Sašo Živanovic
University of Ljubljana*
UDK 811.163.6'367
conjunctive and prepositional comparatives
in slovenian
1 INTRODUCTION
In Slovenian, the comparative complement can be headed either by the conjunction kot, as in (1a), or by the preposition od, as in (1b). I call the two types of comparative structures conjunctive and prepositional, respectively. The objective of this paper is to compare the acceptability of the two types of structures with respect to certain syntactic and morphological factors.
(1) a.
Janko J
nom.m.sg.
je starejši be older 3.sg. nom.m.sg.
kot Metka. than M conj. nom.f.sg.
b. Janko	je starejši	od	Metke.
J	be older	than	M
nom.m.sg. 3.sg. nom.m.sg. prep.	gen.f.sg. 'Janko is older than Metka.'
Pancheva (2006: 252) observes that in Polish and	Serbo-Croatian, temporal adver-bials are not acceptable in comparative complements of prepositional comparatives, (2)-(3).1
The same holds for Slovenian, (4).
(2)	* Ania jest weselsza dzisiaj	od wczoraj.
A is happier today	from yesterday
'Ania is happier today	than yesterday.' (Polish)
(3)	* Ana je još gladnija od	juče.
A is even hungrier than	yesterday
'Ana is even hungrier than	yesterday.' (Serbo-Croatian)
Author's address: Filozofska fakulteta, Oddelek za primerjalno in splošno jezikoslovje, Aškerčeva 2, 1000 Ljubljana, Slovenia. Email: saso.zivanovic@guest.arnes.si
1 Pancheva (2006) uses the terms referential partitive and predicative partitive comparatives for conjunctive and prepositional comparatives, respectively. She bases her terminology on her semantic analysis. As far as I can see, our terminologies are extensionally equivalent.
(4) * Ana	je se bolj lacna	od vceraj.
A	be even more hungry	than yesterday
nom.f.sg. 3.sg.	nom.f.sg. prep.
'Ana is even hungrier than yesterday.'
Pancheva suggests the possibility that such examples are ungrammatical due to "case-resistance on the part of yesterday, or a problem with the topicalization of the adverb" Pancheva (2006: 251). Neither of these suggestions can work: in (5), adverb zjutraj is a complement of preposition od (here meaning 'from'), and the complex temporal adverbial od zjutraj do pete ure popoldne is most likely the topic.
(5) Od zjutraj	do pete ure popoldan	sem
from in the morning to fifth hour in the afternoon be gen.f.sg. gen.f.sg.	l.sg.
bila	tega
be	this
ptc.f.du. gen.m.sg.
dne pri Jani, ob petih pa sta prišli še mama in sestra. day at J at five but be come also mother and sister gen.m.sg. loc.f.sg. loc.pl.	l.du. ptc.f.du. nom.f.sg.	nom.f.sg.
'From the morning to five PM, I was at Jana's that day, and at five my mother and sister also arrived.' (FidaPlus: Jana 3662)
Pancheva also notes that in Russian temporal adverbials are acceptable in comparative complements of prepositional comparatives, (6), although not with any kind of adverbial. For example, place adverbials are unacceptable (*moskovskogo 'Moscow-adj.gen.') and not even any temporal adverbial will do (*proslogodnego 'last year-ADJ.GEN.') (2006: 251).
(6) Maša segodnja veselee včerašnego. M today jollier yesterday
adj.gen.
'Masha is jollier today than yesterday.' (Russian)
It is the objective of this paper to answer Pancheva's call for further cross-linguistic research on the topic by exploring the patterns of acceptability of prepositional comparatives in Slovenian with respect to various syntactic and morphological factors, while additionally contrasting them with conjunctive comparatives.
2 A NOTE ON TERMINOLOGY
The literature on comparatives usually distinguishes clausal comparatives (7), where than introduces a clause (which is most often reduced), and phrasal comparatives (8), where than introduces a single phrase (see e.g. Lechner 2004). Complements of than in clausal comparatives are uncontroversially taken to be derived from full clauses by some kind of ellipsis, (7a). On the other hand, the question whether the same holds
for phrasal comparatives is one of the central issues of the debate. Proponents of the ellipsis analysis argue that phrasal comparatives such as (8) are simply the endpoint of ellipsis, (8a). On the contrary, proponents of the direct analysis hold that there is no covert structure in phrasal comparatives, (8b).2
(7)	The book is longer than the article is.
a. The book is longer than the article is long.
