The Perplexity Of Dat Nom Verbs In Icelandic
The Perplexity of Dat-Nom Verbs in
Icelandic
Jo´hanna Barðdal
Barðdal, J. 2001. The perplexity of Dat-Nom verbs in Icelandic. Nordic Journal of
Linguistics, 24, 47–70.
This paper presents data from Modern Icelandic of a small group of Dat-Nom verbs
which select for two arguments: a Dative human argument and a Nominative stimulus.
When applying independently established subjecthood tests on these arguments it turns
out that both pass the tests, i.e. both arguments can behave like subjects and like objects,
but not at the same time. An examination of the lexical meanings of these predicates
reveals that they can be divided into the following main groups: Emotive verbs,
Perception Verbs, Cognition Verbs, Verbs of Attitude and Benefactive verbs. A
Construction Grammar analysis is proposed, assuming two different syntactic
constructions to exist in Icelandic, i.e. a Dat-Nom construction versus a Nom-Dat
construction. It is argued that the occurrence of these predicates in the two constructions
follows directly from their conceptual causal structure. Furthermore, the choice of
subject seems to be contextually determined, i.e. the more topical argument takes on the
subject function. The relation between the two constructions, i.e. the Dat-Nom and the
Nom-Dat, seems to be like the relation between an ordinary transitive construction and
its topicalization construction, in that when the lower argument is “topicalized” to first
position the other construction is activated, hence the ordinary topicalization
construction in Icelandic is not as readily available to these verbs as the other
construction is.
Jo´hanna Barðdal, Department of Scandinavian Linguistics, Lund University, Helgona-
backen 14, S-223 62 Lund, Sweden. Department of Linguistics, University of Man-
chester, M13 9PL Manchester, U.K. E-mail: johanna.barddal@nordlund.lu.se or
johanna.barddal@man.ac.uk
1. INTRODUCTION
A peculiarity of Icelandic, more or less unnoticed in the syntactic litera-
ture, is the special subtype of psych-verbs, which can occur in two syn-
tactic frames, i.e. both as fear-verbs and as frighten-verbs.1 A verb’s
occurrence in two syntactic frames is not particularly noteworthy consid-
ering that for instance the verb give can occur both as a ditransitive ‘I
gave him the book’ and with a prepositional variant ‘I gave the book to
him’. The psych-verbs to be discussed here differ radically from such
examples since they include reordering of grammatical functions:
(1) a. Hentar ÞETTA ÞE
´ R?
pleases this (nom-subj) you (dat-obj)
‘Does this please you?’
b. Hentar ÞE
´ R ÞETTA?
pleases you (dat-subj) this (nom-obj)
‘Are you pleased with this?’2
47
48
In (1a) the nominative stimuli þetta ‘this’ is the grammatical subject, as
is obvious from the fact that it inverts with the verb in questions, while
in (1b) it is the dative human argument þe´r ‘you’ which is the grammati-
cal subject, since it also inverts with the verb in questions (see Thra´ins-
son (1979), Berno´dusson (1982), Zaenen, Maling & Thra´insson (1985),
Sigurðsson (1989, 1992) and Jo´nsson (1997–98) on oblique subjects in
Icelandic, and Berno´dusson (1982), Zaenen, Maling & Thra´insson
(1985), Sigurðsson (1990–91) and Maling & Jo´nsson (1995) on nomina-
tive objects in Icelandic). From the examples in (1) we can gather that
both arguments of henta can occur as a subject and both can occur as an
object, though not at the same time, of course.
These verbs were rst discovered by Berno´dusson in 1982 and have
since received scanty attention (see a note in Zaenen, Maling & Thra´ins-
son 1985: 469, and a short mention in Jo´nsson 1997–98: 14–15. For the
theoretical implications of such verbs in historical linguistics, see Allen
(1995), Ro¨gnvaldsson (1996a: 65) and Barðdal (1997: 44–45 and 1998).
The reason for this may be that most modern syntacticians working on
Icelandic have carried out their research within a transformational tradi-
tion, in which it is not clear how to account for the behaviour of these
psych-verbs in a straightforward way. It is a fact that within most con-
temporary syntactic theories that a given argument of a predicate is the
subject and that it is always the subject, provided that the diathesis has
not been altered, but not that subject status can be subject to variation
within the active diathesis. However, implementing this fact of the Ice-
landic language into Construction Grammar entails no complications at
all because Construction Grammar has a uniform way of representing all
grammatical knowledge, namely as a form-meaning correspondence, i.e.
as a construction (see section 3 below).
Before I proceed to the main body of this paper, an overview of the
constructions Icelandic psych-verbs occur in is in place, and accordingly
a closer speci cation of our research object.
Psych-verbs in Icelandic can be found in the following constructions:
(2) a. SubjNom/hum V ObjAcc/stim E´g hrœðist hunda
I fear dogs
b. SubjNom/stim V ObjAcc/hum Hundar hrœða mig
Dogs frighten me
c. SubjAcc/hum V ObjAcc/stim Mig dreymdi o¨mmu
I dreamt of grandma
d. SubjAcc/hum V (loc)
Mig verkjar õ´ magann
I ache in the stomach
49
e. SubjDat/hum V AdjAgr¡(loc) Me´r er illt õ´ maganum
I am ill in the stomach
f. SubjNom/hum V AdjAgr‡
E´g er illur
I am angry
g. SubjDat/hum V ObjNom/stim Me´r lõ´kar Guðmundur
I like Guðmundur
h. SubjNom/stim V ObjDat/hum Þetta hentar me´r
This pleases me
These are the most prominent constructions Icelandic psych-verbs occur
in. The rst is clearly an instance of a more general construction, i.e. the
transitive construction, the last two case patterns are also utilized by
passives of ditransitive verbs in Icelandic, (f) is common for depictive
statives, while the remaining constructions are more or less reserved for
psych-verbs. In this paper I discuss only the last two examples: the one
with the Dative human argument subject and the Nominative stimulus
object and conversely the one with the Nominative stimulus subject and
the Dative human argument object.
