattern to "blocking," it is not made evident by this test. It is clear 
that some common words have potential "blocking" synonyms, and some uncommon 
words do not; it is clear that frequency of use and "blocking" are rather vaguely 
correlated; but neither of these conclusions is new. The results are not uninformat- 
ive, however. It should be noted that several of the "made-up" words were under- 
stood as inchoatives or factitives, respectively, by most of the repondents: for in- 
stance, 59 students (48 of the total), nearly all those who rejected the derivative 
leneti, wrote in words and paraphrases which conveyed the idea of "becoming lazy," 
(e.g., polenariti, polenjati se, bolj je Zen, lena sem ko fuks); see also the figures 
for samotneti, kruteti, porevneti, plaieti, drobniti; on the other hand, slabotniti, po- 
revniti, srepiti and grobiti were either not understood as factitives, or if they were, 
the task of devising a synonymous expression was too difficult for the respondents. 
Moreover, whereas the derivation of leneti can be considered "blocked" by the prior 
existence of such words as polenariti, the same can definitely not be said of the de- 
rivation of kruteti: the verb was rejected by a majority of respondents, but only one 
thought of a synonymous word (krutiti se) whereas 49 suggested paraphrases (otrdeti 
v srcu, biti krutejii, biti okruten, etc.). If this particular test is replicated with a 
large number of "made-up" words, it should be very informative about the role 
played by "blocking," for it should become clear to what extent there are already-ex- 
isting words which pre-empt the role of potential derivatives. 
One other matter of interest, deriving from Table VIII, is mentioned in 4.1. 
4 Other Results 
Two more results, both incidental to the question of productivity, deserve men- 
tion. 
4.1 Indicators of frequency in the SSKJ 
In addition to the comments on frequency in 3.2, above, two other important 
points should be made: one is apparent from Tables IV-V, the second from Table 
VIII. First, note that on Table I we see that the subjects tended to rate their own use 
of verbs (question (4)), in general, as more rather than less frequent: the mean for 
all 33 -eti verbs is 3.79, and the mean for all 33 -iti verbs is 3.46, in both instances 
much higher than the "povpreEno" rating which was scored at 3.00. When we com- 
pare, on Table IV, means for question (4) on the one hand for verbs without indi- 
20 Slovenski iezik - Slovene Linguistic Studies 2 (1999) 
Table VIII. "Replacements" for Verbal Cues Rated "Not Used" 
A = number of one-word replacements (tokens) 
B = number of paraphrase-type replacements (tokens) 
in bold: verbs marked "redko" in the SSKJ 
in italics: "made-up" words 
-eti verbs -iti verbs 
rjaveti 
medleti 
hrometi 
trdeti 
dreveneti 
Evrsteti 
debeleti 
nemeti 
jekleneti 
suroveti 
kosmateti 
kasneti 
mrtveti 
bakreneti 
srepeti 
apneneti 
pogrdeti 
plitveti 
leneti 
prestareti 
samotneti 
ogrbeti 
otolsteti 
kruteti 
goleti 
pobedasteti 
mrzleti 
iolteti 
porevneti 
globeti 
plafeti 
osmeleti 
godneti 
blaiiti 4.60 1 0 
jasniti 4.41 0 0 
sladiti 4.32 6 0 
tesniti 4.31 0 0 
one8stiti 3.84 9 0 
oiiti 3.55 1 0 
ledeniti 3.53 1 0 
oleseniti 3.35 0 0 
rjaviti 3.33 0 1 
besniti 3.28 2 3 
leniti 3.24 0 1 
postekleniti 3.24 3 1 
siviti 3.13 0 0 
tujiti 2.92 1 1 
izljubiti 2.90 11 1 
dreveniti 2.89 1 1 
gluSiti 2.85 2 4 
blazniti 2.84 2 5 
kisliti 2.82 17 1 
jekeleniti 2.76 0 3 
mrtviti 2.74 7 0 
slabotniti 2.50 7 0 
drobniti 2.47 25 2 
porevniti 2.41 8 3 
ogrditi 2.33 10 3 
srepiti 2.32 0 2 
huditi 2.24 23 4 
pomnogoteriti 2.21 37 0 
grobiti 2.20 0 1 
bahatiti 2.05 58 3 
poo8ti 1.80 20 2 
osmeliti 1.32 1 1 
zanoviti 1.31 16 5 
T. Priestlv. On Derivational Productivitv in Slovene ... 21 
cators and on the other for verbs with the indicators "redk.," "star." and "zastar." in 
the SSKJ, we see that the majority of the verbs in the latter category receive high 
frequency ratings (even with an adjustment to allow for the subjects' high ratings 
just described). These include the "rare" debeleti and jekleneti, the "bookish and 
rare" ogrobeti, the "archaic" iolteti, and, most surprisingly, given that the definiton 
of the indicator includes the words "danes je ... mrtva," the "obsolete" globeti. In 
addition, this evidence seems to suggest that dreveneti should be marked "rare" in 
the SSKJ. 