(8)	The book is longer than the article.
a.	The book is longer than the article is long.
b.	The book is longer than the article.
Using the notions of clausal and phrasal comparatives as the primary classificatory tool is useful for languages such as English, where—with respect to the phonological form—a single lexical item introducing the comparative complement can be found (than in English). However, having more than one phonologically distinct comparative complement head is certainly a better hint to the diversity of comparative structures. In Slovenian I therefore primarily distinguish between comparatives employing the conjunction kot and the preposition od as the comparative complement head, dubbing them conjunctive and prepositional comparatives, respectively.3
The non-elided phrases in the comparative complement of conjunctive (clausal) comparatives are usually called remnants, and their counterparts with the same grammatical function in the matrix clause associates. I adopt the same terminology for prepositional comparatives (without an a priori commitment to the ellipsis analysis of prepositional comparatives).
3 SLOVENIAN DATA
The primary objective of this section is to provide data on the acceptability of different varieties of prepositional comparative structures in Slovenian. To provide a background for the endeavour, conjunctive comparatives are examined first, since—as will become clear shortly—their behaviour is more stable across speakers. To provide a clear parallel between the two kinds of structures, only conjunctive comparatives with a single remnant will be considered, such as (9a)-(9b). In comparative clauses with several remnants, such as (9), all the restrictions valid for single remnants apply simultaneously. Thus, (9) is acceptable since both (9a) and (9b) are acceptable.
2	Some authors argue for an intermediate position. Also, phrasal comparatives with explicit standards are usually given a direct analysis, even by the proponents of the ellipsis analysis. (See Lechner 2004: 93)
3	Both kot and od are also used in non-comparative structures (see Bajec 1994). That they are a conjunction and a preposition (in all uses) is fairly obvious: (i) kot is often followed by a full clause or a partially reduced clause; when it introduces a single noun phrase, its case is governed by the grammatical function of its associate in the matrix clause; (ii) od can only introduce a single phrase; if this is a noun-phrase, it must be genitive.
(9) Včeraj je na predavanje prišlo več študentov kot danes na koncert.
yesterday be on lecture come more student than today on concert 3.sg. acc.n.sg. ptc.3.n.sg. gen.m.pl. conj.	acc.m.sg.
'More students came to the concert yesterday than to the lecture today.'
a.	Včeraj je na predavanje prišlo več študentov kot danes. Yesterday be on lecture come more student than today
3.sg. acc.n.sg. ptc.3.n.sg.	gen.m.pl. conj.
'More students came to the lecture yesterday than today.'
b.	Na predavanje je včeraj prišlo več študentov kot na koncert. on lecture be yesterday come more student than on concert
acc.n.sg. 3.sg.	ptc.3.n.s	gen.m.pl. conj. acc.m.sg
'Yesterday, more students came to the lecture than to the concert.'
The acceptability of comparative structures is investigated with respect to three dimensions. The first one is the morphological environment of the comparative morpheme: amount and quality comparatives are distinguished.4
Amount comparatives are used for comparision of amounts of objects. Comparative determiners are typical examples of amount comparatives. In Slovenian, the amount comparative morpheme is več. It can be a part of morphologically complex words such as večkrat 'more times' or največ 'most'; standalone, as in (10), več 'more' functions as a comparative determiner.
Quality comparatives compare the degree to which the objects exhibit a certain gradable property. A typical quantity comparative is a comparative adjective. There are two forms of the quality comparative morpheme in Slovenian: the synthetic form -(ej)š- and the analytic form bolj 'more'.
The other two dimensions are the syntactic environments of the comparative morpheme and the associate; specifically, the grammatical function of the major phrases containing them. The grammatical functions investigated in this paper are the following:
4	While the term 'amount comparative' has been previously used (e.g. by Hackl 2001, Lechner 2004), 'quality comparative' is my own invention: I simply could not find a term covering all the non-amount comparatives I discuss in this paper.
5	Depending on the speaker, več is either declinable (only plural forms exist: nom., gen., acc. več, dat. večim, loc. večih, ins. večimi) or not. Some numerals exhibit the same variation.
(nominative) subject (S);6
•	(nominative) subject (S);
•	(accusative) direct object (DO);
•	(dative) indirect object (IO);
•	prepositional object (PO);
•	locative adverbial (L);
•	temporal adverbial (T);
•	other adverbials (Adv);
•	predicate (P).
