Proceedings of the Third ACL-SIGSEM Workshop on Prepositions, pages 23–28,
Trento, Italy, April 2006. c©2006 Association for Computational Linguistics
Marked Adpositions 
 
S.A.M. Lestrade 
Radboud University Nijmegen 
S.Lestrade@let.ru.nl 
 
 
 
 
Abstract 
 
This paper discusses the partitive-genitive 
case alternation of Finnish adpositions. This 
case alternation is explained in terms of 
bidirectional alignment of markedness in 
form and meaning. Marked PP meanings are 
assigned partitive case, unmarked ones 
genitive case. 
 
1 Introduction1 
 
This paper explores the relation of spatial 
adpositions and case in Finnish. Two major 
classes of adpositions in Finnish can be 
distinguished, one combining with genitive case, 
the other with partitive case. A third, very small 
set can combine with both. The meaning of these 
latter adpositions depends on whether they are 
combined with partitive case or genitive case, as 
illustrated in the following examples: 
 
(1) juoks-i-mme  ympäri  kaupunki-a 
 run-PAST-1PL around  city-PART 
‘We were running around in the city’  
 
(2) juoks-i-mme kaupungi-n   ympäri 
 run-PAST-1PL  city-GEN         around 
 ‘We ran around the city’  
 
In Section 2 and 3, I will show what determines 
the markedness of adpositional phrases (PPs). In 
Section 4, I will give a short introduction of 
bidirectional OT, with which I will analyze the 
Finnish data in Section 5. Before ending with 
conclusions, I will shortly  discuss the role of 
                                              
1 Many thanks to Lotte Hogeweg, Helen de Hoop, Peter de 
Swart, and Joost Zwarts, as well as to three anonymous 
reviewers, for useful comments on an earlier version of this 
paper. I would especially like to thank Seppo Kittilä for 
discussion of the Finnish examples. Also, I gratefully 
acknowledge the Netherlands Organisation of Scientific 
Research (NWO) for financial support, grant 220-70-003, 
principal investigator Helen de Hoop (PIONIER-project 
“Case cross-linguistically”).  
word order (i.e., pre- vs. postpositional use of the 
adposition).  
 
2 Qualities of Adpositional Phrases: 
Markedness of Meaning 
 
There are different ways to diagnose 
markedness. De Hoop et al. (2003) mention 
amongst others order of acquisition (the 
unmarked form is acquired before the marked 
one), context (the unmarked interpretation of a 
form is the interpretation it gets in neutral 
context), syntactic/morphological complexity (a 
simple form or construction is less marked than a 
complex form or construction), and frequency 
(the unmarked option occurs more often than the 
marked one). In what follows I will use these 
criteria on properties of adpositions.  
Spatial meaning is usually divided into 
locational and directional meaning. The former 
could be said to be less complex than the second, 
since directional meaning by definition implies a 
change in place (cf. Helmantel 1998; Jackendoff 
1983). This difference in complexity is also 
reflected in the order of acquisition of spatial 
prepositions, as locatives are acquired before 
directionals (Bowerman and Choi 2001). I 
propose that some directional meaning can be 
even more complex. Compare the following 
schematizations of locative meaning (3a), simple 
directional meaning (3b/c), and complex 
directional meaning (3d). The squares represent 
the Ground, each  round the Figure at a different 
moment in time:2 
 
Figure 3. Schematizations of spatial meaning 
 
   
(3a) Locative   (3b) Simple directional 
meaning:  ‘besides’ meaning: ‘from’ 
                                              
2 The notions of Figure and Ground refer to the located 
object and the object of reference, respectively (cf. Talmy 
2001). 
23
 
 
   
(3c)  Simple directional (3d) Complex directional 
meaning: ‘around’ meaning: ‘around’ 
   
m 
Following Zwarts (1997, 2003a, 2003b), we can 
describe the relation between a Figure and a 
Ground by means of vectors. The simple 
directional meaning puts restrictions on either the 
direction (in case of a straight line (3b)) or the 
length (in case of a circle (3c)) of the vectors. 
Complex directional meaning imposes no 
restrictions whatsoever (3d). That the distinction 
between (3c) and (3d) is linguistically relevant 
can be seen in the following Ancient Greek 
example. In (4a), dia combined with the genitive 
case indicates a simple, straight line, whereas the 
combination with accusative case in (4b) refers 
to a complex directional meaning, in which the 
Ground is almost completely covered by the 
path: 
 