The same comparison on Table V (with, as previously, an adjustment) shows a 
similar picture: high average ratings are given to blaiiti and bahatiti (both "rare"), 
to blazniti and siviti (both "bookish"), and to postekleniti and izljubiti (both "rare" 
and "bookish"). In addition, this evidence suggests that oiiti and oneEistiti should be 
furnished with indicators in the SSKJ. The results summarized in this and the pre- 
ceding paragraph provide very strong grounds for suggesting that one of two facts 
must obtain: either the indicators in the SSKJ (at least for these verbs, but - given 
their random selection - probably for others) are inaccurate, and require revision; 
or, at best, that these indicators reflect the language of older generations, not that 
of the students who took these tests?' These results call, at the very least, for furth- 
er empirical research into the "rarity," the "archaicness" and the "obsoleteness" of 
words in the SSKJ?' 
For the second point, we recall that the definition of the kvalifikator "redko" 
in the SSKJ is, "beseda, ki je sicer v splogni rabi, pa je v razmerju do sinonima 
precej redka." On Table VIII, verbs with the indicator "redko" are shown in bold 
type. It can be seen that 58 of the student repondents wrote in single-word syno- 
nyms for bahatiti - from a total of 123 students - and the number of students who 
could think of synonyms for the other "rare" verbs was 38 (mrzleti), 18 (debeleti], 
and 14 (otolsteti); for the other 12 verbs, less than 10 of the 123 students could think 
of synonyms. In some instances, they found it easy to express the meaning of the 
verb - thus, 36 students wrote in a paraphrase for kosmateti - but, overall, very 
few of the "rare" words did in fact have synonyms which were easily brought to 
mind. Again, some re-examination of the indicator "redko" is suggested. 
4.2 Awareness of Norm 
It was argued, prior to administering these tests, that question (1) - Ali se po 
vagem beseda nahaja v knjiini slovengEini? - might be informative in quite a dif- 
ferent way; specifically, that answers to this question should be assumed to reflect 
an awareness of the norm. If some subjects rated verbs higher on this scale than 
did other subjects, they might be said to show a greater self-assuredness in their 
judgments as to which verbs belong more, and which belong less, to Standard Liter- 
ary Slovene. This was not assumed to mean that their judgments would necessarily 
be more reliable (although this might also be true); but it would indicate, at least, 
that they believed that their judgments were more reliable. It was therefore hypo- 
thesized that various extralinguistic factors might affect the responses to question (1) 
For average ages, see footnote 13. 
'' Re-evaluation of ,ne other subset of indicators (the 'expressive' ones) for the SSKJ is pro- 
posed and described in Zele 1993. 
22 Slovenski jezik - Slovene Linguistic Studies 2 (1999) 
- factors having to do with the amount to which individual speakers had been ex- 
posed to Standard Literary Slovene during their lives to date, which exposure would 
influence their self-confidence. 
Of the personal details collected from respondents, the following were used for 
analysis: (1) sex, (2) current level of education, (3) being in a "language-oriented" 
educational stream or not, (4-5) recent scholastic level of attainment, (6-8) place of 
longest residence of self and parents, (9-10) education of parents, and (11-12) socio- 
-economic status of parents. For (2) data from secondary school students were tabu- 
lated separately from those from university students. For (3) subjects were divided 
into groups according to their specialization: those studying at the Upravno-admini- 
strativna iola, the Gimnazija, the Filozofska fakulteta and the Pedagogka akademija 
were classified as language-oriented, and those in the others as non-language, i.e., as 
technically, ~riented.'~ For (4) and (5) subjects reported their most recent average 
grades, respectively, in language classes and in all subjects; these were divided into 
"lower" and "higher" grades. For (6) through (8) subjects listed their own and their 
parents' places of residence during their lives; the place where each spent most of 
their linguistically formative years, i.e., from age 1.6 through late teens, was taken 
as indicative of dialect background, whereby Ljubljana, Upper and Lower Carniola 
were somewhat arbitrarily counted as "central" and all other areas within Slovenia 
as "~eripheral."~~ For (9) and (10) they reported whether their parents had finished 
primary, secondary, or post-secondary school; figures for the first and third of 
these subsets were contrasted. And finally for (11) and (12) subjects reported the 
occupation of each parent; these were ranked according to the Pineo Scale (see 
Pineo, Porter and McRoberts 1977) as interpreted for occupational categories in Slov- 
enia; those in the upper half of the scale (18-10) were contrasted with those in the 
lower half (9-1). 