I will refer to various types of comparative structures along the second and third dimension as X comparatives and X-associate comparatives, respectively, where X is any of the above grammatical functions. For examples of various types of comparative structures see the following subsection.
3.1 Conjunctive comparatives
This section provides the data on conjunctive comparatives. Most combinations along the above described dimensions yield an acceptable construction.7
Below I provide several examples, illustrating for every dimension all possible variation within the dimension.8 The data is summarized in Table 1.
(10) Včeraj je prišlo več študentov kot danes. yesterday be come more student than today
3.sg. ptc.3.n.sg.	ptc.3.n.sg. conj.
'More students came yesterday than today.' (am, S, T-ass)
6	Dative subjects were not investigated. My intuition is that they behave like nominative subjects and not like dative objects.
7	Several combinations are not possible for reasons independent of comparatives. In Tables 1 and 2, they are marked as non-available.
8	Listing all logically possible combinations (2*8*8=128) results in a list that is too long to include. In the following examples, the type of a comparative structure is given next to its translation.
For example, consider what could be a quality P DO-associate comparative. The closest candidate might be formed by a "gradable" transitive verb such as prehitevati 'be too fast'. However, such verbs take measure phrases as (accusative) direct objects. When forming a P-comparative, the direct object is prohibited, (i)—actually, the analysis of (i) as a DO-comparative is probably preferable.
(i) Ta ura zaostaja bolj (*pet minut) kot tista.
this watch be too fast more five minutes than that. 'This watch is (*five minutes) slower than that one.'
(11) Huligani so več flaš	razbili kot	popili.
hooligan be more bottle break than	drink
nom.m.pl. 3.pl.	gen.f.sg. ptc.m.pl. conj.	ptc.m.pl.
'The hooligans broke more bottles than they drank.' (am, DO, P-ass)
	amount								quality							
associate	S	DO	IO	PO	L	T	Adv	P	S	DO	IO	PO	L	T	Adv	P
S	V	V	?	V	v?	V	V	V	*	V	V	V	v?	V	V	V
DO	V	V	?	V	?	V	V	n/a	?	?	?	V	V	v?	V	n/a
IO	v?	V	V	V	V	V	V	n/a	?	V	*	V	V	V	V	V
PO	V	V	v?	V	v?	v?	V	n/a	V	v?	v?	?	V	V	V	V
L	V	V	V	V	?	V	V	V	V	V	v?	V	*	v?	V	V
T	V	V	V	V	V	V	V	V	v?	V	v?	V	V	*	V	V
Adv	v?	v?	v?	v?	v?	v?	V	n/a	*	*	*	*	*	*	V	n/a
P	V	V	v?	V	V	V	V	V	*	*	*	*	*	*	V	V
Table 1: Distribution of conjuctive comparatives
(12) Pismo so poslali več(im) politikom kot poslovnežem. letter be send more politician than bussinessman acc.f.sg. 3.pl. ptc.m.pl.g. (dat.pl.) ptc.3.n.sg. conj. dat.m.pl. 'They sent the letter to more politicians than businessmen.' (am, IO, IO-ass)
(13)	Lani si mi pisal iz	več(ih) mest kot letos. last year be I write from more city than this year acc.f.sg.. 3.pl. gen.f.sg. ptc.m.pl. conj. ptc.m.pl.
'Last year you sent me greetings from more cities than this year. (am, L, T-ass)
(14)	Tečem ob več(ih) priložnostih kot	plavam. run at more opportunity than swim pres.1.sg.g. 3.pl. (loc.pl.)pl.g. loc.f.pl. conj. pres.1.sg.
'I grab more oppurtunities for jogging than swimming.' (am, T, P-ass)
(15) Janko nas je obiskal	večkrat kot	Metka.
J we be visit	more times than	M
nom.m.sg. acc.pl. 3.sg. ptc.m.sg.	conj.	nom.f.sg.
'Janko visited us more often than Metka.'	(am, Adv, S-ass)
(16) Z očetom ima več težav kot z mamo. with father have more problem than with mother ins.m.sg. 3.sg.	gen.f.pl. conj.	ins.f.sg.
'He has more trouble with his father than with his mother.' (am, DO, PO-ass)
(17) Janko je pomislil na več podrobnosti kot	Metka.