(4a)  di’ mou   khálkeon 
DIA shoulder:GEN bronze:N/A 
égkhos  êlten  
spear:N/A go:AOR.3SG 
‘the spear of bronze went through his 
shoulder’ [Ilias 4.481-2, taken from 
Luraghi 2003: 168] 
 
(4b)  helixámenos   dià
 turn:PART.AOR.MID.NOM DIA  
bssas  
glen:ACC.PL.F 
‘turning around through the glens’ [Ilias 
17.283, taken from Luraghi 2003: 171] 
 
I argue that this difference in spatial complexity 
corresponds to a difference in semantic 
markedness, (4a) being less marked than (4b).  
An adposition can be non-concrete, or 
non-spatial. Solstad (2002) discusses the 
opposition spatial - non-spatial. Comparing 
spatial and causal meaning of the German 
preposition durch, he shows that in an unmarked 
syntactic position, that is, attached at the VP or 
IP level, the spatial meaning of a durch PP wins. 
In a marked position, that is, attached as CP 
adjunct, the causal meaning wins. Following the 
syntactic complexity criterion, the spatial 
meaning is unmarked, and the causal reading is 
marked.  
It is by definition impossible to locate an 
abstract notion in space. As adpositions prefer a 
spatial meaning, an abstract PP object is more 
marked than a concrete one. This markedness is 
reflected in the complexity of the form: 
Prepositions that typically take abstract objects 
are often morphologically more complex. 
Examples are the following Dutch prepositions, 
and their English equivalents: vanwege ‘because 
of’, betreffende ‘concerning’, met betrekking tot 
‘with regard to’, and aangaande ‘concerning’.  
 Concluding this section, we can say that 
a PP preferably has a simple, concrete, spatial 
meaning.  
 
3 Finnish Adpositions: Markedness 
of Form 
 
As we want to pair meaning with form, let us 
look at the markedness of the form of Finnish 
adpositional phrases. In Finnish in general, there 
are more postpositions than prepositions. 
Postpositions combine more often with genitive 
case than with partitive (54 vs. 10, respectively). 
The prepositions have a small preference for 
partitive case (9 vs. 6). Also for spatial 
adpositions, genitive case is assigned most, as 
can be seen in (5) and (6): 
 
(5) Spatial adpositions assigning genitive case: al 
‘under’, ede-/ete- ‘front’ jälke-/jälje- 
‘after’ luo- ‘to’, ‘by’, pää- ‘on’, sisä ‘in’, ta(ka)- 
‘back’, viere- ‘beside’.  
 
(6) Spatial adpositions assigning partitive case: 
koh- ‘towards’, pitkin ‘along’, ‘all over’, päin 
‘against’, vast- ‘against’, ‘in the opposite 
direction’. 
 
 Assuming that the most frequent form is 
unmarked (cf. the frequency criterion), we can 
conclude that genitive case is the unmarked form 
for the complement of the PP in Finnish.3  
 Before analyzing the Finnish data in 
Section 5, I will first introduce Bidirectional OT, 
the framework I am using. 
 
                                              
3 Another motivation for genitive being the standard case, 
though not pursued here, is the origin of adpositions. 
Postpositions deriving from nouns, like pää < ‘head’, 
usually take genitive case.  
24
 
 
4 Bidirectional OT  
 
Optimality Theory is a model of the system of 
the linguistic knowledge a speaker of a language 
has (cf. Prince and Smolensky 1993). The rules 
in this grammar are violable constraints, which 
can be in conflict with each other. Weak 
constraints can be violated in order to satisfy 
stronger ones. Each language has its own specific 
hierarchy of constraints. In the Generator 
possible output candidates (forms in case of 
production, and meanings in case of perception) 
are formed, that are evaluated in the Evaluator. 
The candidate that has the least serious violations 
of the constraints is the winner.  
Bidirectional OT gives a general 
procedure of optimization of the relation of form 
and meaning, simultaneously optimizing in both 
directions, from meaning to form, and from form 
to meaning.  Hence, BiOT evaluates form-
meaning pairs (Blutner et al. to appear: 90-91). 
In bidirectional OT a form-meaning pair is called 
super-optimal if there is no other super-optimal 
form-meaning pair with a different (i.e. less 
marked) form that expresses the same meaning 
better, and there is no other super-optimal form-
meaning pair with a different meaning that is a 
better interpretation of that same form. This 
yields two super-optimal form-meaning pairs, 
namely the unmarked form with the unmarked 
meaning, and the marked form with the marked 
meaning. The widely attested Markedness 
principle of Horn (1984) which states that 
marked forms in languages go with marked 
meanings (and unmarked forms with unmarked 
meanings) is thus accounted for by bidirectional 
OT.   
 