On each of Tables IX through XI11 mean scores for responses to question (1) 
are displayed, for each of -eti and -iti, and for all verbs except the six made-up 
words in each instance (which were omitted from this analysis since they may not 
belong to the norm). In the third row of each section of the tables is an expression 
of the difference in responses from the two subsets of respondents, viz., the differ- 
ence between the two means, expressed as a percentage of the lower mean. Thus, in 
Table IX, the third row of the first section reads "university : high school ... [eti] 
8.1 ... [iti] 4.8"; this is to be read as "for -eti verbs, the mean for those attending 
university was 8.1 higher than the mean for those attending high school; for -iti 
verbs, the former was 4.8 higher than the latter." It will be noted that in only one 
instance (the "total grades" figures) do the indications for -eti verbs on the one 
hand, and for -iti verbs on the other, disagree; this general agreement shows, surely, 
the strength of the conclusions listed below. It will also be noted that the indications 
for -iti verbs are, in many but not all cases, much higher than those for -eti verbs. 
This result requires further study; is it linked to the hypothesized and - hopefully - 
demonstrated higher productivity of -iti? 
Under the general heading "educational and intellectual development" we con- 
23 Full details of educational institutions in note 2. 
24 Subjects who reported their own, their mothers' or their fathers' formative places of resid- 
ence as being outside Slovenia were omitted from the relevant calculations. 
T. Priestlv. On Derivational Productivitv in Slovene ... 23 
sider the following results (see Table IX). Students at university had higher mean 
scores than students in high schools; students with "language-oriented" specializa- 
tion, here labelled "humanities," had higher mean scores than "technical" students. 
Those who had higher grades in the previous year tended to have higher scores than 
those who had lower ones. Of these factors, it is striking that the most influential 
appears to be the orientation toward or away from the use of language. Three par- 
ameters can be examined under the heading "dialect influence," as on Table X. In 
all three instances, residence in central Slovenia for the subjects themselves and for 
their parents correlates with high assessments, as compared with residence in more 
peripheral areas of the country. Interestingly, the places where the subjects' parents 
grew up is of more importance than the places where the subjects themselves did so; 
and, in addition, the mother's place of residence is more influential than that of the 
father. 
Third, we consider the education of the parents (Table XI). Subjects whose par- 
ents had graduated from higher education had higher scores than subjects whose 
parents had only had primary education. Here again, mothers figure more promi- 
nently than fathers. Fourth, let us examine Table XII, which reports scores accord- 
ing to the socio-economic status of the parents. Subjects whose parents had a higher 
socio-economic status had higher scores than subjects with parents with lower status; 
in this instance, fathers figure more prominently than mothers. 
Finally, let us have a look at sex differences, as displayed in Table XIII: it is 
clear that females had higher mean scores than did males. 
Table IX. Extralinguistic Variables: Educational and Intellectual Level 
attending university 3.06 2.83 
attending high school 2.83 2.70 
university : high school % +8.1 +4.8 
specializing in humanities 3.21 2.95 
specializing in technical subjects 2.71 2.58 
humanities : technical % +18.5 +14.3 
high language grades 2.91 2.85 
low language grades 2.87 2.65 
higher : lower % +1.4 +7.5 
high total grades 2.89 2.85 
low total grades 2.89 2.66 
higher : lower % - +7.1 
24 Slovenski iezik - Slovene Linguistic Studies 2 (1999) 
Table X. Extralinguistic Variables: Place of Longest Formative Residence 
self: resident central Slovenia 3.00 2.85 
self: resident elsewhere in Slovenia 2.91 2.61 
central : elsewhere % +3.1 +9.2 
father from central Slovenia 3.09 2.91 
father from elsewhere in Slovenia 2.90 2.62 
central : elsewhere % +6.6 +10.7 
mother from central Slovenia 3.12 3.00 
mother from elsewhere in Slovenia 2.88 2.61 
central : elsewhere % +8.3 +14.9 
Table XI. Extralinguistic Variables. Parents' Education 
father higher education 
father lower education 
higher : lower % +0.3 +2.9 
mother higher education 3.06 2.94 
mother lower education 2.81 2.63 
higher : lower % +8.9 +11.8 
Table XII. Extralinguistic Variables: Parents' Socio-Economic Status 
father higher socio-economic status 3.04 2.80 
father lower socio-economic status 2.81 2.62 
higher : lower % +8.2 +6.9 
mother higher socio-economic status 3.00 2.70 
mother lower socio-economic status 2.84 2.59 
higher : lower % 
T. Priestlv. On Derivational Productivitv in Slovene ... 25 
Table XIII. Extralinguistic Variables: Sex of Respondents 
females 
males 
females : males % +8.9 +13.1 
To summarize: a higher awareness of the norm has been demonstrated among 
certain specific subsets of respondents in this study; namely, a heightened awareness 
was reported by those at a more advanced intellectual or educational level; by those 
who had a personal and family background based in central, rather than peripheral, 
dialects of Slovene; by those whose parents were better educated and of a higher 
socio-economic status; and by those who used, or were being trained to use, lan- 
guage more rather than less in their occupations. In addition, females reported a 
more heightened awareness of the norm than males. Most of these conclusions are, 
indeed, what one may have expected: a greater awareness of the norms of Standard 
Literary Slovene may be predicted among those who are brought up by and among 
central dialect-speakers, given that the standard language is closer to these varieties. 