J	be think on more detail	than	M
nom.m.sg. 3.sg. ptc.m.sg.	acc.f.pl.	conj.	nom.f.sg.
Janko considered more details than Metka.' (am, PO, S-ass)
(18) Metka je lepša kot Janko. M	be prettier than J
nom.f.sg. 3.sg. nom.f.sg. conj. nom.m.sg. 'Metka is prettier than Janko.' (q, P, S-ass)
(19) Ta	članek sem napisal hitreje kot tistega.
this article be write faster than that acc.m.sg. acc.m.sg. 1.m.sg. ptc.m.sg. adv. conj. acc.m.sg. 'I wrote this article faster than that one.' (q, Adv, DO-ass)
(20) Na Krku sem ujel večjo ribo kot v Izoli. on K	be catch larger fish than in I
loc.m.sg. 1.sg. ptc.m.sg. acc.f.sg acc.f.sg conj.	loc.f.sg.
'I caught a bigger fish on island Krk than in Izola.' (q, DO, L-ass)
(21) * Pogosteje govorim hitro kot preudarno. more often speak fast than deliberately
1.sg.	conj.
'I speak fast more often than deliberately.' (q, Adv, Adv-ass)
(22) Janko je .bolj .ponosen na mamo kot na očeta. J	be more proud on mother than on father
nom.m.sg. 3.sg.	nom.m.sg. acc.f.sg. conj. acc.m.sg.
'Janko is prouder of his mother than his father.' (q, P, PO-ass)
As the summary in Table 1 shows, all structurally possible varieties of amount comparatives are (more or less) acceptable. In quality comparatives, the situation is different. There are two classes of unacceptable combinations: (i) most of Adv- and P-associate comparatives, and (ii) half of the structures where the comparative morpheme is embedded in the associate (henceforth CIA comparatives).
We first turn to Adv- and P-associate quality comparatives. Their general unaccept-ability is exemplified by (23) and (24);9 cf. parallel amount comparatives (25), which are perfectly well-formed. Adv and P comparatives ((26)-(27), and the already given (21)) present an exception in the class of Adv- and P-associate comparatives: they are perfectly acceptable.10
(23)	* Izpit so naredili pametnejši študentje kot padli.
exam be make smarter student than fall acc.m.sg. 3.pl. ptc.m.pl nom.m.pl. nom.m.pl. conj. ptc.m.pl Intended: 'The students who passed the exam were smarter than those who failed it.' (q, S, P-ass)
(24)	* Študentke so pismo prinesle lepšim	profesorjem kot poslale.
student be letter bring more beautiful professors than send nom.f.pl. 3.pl. acc.f.sg. ptc.f.pl. dat.m.pl.	dat.m.pl. conj. ptc.f.pl.
Intended: 'The professors that were brought the letter by the students were more beautiful than the professors that were sent the letter.' (q, IO, P-ass)
(25)	Predstojniku je pismo več študentov poslalo kot prineslo. head	be letter more students send than bring m.sg.dat. 3.sg. n.sg.acc.	m.pl.nom. ptc.n.sg.. conj. ptc.n.sg. 'The number of students who sent the letter to the head was greater than the number of students who brought the letter to him.' (am, S, P-ass)
(26)	Predstojniku so študentje pismo hitreje poslali kot prinesli. head	be student letter quicklier send than bring dat.m.sg. 3.pl. nom.m.pl. acc.n.sg.	ptc.m.pl. conj. ptc.m.pl. 'The students were quicker in sending the letter to the head than in bringing it to him.' (q, Adv, P-ass)
9	In Slovenian, verbs resist occurring in the sentence-initial position, therefore the examples are constructed to avoid such situation. Parallel verb-initial examples are judged even worse.
10	Actually, quality P-associate P comparatives present a classical example of comparative subdele-tion. Adv-associate P comparatives are structurally impossible.
(27)	Ta miza je bolj dolga kot	široka. this table be more long than	wide nom.f.sg. nom.f.sg. 3.sg. nom.f.sg. conj.	nom.f.sg. 'This table is longer than it is wide.' (q, P, P-ass)
Next, consider CIA comparatives. S-, IO-, L- and T-CIA comparatives11 are unacceptable, (28)-(31). Since there can be only one predicate per sentence, it was already shown in the discussion of P-associate quality comparatives that P-CIA comparatives are acceptable, (27). Example (32) further shows that Adv-CIA comparatives are fully acceptable. It is quite surprising, however, that DO- and PO-CIA comparatives are also acceptable, although only partially.