In section 2 we saw which properties of a PP are 
marked. An unmarked PP preferably expresses a  
simple, concrete, spatial meaning. Put into 
markedness constraints, we get:  
 
*COMPLEXITY: avoid semantic complexity 
BC (BE CONCRETE): have a concrete meaning 
 
In my analysis, simple directional meaning 
violates *Complexity once, whereas complex 
directional meaning yields a double violation. In 
section 3, genitive case was shown to be the most 
frequent, hence unmarked form. We can put this 
finding in a markedness constraint on the form of 
the PP: 
 
GEN: use genitive case on the object of an 
adposition.   
 
5 Analysis of the Data 
 
Now let us have a look at the Finnish data. The 
general distribution of spatial adpositions was 
given in (5) and (6) above. This general 
distribution seems to follow the opposition 
locative-directional: as ‘towards’, ‘along’, 
‘against’ are directional and the others (roughly) 
locational.4  
 
As illustrated in Tableau 1, directional meanings 
violate *Complexity, and partitive forms violate 
UG.5  
 
Tableau 1. The general distribution 
 <form, 
meaning> 
GEN BC *COMP
LEXITY 
 <gen, loc>    
 <gen, dir>   * 
 <part, loc> *   
 <part, dir > *  * 
 
Since the form-meaning pair <gen, loc> is the 
only candidate that satisfies all constraints it is 
the first super-optimal pair. In a second round of 
optimization the pairs <gen, dir> and <part, loc> 
fall out, because there is another form-meaning 
pair, namely <gen, loc>, with either a less 
marked meaning that is expressed by the same 
form, or with a less marked form that expresses 
the same meaning. This yields the second super-
optimal form-meaning pair, namely <part, dir> 
which combines the marked form with the 
marked meaning.  
The unmarked locational meaning 
corresponds to the unmarked genitive form, and 
the marked directional one to the marked 
partitive. Thus, the general distribution of 
Finnish adpositions is accounted for.  
But, as mentioned before, besides these 
two groups, Finnish has a third set of adpositions 
combining with both cases: 
 
                                              
4 The following test can be used to discriminate locatives 
from directional adpositions: the former can be combined 
with the verb to be, the latter with verbs of motion, such as 
to go. 
5 For further explanation on the use of tableaux, I refer to 
Blutner et al. (to appear) 
25
 
 
(7) Adpositions taking both genitive and partitive 
case: ympäri- ‘around’, keske-‘middle’, lähe- 
‘near’, yli(-) ‘over’. 
 
The meaning of such adpositions in combination 
with genitive case is different from that of a 
combination with partitive case. In what follows, 
I will give an account of the form-meaning pairs 
involving these adpositions. First, let us look at 
the alternation of ympäri ‘around’ shown in 
examples (1) and (2), repeated below for 
convenience.  
 
(1) juoks-i-mme    ympäri  kaupunki-a 
 run-PAST-1PL   around  city-PART 
‘We were running around in the city’  
 
(2) juoks-i-mme kaupungi-n   ympäri 
 run-PAST-1PL  city-GEN         around 
 ‘We ran around the city’  
 
In (1) around indicates a complex, random path 
comparable to the Greek example (4b), 
schematized in (3d). In (2) however, around 
describes a circle with the city as its centre (cf. 
(3c)). The constraint on complexity is playing a 
role in this alternation, as is illustrated in Tableau 
2: 
 
Tableau 2. Ympäri 
 <form, 
meaning> 
GEN BC *COMP
LEXITY 
 <gen, circle>   * 
 <gen, criss-
cross> 
  ** 
 <part, circle> *  * 
 <part, criss-
cross> 
*  ** 
 
As expected, the genitive case has the unmarked 
simple meaning, whereas the partitive case 
combines with the complex one. Simple 
directional meaning violates *Complexity once, 
as it is one step more complex than locational 
meaning; complex directional meaning yields a 
double violation.  
The alternation of keskellä ‘in the 
middle’ can be explained along the same lines. 
Again, genitive case combines with the 
unmarked meaning in (9), whereas the partitive 
is assigned the marked notion ‘all over’ in (8): 
 