It may be predicted among those who are better educated, and among those who are 
being trained not in technical subjects but in the humanities. It may be predicted 
among those with better-educated and more socio-economically advantaged parents. 
It may also, given what has been shown for sociolinguistic differences between the 
sexes, be predicted for females rather than males.'' It is nevertheless of interest to be 
able to demonstrate that these predictions hold; and some of the details - the greater 
role of the mother's education on the one hand, and of the father's socio-economic 
status on the other - are striking. These results, when replicated, may have implica- 
tions for those who are involved in teaching Standard Slovene in Slovene schools, 
who may thus with justification take extra care in the teaching of the language to 
schoolchildren in particular subgroups and with particular backgrounds; for, al- 
though an awareness of the standard norm may not reflect ability to use that norm, 
it will surely, at least, be reflected in individual speakers' self -confidence. 
5 Suggestions and Conclusions 
5.1 Suggestions 
It cannot be over-emphasized that this was a first hesitant step in empirical re- 
search into derivational productivity in Slovene. Apart from further replication of 
the tests outlines above, and in particular of the tests resulting in p-ratings and u- 
ratings, there are many other aspects of this problem to be explored. Among them 
are the following: 
Derivations are not all of the same kind; especially, some are, to use the 
words of Dressler (1986), more and some are less morphosemantically transparenLZ6 
25 On the tendency for adult females (especially those in certain socio-economic groups) to 
prefer more standard forms, see Chambers 1995. 
26 This concept has much in common with that of "motivatedness," see, e.g., Lopatin and 
Uluxanov (1966) and Ermakova and Zemskaja (1991) on motivirannost'; see Armstrong (1978) and 
26 Slovenski jezik - Slovene Linguistic Studies 2 (1999) 
Derivations which are more so may, I suggest, be more amenable to tests of the 
kind described here;27 the denominal/deadjectival suffixes -oba, -ost and -stvo may 
for instance provide rewarding material for such tests. Replication of these tests with 
the same or similar cues should be performed, but with attention to two aspects of 
the problem which I was not able, for feasibility reasons, to incorporate. First, it 
would be informative to contrast derivatives which involve affixation on a simple 
stem with those involving more complex affixation. Among the cues for the 1993 
tests, for instance, the majority consisted of the -e- or the -i- suffix being added to 
simple stems; but in several the suffix was added to already-suffixed stems - several 
with -en-, e.g., bakr-en-e-ti, stekl-en-i-ti; a few with phonologically more complex 
components, e.g., kos-mat-e-ti, mnogo-ter-i-ti; two, even, which already had two suf- 
fixes, sum-ot-n-e-ti and slab-ot-n-i-ti - and in others the suffix was added to al- 
ready-prefixed stems (pre-star-e-ti, o-ne-tist-i-ti). One cue, indeed, has a combina- 
tion of these features - po-bed-ast-e-ti. Inspection of Tables IV and V does not de- 
tect any pattern which sets the results for these verbs off from the results for more 
simple derivations. Attention to this factor could - and, for those who may wish to 
hypothesize that derivation involves processes with greater or lesser "depth" - should 
be the basis for further testing. 
Second, there was no opportunity in the experiments reported here to incorp- 
orate variables representing the role of semantics within the scope of operation of 
individual DPs. As mentioned briefly above (2.1), most of the derivations were sem- 
antically transparent (thus, the relationship between rjaveti "to become brown" and 
rjav "brown"), but a few were semantically more opaque (e.g., the relationship is not 
so obvious between dreveneti "to stiffen" and dreven "made of wood," and that be- 
tween izljubiti "to adore" and ljub "dear" is even more tenuous). An experiment 
should be devised with cues ranging from the perfectly transparent to the extremely 
opaque, to see if this factor is involved in productivity. In addition, a different kind 
of test - one in which subjects would have to interpret the meanings of "made-up" 
words - would clarify this aspect of the question. 
5.2 Conclusions 
As far as productivity is concerned, there are three main conclusions. First, it 
is very clear that, for these DPs, productivity is a "cline," but in two quite distinct 
meanings of the term (3.1). Second, there is an enormous difference between type 
frequency and productivity (3.2). Third, the formula (p  - u) + u, as described, 
appears to provide a measure of the relative productivity of different DPs, and 
specifically shows that -iti is more productive than -eti (3.3). The applicability of 
this formula should be checked by tests with much more extensive lists of cues, for 
it is, to the best of my knowledge, something new. And the results of the investiga- 
tion into "blocking" (3.4) were suggestive, but inconclusive. 