(28)	* Predstojniku so pisali pametnejši študentje kot profesorji.
head	be write smarter student than professor
dat.m.sg. 3.pl. ptc.m.pl. nom.m.pl. nom.m.pl. conj. nom.m.pl. Intended: 'The students who wrote to the head were smarter than the professors who did that.' (q, S, S-ass)
(29)	* Janko je pisal lepši plesalki kot pevki.
J	be write prettier dancer than singer
nom.m.sg. 3.sg. ptc.m.sg. dat.f.sg. dat.f.sg. conj. dat.f.sg.
Intended: 'The dancer that Janko wrote to is prettier than the singer he wrote to.'
(q, IO, IO-ass)
(30)	*Novinar je poročal iz bolj umazanega mesta kot vasi.
journalist be report from more dirty city than village nom.m.sg. 3.pl. ptc.m.sg.	gen.n.sg. gen.n.sg. conj. gen.f.sg.
Intended: 'The city that the journalist reported from was dirtier than the village he reported from.' (q, L, L-ass)
(31)	*V reki so se kopali (nekega) hladnejšega jutra kot večera.
in river be SE take a bath (some) colder morning than evening
loc.f.sg. 3.pl. refl. ptc.m.pl. gen.n.sg. gen.n.sg. gen.n.sg. conj. gen.m.sg. Intended: 'The morning when they took a bath in the river was colder than the evening when they did that.' (q, T, T-ass)
11 X-CIA comparative is an X-associate X comparative.
(32) Janko	govori	bolj	hitro	kot	glasno.
J	speak	more	fast	than	loud
nom.m.sg.	pres.3.sg.		adv.	conj.	adv.
'Janko speaks more fast than loud.' (q, Adv, Adv-ass)
(33)	? Janko je napisal daljši roman kot pesem.
J	be write longer novel than poem
nom.m.sg. 3.sg. ptc.m.sg. acc.m.sg. acc.m.sg. conj. acc.f.sg.
'The novel that Janko write is longer than the poem that he wrote.' (q, DO, DO-ass)
(34)	? Janko je padel v globljo jamo kot jarek.
J	be fall in deeper cave than ditch
nom.m.sg. 3.sg. ptc.m.sg. acc.f.sg. acc.f.sg. conj. acc.m.sg
'The cave Janko fell into was deeper than the ditch he fell into.' (q, PO, PO-ass)
Finally, note that although speakers differ in their judgements on the above data (the question marks in Table 1 are actually mainly the average of polarized judgements), there is a sharp difference between acceptable and unacceptable sentences. Sentences were judged on a scale from 1 to 5. There is a gap from 2 to 3. No average score falls within this area: all unacceptable sentences have the average score between 1 and 2. Thus, the partitioning into acceptable and unacceptable sentences seems well motivated.
3.2 Prepositional comparatives
Compared to conjunctive comparatives, prepositional comparatives are in general much more restricted: only S- and DO-associate comparatives are acceptable at all. The severe unacceptability of IO-, PO-, L-, T-, Adv- and P-associate prepositional comparatives is illustrated in (35)-(40).
(35)** Janko je poslal predsedniku vec pisem kot generalu. J	be send president more letter than general
m.nom.sg. 3.sg. ptc.m.sg. dat.m.sg.	gen.f.pl. prep. gen.m.sg.
Intended: 'Janko sent more letters to the president than to the general.'12 (am, DO, IO-ass)
12
12 Example (35) is acceptable in the non-relevant reading 'Janko sent more letters to the president than the general did,' where the associate is the subject. The same holds for (36) (first option) and (42).
(36)** Bolj je ponosen na očeta od {mame / na mamo]. more be proud on father than mother on mother 3.sg. nom.m.sg. acc.m.sg. prep. gen.f.sg.	acc.f.sg.
Intended: 'He is prouder of his mother than father.' (q, P, PO-ass)
(37)**Na Krku sem ujel večjo ribo od {Izole / v Izoli]. on K	be catch larger fish than I	in I
loc.m.sg. 1.sg. ptc.m.sg. acc.f.sg. acc.f.sg. prep. gen.f.sg. loc.f.sg. Intended: 'I caught a bigger fish on island Krk than in Izola.' (q, DO, L-ass)
(38)	** Zjutraj	ponavadi spijem več kav od zvečer.
in the morning usually drink more coffee than in the evening
pres.1.sg.	gen.f.pl. prep.