(8) lelu-t  o-vat  
 toy-PL.NOM be.PRES-3PL  
 keskellä lattia-a 
in.the.middle.of floor-PART 
‘The toys are all over the floor’ 
 
(9) lelu-t  ovat   
 toy-PL.NOM be.PRES-3PL  
lattia-n  keskellä 
floor-GEN in.the.middle.of 
‘The toys are in the middle of the floor’ 
(lit. at the centre) 
 
The third adposition to deal with is lähellä 
‘near’. For the previous adpositions we saw that 
the case alternation was only used for an 
extended meaning. Rather unexpectedly, for 
lähellä partitive case is also possible within the 
simple spatial domain, as shown in (10).6 
 
(10) auto    on  talo-n 
 car.NOM   be.SG.PRES house-GEN 
 lähellä  ( / talo-a) 
near house-PART 
‘The car is near the house’  
 
As soon as the concreteness constraint BC 
becomes of importance, however, partitive case 
is used, as is clear from (11): 
 
(11) ole-mme lähellä ratkaisu-a 
 be.PRES-1PL near solution-PART 
 ‘We are close to a solution’ 
 
In this example, the constraint BC comes into 
play. The abstract meaning combines with 
partitive case, whereas the concrete one can have 
both cases. This is shown in the following 
tableau, in which the unpredicted pair is 
indicated with a question mark: 
 
Tableau 3. Lähellä  
 <form, 
meaning> 
GEN BC *COMP
LEXITY 
 <gen, concr>    
 <gen, abstr>  *  
? <part, concr> *   
 <part, abstr> * *  
 
 
                                              
6 Indeed, it is not always the case that there is a clear one-to-
one correspondence of form and meaning. For further 
discussion on optionality I refer to De Hoop et al. (2003). 
26
 
 
Finally, let us turn to yli ‘over’. The 
preposition over has extensively been studied for 
English (cf. Lakoff 1987; Brugman 1988; Dewell 
1994). The more central meanings are the 
locative ‘be in the space above something’ and 
‘be on the other side of something’, and the 
directional ‘starting from one side of something, 
go to the other side via the covering space’ and 
‘starting from one side of something, go to the 
other side via its (top) surface’. The reading ‘to 
exceed something’ that ‘over’ also has, is further 
removed from the central meaning. In Finnish we 
see that the more central notions are covered by 
the genitive case, as is shown in (12) and (13): 
 
(12) lintu     lentää  puu-n bird.nom  
fly.pres.3sg tree- GEN 
 yli 
over 
‘The bird flies over the tree’ 
 
(13) mies  kävelee           kukkula-n  yli 
man  live.pres.3sg  hill-GEN over 
‘The man lives over the hill’ 
 
Only within the extended ‘Excess’ meaning, the 
partitive case can occur, as can be seen in (14): 
 
(14) tämä-n   auto-n    hinta              
this-GEN car-GEN  price.NOM    
on                  yli      2500  euro-a 
be.3SG.PRES  over   2500  euro-PART  
 ‘The price of this car is higher than 
2500 Euro’ 
 
It is still possible to use genitive case here, but 
the 2500 Euro is pragmatically different for both 
constructions. With partitive case, it is an 
arbitrary amount, whereas it is precisely the 
amount I intended to spend on the car with the 
genitive. Here, the prominence of the amount 
makes it more concrete.  
 
Tableau 4. Yli ‘exceeding’ 
 <form, 
meaning> 
GEN BC *Compl
exity 
 <gen, concr>    
 <gen, abstr>  *  
 <part, concr > *   
 <part, abstr > * *  
  
 
 
6 About Word Order 
 
As noted above, there is a correlation between 
the syntactic position and case in the Finnish 
adpositional system. How do we know that it is 
not the syntactic position that determines the 
markedness, instead of case (cf. (1) and (2))? 
First, the correlation should not be taken as a 
strict rule, as both postpositional and 
prepositional lähellä can govern partitive case: 
 
(15) auto     on  talo-a 
car.NOM  be.SG.PRES house-PART    
lähellä (/ talo-a) 
near house-PART 
‘The car is near the house’  
 
Other examples of the indifference of the 
partitive with respect to its position are kohti and 
ennen: 
 