In addition, the kvalifikatorji in the SSKJ deserve re-examination and, if pos- 
sible, empirical justification (4.1); and it has been shown possible to determine 
which speakers of Slovene have a greater, and which ones have a lesser, awareness 
of the standard norm (4.2). 
G6rska (1983) who also discuss relative productivity in terms of semantic relationships among 
morphemes. 
27 This was pointed out to me by Wolfgang Dressler, personal communication, 1993. 
T. Priestlv. On Derivational Productivitv in Slovene ... 27 
Appendix I: Cues for 1993 Tests 
Underlined verbs are made-up derivations. 
Inchoative 
apnen 'calceous' apneneti; bakren 'made of copper' bakreneti; bedast 'foolish' po- 
bedasteti; c'vrst 'firm' c'vrsteti; debel 'fat' debeleti; dreven 'made of wood' dreveneti; 
globok 'deep' globeti; goden 'ripe' godneti; go1 'naked' goleti; grd 'ugly' pogrdeti; 
grob 'rough' ogrobeti; hrom 'lame' hrometi; jeklen 'made of steel' jekleneti; kasen 
'late' kasneti; kosmat 'hairy' kosmateti; krut 'cruel' kruteti; Zen 'idle' m; medel 
'flabby' medleti; mrtev 'dead' mrtveti; mrzel 'cold' mrzleti; nem 'dumb' nemeti; plah 
'timid' plaieti; plitev 'shallow' plitveti; pre-star 'too old' prestareti; reven 'poor' po- 
revneti; rjav 'brown' rjaveti; samoten 'lonely' samotneti; smel 'bold' osmeleti; srep 
'staring' srepeti; surov 'crude' suroveti; tolst 'fat' otolsteti; trd 'hard' trdeti; iolt 'yel- 
low' iolteti 
Factitive 
bahat 'proud' bahatiti; besen 'angry' besniti; blag 'gentle' blaiiti; blazen 'insane' bla- 
zniti; dreven dreveniti; droben 'tiny' drobniti; gluh 'deaf' gluiiti; grd ogrditi; grob 
grobiti; hud 'bad' huditi; ljub 'dear' izljubiti; jasen 'clear' jasniti; jeklen jekleniti; ki- 
sel 'sour' kisliti; leden 'icy' ledeniti; Zen leniti; lesen 'made of wood' oleseniti; mnogo- 
ter 'manifold' pomnogoteriti; mrtev mrtviti; ne-c'ist 'unclean' onec'istiti; nov 'new' za- 
noviti; oc'it 'obvious' pooc'ititi; ozek 'narrow' oiiti; reven porevniti; rjav rjaviti; siv 
'grey' siviti; slaboten 'feeble' slabotniti; sladek 'sweet' sladiti; smel osmeliti; srep 
piti; steklen 'made of glass' postekleniti; tesen 'tight' tesniti; tuj 'strange' tujiti 
Appendix 11: Cues for 1994 Tests 
Glosses are provided only for adjectives not used for derivatives for the 1993 test. 
Inchoative 
apnen apneneti; bakren bakreneti; c'vrst c'vrsteti; dreven dreveneti; fanatic'en 'fanati- 
cal' fanatic'neti; gibc'en 'supple' gibc'neti; gotov 'complete' dogotoveti; hrom hrometi; 
jeklen jekleneti; kasen kasneti; klavrn 'low-spirited' klavmeti; kosmat kosmateti; 
krut kruteti; Zen m; ljubezniv 'kind' ljubezniveti; mlac'en 'lukewarm' mlac'neti; 
moc'an 'strong' omoc'neti; mrtev mrtveti; nem nemeti; nemoc'en 'feeble' onemoc'neti; 
plah plaieti; rjav rjaveti; roinat 'pink' roinateti; samoten samotneti; slaboten & 
botneti; smel osmeleti; srep srepeti; surov suroveti; itorast 'clumsy' itorasteti; iolt 
iolteti 
Factitive 
besen besniti; blag blaiiti; blaten 'muddy' blatniti; blazen blazniti; dreven dreveniti; 
droben drobniti; gluh gluiiti; grd ogrditi; grob grobiti; jeklen jekleniti; kisel kisliti; 
leden ledeniti; Zen leniti; mriav 'lean' mriaviti; mrtev mrtviti; ne-c'ist onec'istiti; 
ne-strupen 'not poisonous' onestrupeniti; nov zanoviti; ozek oiiti; ploden 'fertile' 
plodniti; popularen 'popular' popularniti; reven porevniti; rjav rjaviti; roinat 'pink' 
roinatiti; slaboten slabotniti; srep srepiti; strupen 'poisonous' ostrupeniti; tesen tesni- 
ti; iivc'en 'nervous' iivc'niti; iolt oioltiti 
28 Slovenski jezik - Slovene Linguistic Studies 2 (1999) 
Bibliography 
Anshen, Frank, and Mark Aronoff. 1981. "Morphological productivity and phonologi- 
cal transparency," Canadian Journal of Linguistics 16: 63-72. 