Intended: 'In the morning I usually drink more coffees than in the evening.' (am, DO, T-ass)
(39)	** Pogosteje govorim hitro od preudarno.
more often speak fast than deliberately
1.sg.	prep.
Intended: 'I speak fast more often than deliberately.' (q, Adv, P-ass)
(40) ** Huligani so več flaš razbili od popili. hooligan be more bottle break than drink nom.m.pl. 3.pl.	gen.f.sg. ptc.m.pl. prep. ptc.m.pl.
'The hooligans broke more bottles than they drank.' (am, DO, P-ass)
Table 2 provides the data on the acceptability of S- and DO-associate comparatives. It must be emphasized, though, that the data in Table 2 is less reliable than the data in Table 1. In contrast to conjunctive comparatives, the acceptable and unacceptable sentences are not clearly delimited: the average scores Table 2 is based on are evenly distributed. Also, their acceptability seems more idiosyncratic. For example, one of the informants accepted almost no prepositional comparatives, the only exception being quality Adv and P comparatives. The informant was disregarded in the statistics for the table.
Disclaimer provided, the general trends of acceptability that can be read off Table 2 are as follows.
First, S-associate comparatives are more acceptable than DO-associate comparatives.
	amount							quality								
associate	S	DO	IO	PO	L	T	Adv	P	S	DO	IO	PO	L	T	Adv	P
S	*	V	*?	?	?	?	V?	?	*?	V	?	?	V?	V?	V	V
DO	*	*	*	*?	*	*?	V?	n/a	*	?	*	*?	*?	?	?	n/a
Table 2: Distribution of prepositional comparativesassociate
(41)	V? Študentje so jedli v dražji	restavraciji od profesorjev.
student be eat in more expensive restaurant than professor nom.m.pl. 3.pl. ptc.m.pl. loc.f.sg.	loc.f.sg. prep. gen.m.pl.
'The students ate in a more expensive restaurant than the professors.' (q, L, S-ass)
(42)	V? Raco so jedli v dražji	restavraciji od piščanca.
duck be eat in more expensive restaurant than chicken acc.f.sg. 3.pl. ptc.m.pl. loc.f.sg.	loc.f.sg. prep. gen.m.pl.
'They ate duck in a more expensive restaurant than they ate chicken in.' (q, L, DO-ass)
Second, a rough scale of acceptability with respect to the syntactic environment of the comparative morpheme is the following: [DO, PO, Adv, P] > [L, T] > {IO, S}. See (43)-(45) for quality and (46)-(48) for amount comparatives.
(43)	V? Janko je	napisal daljšo pesem od Metke.
J	be	write longer poem than M
nom.m.sg. 3.sg. ptc.m.sg. acc.f.sg. acc.f.sg. prep. gen.f.sg. 'Janko wrote a longer poem than Metka.' (q, DO, S-ass)
(44)	V? Janko je zamudil v službo na pomembnejši dan od Metke.
J	be be late in work on more important day than M
nom.m.sg. 3.sg. ptc.m.sg. acc.f.sg. acc.m.sg.	acc.m.sg. prep. gen.f.sg.
'Janko was late for work on a more important day than Metka.' (q, T, S-ass)
(45)	V? Izpit za avto so delali starejši fantje od deklet.
exam for car be work older boy	than girl
acc.m.sg.	acc.m.sg. 3.pl. ptc.m.pl. nom.m.pl. nom.m.pl. prep. gen.f.pl.
Intended: 'The boys who were trying to pass the driving license were older than the girls who were doing that.' (q, S, S-ass)
(46)	V Janko je pojedel več krofov od Metke.
J	be eat more doughnut than M
nom.m.sg. 3.sg. ptc.m.sg.	gen.m.pl. prep. gen.f.sg.
'Janko ate more doughnuts than Metka.' (am, DO, S-ass)
(47)	? Janko se je izgubil ob več(ih) priložnostih od Metke.
J	se be lose at more opportunity than M
nom.m.sg. refl. 3.sg. ptc.m.sg. (loc.pl.) loc.f.pl. prep. gen.f.sg. 'Janko got lost at more opportunities than Metka.' (am, T, S-ass)
(48)	* Izpit je naredilo več	student od študentk.
exam be make more gen.m.pl. than student acc.m.sg. 3.sg. ptc.n.pl. študentov	prep. gen.f.pl.