(16)  auto     tul-i  
 car.NOM  come-3SG.PAST  
minu-a          kohti (/ minu-a) 
1SG-PART  towards 1SG-PART 
 ‘The car was coming towards me’ 
 
(17)  hän     saapu-i 
 s/he.NOM   arrive-3SG.PAST 
 minu-a   ennen (/ minu-a) 
 1SG-PART before 1SG-PART 
 ‘S/he arrived before me’ 
 
A more complex example is given in (18) and 
(19). Sometimes, the non-standard combination 
of partitive with a postposition yields a 
metaphorical reading, whereas the standard 
combination with a preposition yields a concrete 
reading: 
 
(18)  tämä      aihe  on
 this.NOM  topic.NOM be.3SG.PRES 
 minu-a  lähellä 
 1SG-PART near/close 
 ‘This topic is close to my heart’  
 
(19) tämä  on   
 this.NOM be.SG.PRES  
lähellä minu-a 
near 1SG-PART 
 ‘This is close to me’ 
 
Hence, markedness is not just a matter of the 
overall less frequent use of a preposition – 
otherwise (19) should have the metaphorical 
27
 
 
reading. Apparently, there is an additional 
constraint on the form which does not say “use 
postpositions rather than prepositions”, but “use 
postpositions for genitive case and prepositions 
for partitive”. 
 
7 Conclusion  
 
In this paper I used a bidirectional OT approach 
to account for the case distribution in the Finnish 
spatial adpositional system. I argued that 
adpositions in general can have a more or less 
marked meaning. In Finnish this markedness is 
reflected in the assignment of case. Partitive 
case, as the marked form, is assigned to marked 
adpositions, and the unmarked genitive case is 
used for the unmarked ones. Within case 
alternating adpositions, this same tendency is 
found, albeit along different lines of markedness.  
 
References  
Blutner, R., de Hoop, H., and Hendriks, P. (to 
appear). Optimal Communication. CSLI 
 
Bowerman, M. and Choi, S. (2001) “Shaping 
meanings for language: universal and 
language-specific in the acquisition of 
spatial semantic categories”. In: M. 
Bowerman and S. C. Levinson (Eds) 
Language acquisition and conceptual 
development, 475-511. Cambridge: 
Cambridge University Press.  
 
Brugman, C. (1988) The Story of Over. New 
York: Garland Publishing, Inc. 
 
Dewell, R. (1994) Over again: Image-schema 
transformations in semantic analysis. 
Cognitive Linguistics 5 (4): 351-380.  
 
Helmantel, M. (1998) Simplex adpositions and 
vector theory. The Linguistic Review 15: 
361-388. 
 
de Hoop, H., Haverkort, M. and van den Noort, 
M. 2003. “Variation in form versus 
variation in meaning”. Lingua 114: 
1071-1089 
 
Jackendoff, Ray. 1983. Semantics and cognition. 
Cambridge, MA: MIT Press 
 
Lakoff, G. 1987. Women, fire, and dangerous 
things : what categories reveal about the 
mind. Chicago: University of Chicago 
Press 
 
Prince, A., & Smolensky, P. 1993. Optimality 
theory: Constraint Interaction in 
Generative Grammar. Rutgers Center 
for Cognitive Science: Technical Report 
RuCCSTR-2. Published in 1993 by 
Malden, MA: Blackwell. 
 
Solstad, T. 2002 “Optimal Interpretations and 
Harmonic Alignment: Towards the 
Constrained Lexicon”. Talk held at the 
workshop on Optimality Theory and 
Pragmatics, ZAS Berlin, June 8-10.  
 
Talmy, L. 2001 Toward a cognitive semantics. 
London: MIT Press. 
 
Zwarts, J. 2003a Lexical Competion: ‘Round’ in 
English and Dutch. In P. Dekker and R. 
van Rooij (Eds.) Proceedings of the 
Fourteenth Amsterdam Colloquium. 
Amsterdam: ILLC: 229-234 
 
Zwarts, J. 2003b Vectors across Spatial 
Domains: From Place to Size, 
Orientation, Shape, and Parts. In: E. van 
der Zee, & J. Slack (Eds.) Representing 
Direction in Language and Space. 
Oxford: Oxford UP: 191-208. 
 
Zwarts, J. 1997 Vectors as Relative Positions: A 
Compositional Semantics of Modified 
PPs. Journal of Semantics 14: 57-86. 