Armstrong, Daniel. 1978. "Toward a comparative semantic analysis of suffixal word 
formation in Slavic," 45-81 in H. Birnbaum, ed., American Contributions to 
the Eighth International Congress of Slavists. Columbus, OH: Slavica. 
Aronoff, Mark. 1976. Word Formation in Generative Grammar. Cambridge MA: 
MIT Press. 
- - and R. Schvaneveldt. 1978. "Testing morphological productivity," Annals of the 
New York Academy of Sciences 318: 106-114. 
Baayen, Harald. 1992. "Quantitative aspects of morphological productivity," 104-49 
in Booi and Marle 1992. 
- - and Rochelle Lieber. 1991. "Productivity and English derivation: A corpus based 
study," Linguistics 29: 801-43. 
Bauer, Laurie. 1988. Introducing Linguistic Morphology. Edinburgh: Edinburgh Uni- 
versity Press. 
Besediite 1987 = Milena HajnSek-Holz, Marjeta Humar and Franc Jakopin, eds. Be- 
sediite slovenskega jezika I-II. Ljubljana: Znanstvenoraziskovalni center SAZU. 
Ingtitut za slovenski jezik Frana RamovSa. 
Booi, Geert, and Jaap van Marle. 1988. Yearbook of Morphology. Dordrecht: Kluwer. 
- - and - -. 1992. Yearbook of Morphology 1991. Dordrecht: Kluwer. 
Browne, Wayles. 1981. "Slavic -ba and English *slil: Two persistent constraints," Folia 
Slavica 411-3: 219-226. 
Chambers, Jack. 1995. Sociolinguistic Theory: Language Variation and its Social 
Implications. Oxford: Blackwell. 
Corbin, Danielle. 1987. Morphologie dkrivationelle et structuration de la lexique 
I-II. Tiibingen: Niemeyer. 
Derwing, Bruce L. 1976. "Morpheme recognition and the learning of rules for deri- 
vational morphology," Canadian Journal of Linguistics 2111: 38-66. 
Di Sciullio, A.-M., and E. Williams. 1987. On the Definition of Word. Cambridge 
MA: MIT Press. 
Dressler, Wolfgang. 1987. "Explanation in natural morphology, illustrated with com- 
parative and agent-noun formation," Linguistics 24: 519-548. 
Gjurin, Velemir. 1985. "Priponsko obrazilo -a$ v slovengEini," Slavistitna revija 33: 
195-222. 
G6rska, Elzbieta. 1983. "The methods of establishing the productivity of word forma- 
tion rules," Papers and Studies in Contrastive Linguistics 16: 91-98. 
Greenberg, Sidney, and Randolph Quirk. 1970. Elicitation Experiments in English. 
Linguistic Studies in Use and Attitude. London: Longmans, 1970. 
Lieber, Rochelle. 1992. Deconstructing Morphology. Word Formation in Syntactic 
Theory. Chicago: University of Chicago Press. 
T. Priestlv. On Derivational Productivitv in Slovene ... 29 
Lopatin, V. V., and Uluxanov, I. S. 1966. "Glagoli na -et' v sovremennom russkom 
jazyke ( 0  produktivnosti i reguljarnosti slovoobrazovatel'nogo tipa)," 127-41 in 
Zemskaja, E. A., and 3meleva, D. N. 1966. Razvitie slovoobrazovanija sovre- 
mennogo russkogo jazyka. Moscow: Nauka. 
Mader, Elfriede. 1981. Riicklaufiges Worterbuch des Slowenischen. Klagenfurt: Klag- 
enfurter Sprachwissenschaftliche Gesellschaft. 
Marle, Jaap van. 1988. "On the role of semantics in productivity change," 139-54 in 
Booi and Marle 1988. 
Pineo, Peter C., John Porter and Hugh A. McRoberts. 1977. "The 1971 census and 
the socioeconomic classification of occupations," Canadian Review of Sociolo- 
gy and Anthropology 1411: 91-102. 
PleterSnik, Maks. 1894. Slovensko-nemiki slovar 1-11. Ljubljana: KnezoSkofijstvo. 
Priestly, Tom. 1987. "Slovene verb productivity," Slovene Studies 911-2: 195-203. 
Rainer, Franz. 1988. "Towards a theory of blocking: The case of Italian and German 
quality nouns," 155-85 in Booi and Marle 1988. 
Schultinck, Henk. 1961. "Produktiviteit als morfologisch fenomeen," Forum der Let- 
teren 2: 110-25. 
SSKJ = Slovar slovenskega knjiinega jezika I-V. Ljubljana: Slovenska akademija 
znanosti in umetnosti. 
ToporiSiE, Joie. 1984. Slovenska slovnica. Maribor: Obzorja. 
Vakareliyska, Cynthia. 1994. "Na-drop in Bulgarian," Journal of Slavic Linguistics 
211: 121-150. 