Intended: 'More male than female students passed the exam.' (am, S, S-ass)
Finally, quality comparatives are slightly more acceptable than amount comparatives.13
(49)	V Janko nas je obiskal bolj pogosto od Metke.
J	we be visit more often than M
nom.m.sg. acc.pl. 3.sg. ptc.m.sg.	prep. gen.f.sg.
'Janko visited us more often than Metka.' (q, Adv, S-ass)
(50)	V? Janko nas je obiskal večkrat od Metke.
J	we be visit more times than M
nom.m.sg. acc.pl. 3.sg. ptc.m.sg.	prep. gen.f.sg.
'Janko visited us more often than Metka.' (am, Adv, S-ass)
13 Adv-associate DO comparatives seem to be an exception to this generalization. However, note that quality Adv (and P) comparatives were the only ones accepted by the informant who generally disliked prepositional comparatives, and that that informant is disregarded Table 2. Including the informant in the statistics, the exception ceases to exist.
4 CONCLUSION
Unfortunately, the best conclusion to draw on prepositional comparatives from the above presented data might be to not jump to any rash conclusions. The situation seems even worse than anticipated by Pancheva. There is not only cross-linguistic, but also a great amount of idiolectal variation. Judgements on prepositional comparatives do not fall into clearly delimited categories, but rather form a "continuum."
There is, however, a sense in which the idiolectal indeterminacy of prepositional comparatives is a welcome result for a semanticist. Comparison of Slovenian conjunctive and prepositional comparatives immediately reveals striking dissimilarities between the two. First, the distribution of prepositional comparatives is much more restricted. Second, within the (at least partially) acceptable paradigms, conjunctive comparatives exhibit some clear strucutral restrictions (prohibition against Adv-asso-ciate, P-associate and CIA quality comparatives, both modulo the exclusion of Adv and P comparatives), whereas the restrictions on such prepositional comparatives are less clearly defined, and actually more aptly called tendencies than restrictions. These differences present a serious problem for any reduced clause analysis of prepositional comparatives. In a nutshell, Slovenian provides an argument against a reduced clause analysis of prepositional comparatives.
Of course, this does not mean that no phrasal comparatives have a clausal source. It has only been argued that prepositional comparatives, which are all phrasal, do not have a clausal source. Obviously, Slovenian conjunctive comparatives with a single remnant have a clausal source, although they are phrasal. So, in a language such as English, it only makes sense to claim that certain phrasal comparatives are structurally ambiguous between the conjunctive and prepositional structure, the former having a clausal source and the latter not. (Given the results of section 3.2, I would of course predict that not all English phrasal comparatives are ambiguous in this way.)14
I turn to the question where the difference between conjunctive and prepositional comparatives could come from. The answer I will try to provide is by no means exhaustive, but rather a pointer for further research.
The first thing to note is that while Slovenian preposition od can only introduce the comparative complement of a comparative adjective or adverb, conjunction kot can also be used in several other types of "comparative" structures, (51)-(53).
(51) Smo tudi mi tako lepi	kot naša dežela?
be also we so beautiful than our country l.pl	nom.	nom.m.pl. conj. nom.f.sg. nom.f.sg.
'Are we as beautiful as our country?' (FidaPlus: VECER 408)
14 Also note that arguing against a reduced (full) clause analysis does not immediately imply a direct analysis. Pancheva (2006), for example, argues for a reduced small clause analysis of prepositional comparatives.
(53) Možnosti je veliko, tako novih kot že uveljavljenih. option be many so new than already established gen.f.pl. 3.sg.	gen.f.pl conj.	gen.f.pl.
'There is a lot of options, both new and established.' (FidaPlus: MONITOR 51)
(52) ... da so živali na mraz drugače občutljive kot človek.
that be animal on coldness differently sensitive than human
3.pl. nom.f.pl. acc.m.sg.	nom.f.pl. conj. nom.m.sg.
'... that animals have a different sensitivity to cold than humans.' (FidaPlus: MALI.SVET 216)
A natural way of interpreting this fact is to assume that od is licensed by comparative adjectives (and adverbs), which implies that it must be base-generated in the vicinity of the comparative morpheme, possibly as its complement; see Heim (2001) for discussion. On the other hand, kot clause is independent of the comparative morpheme and can be base-generated higher up in the clause, forming a coordination-like structure (see Lechner 2004).