VidoviE-Muha, Ada. 1981. "Pomenske skupine nekakovostnih izpeljanih pridevnikov." 
Slavistitna revija 29: 19-42. 
- - . 1983. "Zloienke s pomenom lastnosti delov organizma," Slavistitna revija 31: 
359-374. 
iele, Andreja. 1993. "Ekspresivne oziroma Eustvene oznake v slovarstvu," Slavistitna 
revija 41: 529-534. 
Prispelo aprila 1998, sprejeto junija 1998 
Received April 1998, accepted June 1998 
0 produktivnosti izpeljave v slovenSEini s pripombami o besedni 
pogostnosti in zavedanju norme 
Poglavitni cilj priEujoEe raziskave je ugotavljanje moinosti za merjenje produk- 
tivnosti besedotvornih vzorcev oz. besedotvornih postopkov, kar oEitno Se ni bilo iz- 
vedeno za noben jezik. Gradivo za Studijo predstavljata dva nasprotujoEa si izpridev- 
niSka besedotvorna vzorca: ,,inhoativno* obrazilo -eti in ~faktitivnoa obrazilo -iti, kot 
npr. rjav + rjaveti 'postajati rjav' proti rjaviti 'delati rjavo'. Analiza je bila izpeljana 
na osnovi psiholoSkega testiranja, ki je bilo izvedeno v Ljubljani v letih 1993-94. L. 
1993 je bilo testirano 186 dijakov in 180 dodiplomskih Studentov. V nadaljevanju 1. 
30 Slovenski iezik - Slovene Linguistic Studies 2 (1999) 
1994 je bilo testiranje omejeno na tri skupine univerzitetnih gtudentov, ki jih je se- 
stavljalo skupaj 116 oseb. Zastavljena so bila gtiri vpraganja: (1) Ali se po vaiem be- 
seda nahaja v knjiini slovenitini? (2) Ali vi to besedo kdaj uporabljate? (3) Ali je 
ta beseda po vaiem moina in razumljiva? in (4) Kako pogosto sami uporabljate to 
besedo? Testirani so izbirali med petimi moinostmi, ki so bile rangirane po Liker- 
tovi lestvici, in podatki so bili nato statistitno obdelani. Kar zadeva produktivnost, 
se je izkazalo - vsaj za ta dva besedotvorna vzorca -, da jo lahko opigemo kot po- 
Sevni potek, vendar v dveh razlitnih pomenih tega izraza. Prvi pomen je samoume- 
ven: en postopek je lahko bolj produktiven kot drugi. Drugi pomen je ta, da t.i. 
nproduktivna mot<< lahko variira. Ce vzamemo dva primera iz srednjih odgovorov na 
vpraganje (3), je razvidno, da je bila beseda godneti povpreEno ocenjena kot skrajno 
nmoina in razumljiva<<, medtem ko je bila na drugem koncu lestvice beseda plaieti 
ocenjena kot skorajda nemogoEa. Med tema dvema skrajnostma se vrstijo vse druge 
vpragane besede. 
DrugiE, obstaja velika razlika med vrstno pogostnostjo in produktivnostjo. Na 
to pogosto kaiejo precejgne razlike med odgovori v zvezi z isto vpragano besedo, na 
vpraganje (4) in druga vpraganja. TretjiE, zdi se, da nam merilo za relativno produk- 
tivnost ralitnih vzorcev daje posebna formula. Ta formula je (p  - u)  t u, pri Ee- 
mer je p srednja vrednost za odgovore na vpraganje (3), u pa sredina odgovorov na 
vpraganje (2). Formula izraia kolitino, za katero odgovori na vpraganje (3) (pov- 
preEno) presegajo odgovore na vpraganje (2). Ta razlika je drugaEna za dejanske be- 
sede in za izmigljene besede. Podatki iz raziskave v Ljubljani 1. 1993, z uporabo 
omenjene formule, kaiejo, da je - kot je bilo pritakovano - obrazilo -iti bolj pro- 
duktivno kot -eti. Za preverjenje tega rezultata je bilo izvedeno drugo testiranje 1. 
1994 in rezultati so bili pritrdilni. Kljub temu so ti rezultati v vet pogledih le pri- 
pravljalne narave. PriporoEamo obnovitev in razgiritev testiranja. 
Poleg tega, je ti1 kot podpostopek testiranja v zvezi z vpraganjem (2) raziskan 
pojav ,,blokiranjaa. Ceprav rezultati niso dokonEni, so zgovorni, in test lahko prila- 
godimo za raziskavo tega pojava. 
PriSli smo Se do dveh drugih zakljuEkov, ki pa nista v zvezi z besedotvorno 
produktivnostjo. Nekateri kvalifikatorji, ki jih navaja SSKJ za glagole v testih, se 
zdijo na osnovi srednjih odgovorov na vpraganje (4) nepravilni. Ta rezultat morda 
velja le za subjekte testiranja, ki so veEinoma stari 16 do 24 let. Menimo pa, da bi se 
kvalifikatorje izplaEalo ponovno oceniti in, Ee je mogoEe, tudi empiritno utemeljiti. 
Podatki o subjektih, ki so opravili teste, so bili analizirani glede na odgovore na 
vpraganje (1). Izsledki te analize kaiejo, da je zavedanje slovenske knjiine norme 
veEje, Ee ima subjekt vigjo intelektualno inlali izobrazbeno stopnjo, Ee je iz osred- 
njih slovenskih nareEnih skupin, Ee ima starge z vigjo izobrazbo inlali vigjim drui- 
beno-gospodarskim poloiajem in Ee je bil izgolan za veEjo rabo jezika pri delu. Po- 
leg tega se ienske bolj zavedajo norme kot mogki. Ti rezultati imajo pomen za golske 
uEitelje slovengtine. 
T. Priestlv. On Derivational Productivitv in Slovene ... 31 
On Derivational Productivity in Slovene with Notes on Lexical 
Frequency and Awareness of the Norm 
The major aim of the study reported here was to discover a way of measuring 
the productivity of derivative patterns or processes, a task which appears never to 
have been accomplished for any language before. As specific material for the study, 
two contrasting deadjectival verbal derivative patterns were chosen: the 'inchoative' 
-eti and the 'factitive' -iti, as occurring in, e.g., rjav 'brown', rjaveti 'to become 
brown', rjaviti 'to make (someone or something) brown.' The method involved psy- 
cholinguistic tests, administered in Ljubljana in 1993-94. The subjects for the tests 
in 1993 were 186 secondary and 180 post-secondary students. For the 1994 follow-up 
study the tests were limited to three groups of university respondents, totalling 116 
in all. Cues comprised four questions: (1) Ali se po vaiem beseda nahaja v knjiini 
slovenitini? (2) Ali vi to besedo kdaj uporabljate? (3) Ali je ta beseda po vaiem 
moina in razumljiva? and (4) Kako pogosto sami uporabljate to besedo? Subjects se- 
lected responses on five-point Likert scales; the data were analyzed statistically. As 
far as productivity is concerned, there are three conclusions. First, it is clear that, at 
least for these derivative patterns, productivity is a "cline," but in two quite distinct 
meanings of the term. First and most obviously, one process may be synchronically 
more productive than another. Second, what may be called the productive strength 
of any one process also varies. To take just two examples from the mean responses 
to question (3), we see that the cue godneti was assessed, on average, as extremely 
"possible and understandable," while at the other end of the scale the cue plaieti 
was assessed, on average, as well nigh impossible and incomprehensible; the remain- 
ing cues are strung out along the cline in between the two. 
Second, there is an enormous difference between type frequency and produc- 
tivity; this is shown specifically by the often very significant differences between 
answers, for the same cue, to question (4) and the other questions. Third, one parti- 
cular formula appears to provide a measure of the relative productivity of different 
patterns. This formula is expressed as (p  - u) + u,  where p is the mean for re- 
sponses to question (3) and u is the mean for responses to question (2). The formula 
expresses the amount by which (on average) the responses to Question 3 exceed the 
responses to Question 2; and this disparity, in turn, is significantly different for real 
words as compared with made-up words. The data from the experiments in Ljub- 
ljana in 1993 showed, using this formula in these applications, that - as hypothes- 
ized - -iti is more productive than -eti. To test this finding, a second set of tests 
were administered in 1994, with confirmatory results. There are however several re- 
spects in which this finding must be considered preliminary; replication and exten- 
sion of the tests is urged. 
In addition, the phenomenon known as 'blocking' was investigated with one 
sub-routine to tests involving question (2). The results, though inconclusive, are sug- 
gestive, and this test may be adapted for further investigation of this phenomenon. 
Incidentally, two other conclusions, having nothing to do with derivational pro- 
ductivity, were reached. Several of the kvalifikatorji that are provided in the SSKJ 
for verbs used in the tests appear, on the evidence of mean responses to question 
(4), to be incorrect; this finding may only apply to the subjects in the tests, who 
were mostly aged between 16 and 24; but it is suggested that the kvalifikatorji de- 
32 Slovenski iezik - Slovene Linguistic Studies 2 (1999) 
serve re-examination and, if possible, empirical justification. Finally, information 
collected about the subjects who did the tests was analyzed with respect to answers 
to question (1). A higher awareness of the norm of Standard Slovene was reported 
by those at a more advanced intellectual or educational level; by those who had a 
personal and family background based in central, rather than peripheral, dialects of 
Slovene; by those whose parents were better educated and of a higher socio-econom- 
ic status; and by those who used, or were being trained to use, language more rath- 
er than less in their occupations. In addition, females reported a more heightened 
awareness of the norm than males. This has implications for teachers of the Slovene 
language in schools.