However, the analogous interpretation of conjunctive and prepositional comparatives forces us to assume that the prepositional comparative complement moves to the same position where the conjunctive comparative complement is base-generated. It seems that it is this movement which is heavily constrained and causes the difference in the distribution of prepositional and conjunctive comparatives. Of course, further research is needed to confirm of refute this conjecture, especially in the light of extreme cross-linguistic and idiolectal variation in the acceptability of various kinds of prepositional comparatives.
References
Bajec, Anton (ed) (1994) slovar slovenskega knjižnega jezika. Inštitut za slovenski jezik Frana
Ramovša ZRC SAZU. Hackl, Martin (2001) comparative Quantifiers. Ph. D. thesis, MIT.
Heim, Irene (2001) »Degree Operators and Scope.« In: C. Fery/W. Sternefeld (eds), Audiatur
Vox sapientiae. A Festschrift for Arnim von stechow. Berlin: Akademie-Verlag, 214-239. Lechner, Winfried (2004) Ellipsis in comparatives. Berlin: Mouton de Gruyter. (Studies in
generative grammar, 72). Pancheva, Roumyana (2006) »Phrasal and Clausal Comparatives in Slavic.« In: J. Lavine/ S. Franks/M. Tasseva-Kurktchieva/H. Filip (eds), Formal Approaches to slavic Linguistics 14: The princeton Meeting 2005. Michigan Slavic Publications, 236-257.
Fidaplus: corpus of Slovenian language. www.fidaplus.net. (Corpus FidaPLUS is managed by Faculty of Arts, University of Ljubljana.)
Summary
CONJUNCTIVE AND PREPOSITIONAL COMPARATIVES IN SLOVENIAN
The paper deals with the distribution of conjunctive and prepositional comparative structures in Slovenian. They are introduced by the complementiser kot and the preposition od, respectively.
Comparative structures are categorised along three dimensions: (i) the morphological environment of the comparative morpheme (yielding amount and quality comparatives); (ii) the syntactic environment of the comparative morpheme (eight syntactic environments are discussed: (nominative) subject, (accusative) direct object, (dative) indirect object, prepositional object, locative adverbial, temporal adverbial, some other adverbial, and predicate); (iii) the syntactic environment (the same environments as above are discussed) of the associate (the non-elided phrase in the comparative complement is the remnant; its counterpart with the same grammatical function in the matrix clause is the associate).
The comparison of conjunctive and prepositional comparatives shows that the distribution of the latter is more restricted and also exhibits more inter-speaker variation than the former. Conjunctive comparatives are acceptable in virtually all combinations of the above-mentioned parameters, the only exception being quality comparatives with an adverbial or predicate associate where the comparative morpheme is embedded in the associate. The only absolute generalisation that can be made about prepositional comparatives is that the associate must be either a subject or a direct object; all other generalisations are merely tendencies.
Povzetek
VEZNIŠKE IN PREDLOŽNE PRIMERNIŠKE ZGRADBE V SLOVENŠČINI
Prispevek proučuje distribucijo vezniških in predložnih primerniških zgradb v slovenščini. Prve uvaja veznik kot, druge predlog od. Primerniške zgradbe so razporejene v naslednje kategorije: (i) količinske in kakovostne; (ii) glede na skladenjsko okolje primerniškega morfema; (iii) glede na skladenjsko okolje »družabnika«, t.j. stavčnega člena, ki ustreza izraženemu delu primerniškega dopolnila. Proučevana skladenjska okolja so osebek, tožilniški predmet, dajal-niški predmet, predložni predmet, prislovno določilo kraja, prislovno določilo časa, prislov in stavčni predikat. Primerjava vezniških in predložnih primerniških zgradb pokaže, da je distribucija slednjih bolj omejena in od govorca do govorca bolj variira. Vezniške primerniške zgradbe so sprejemljive v skoraj vseh kombinacijah zgoraj omenjenih parametrov. Izjema so kakovostne primerniške zgradbe s prislovnim ali predikatnim družabnikom ter kakovostne primerniške zgradbe, pri katerih je primerniški morfem del družabnika. Edina absolutna zahteva do predložnih primerniških zgradb je, da je družabnik osebek ali premi predmet; vse ostale posplošitve o distribuciji so zgolj težnje.