This paper is organized as follows: In the next section I present the
syntactic tests of subjecthood and provide evidence that both arguments
of our psych-verbs pass all the relevant tests. In section 3, I present the
semantics of the data and argue for a Construction-based analysis of it.
Section 4 is a summary.
2. SYNTACTIC SUBJECTHOOD
The concepts of subject and object are not unproblematic concepts.
Within traditional grammar the subject has been de ned as the argument
carrying the nominative case, and the object as the argument carrying
the accusative, dative or genitive case of transitive verbs. Modern syn-
tactic theories have emphasized the need to look at the syntactic beha-
viour of the arguments in question, and not just their morpho-syntactic
properties, in order to determine their syntactic status. For Icelandic this
has led to a de nition of the subject based solely on syntactic properties,
since it has been shown that morpho-syntactic properties, such as mor-
phological case and subject-verb agreement, do not correlate with the
syntactic properties subjects have in Icelandic, though they correlate
with each other (see Sigurðsson 1990–91). Hence, syntactic subjects in
Icelandic can carry nominative, accusative, dative and genitive case.
Positing a universal category ‘subject’ seems to be theoretically impossi-
ble on grounds of the differences in the behaviour of these arguments in
different languages (see Croft 2001: ch. 4). This is even true for closely
related languages, such as Icelandic and German, since a comparison of
50
the two languages has revealed that out of 13 tests suggested for the two
languages only four tests coincide for both Icelandic and German (see
Barðdal, in prep.). These are only some of the problems that follow from
the assumption that a universal and a uniform category of subjects
exists. Instead, either a language speci c category of subjects has to be
posited or a more radical solution which would entail the abandonment
of the theoretical concept of subjects and objects altogether (see Croft
2001: ch. 4 and Barðdal, in prep.). However, I will not pursue this argu-
ment here. For the purpose of this paper it is suf cient that all left-most
arguments of transitive verbs in ordinary argument linking constructions
in Icelandic show a uniform behaviour in that they pass all the tests that
have been used as subject criteria in Icelandic.
For Icelandic, the following have been proposed as subject properties
(Sigurðsson 1989, 1992, Ro¨gnvaldsson 1996, Zaenen, Maling &
Thra´insson 1985). I refer the interested reader to the above-cited studies
for examples showing that objects behave differently from subjects in
Icelandic with regard to all the properties in (3):
(3) First position in declarative clauses
Subject-verb inversion
First position in subordinate clauses
Conjunction Reduction
Clause-bound re exivization
Long-distance re exivization
Subject-to-object raising
Subject-to-subject raising
Control in nitives
I now present data that show that there is a group of psych-verbs in Ice-
landic of which both arguments pass a number of known tests of sub-
jecthood and thereby t into the category subject.
2.1. Word order and distribution
Both arguments of the verb henta ‘please’ can occur in the position pre-
ceding the verb, both can invert with the nite verb when something else
is topicalized, and both can occupy rst position in subordinate clauses:
(4) a. ÞETTA hefur alltaf hentað ME´R.
‘This has always pleased me.’
b. ME´R hefur alltaf henta ÞETTA.
me has always pleased this
‘I have always been pleased with this.’
51
(5) a. He´ðan af mun ÞETTA alltaf henta ME´R.
from now on will this always please me
b. He´ðan af mun ME´R alltaf henta ÞETTA.
from now on will me always please this
‘From now on I will always be pleased with this.’
(6) a. E´g veit að ÞETTA mun henta ME´R.
I know that this will please me
b. E´g veit að ME´R mun henta ÞETTA.
I know that me will please this
‘I know that I will be pleased with this.’
Both arguments of henta can be reduced in Conjunction Reduction:
(7) a.
ÞETTA FYRIRKOMULAG er a´gœtt og__mun lõ´ka henta ME´R
a´gœtlega.
this arrangement (nom) is ne and will also please me (dat) ne
b.i ME´R lõ´ður vel og__mun o¨rugglega henta ÞETTA a´gœtlega.
I (dat) am ne and will surely be pleased with this (nom)
b.ii E´G er a´nœgð og__mun o¨rugglega henta ÞETTA a´gœtlega.
I (nom) am happy and will surely be pleased with this (nom)
The deleted Dative human argument in (7b.i) is not deleted because it
has the same morphological case as the coordinated constituent. In
(7b.ii) we nd that the dative can also be deleted on identi cation with a
nominative. Accordingly, both arguments of henta can behave like sub-
jects when their word order and distributional properties are considered.
2.2. Re exivization
In Icelandic, we nd both Clause-bound re exivization and Long-dis-
tance re exivization, the former is bound within the simple clause and
the latter is bound across sentence boundaries. Consider rst the Clause-
bound re exivization:
(8) a. HANNi hentar KONUNNI SINNIi a´gœtlega.
he pleases his wife ne
‘He pleases his wife well.’
b. HONUMi hentar FYRIRKOMULAGIÐ SITTi a´gœtlega.
him pleases his arrangement ne
‘He is pleased with his arrangement.’
Regarding Long-distance re exivization, only the Dative human argu-
52
ment can function as an antecedent for a re exive, a property mostly
con ned to subjects:
(9) a. HONUMi hentar að fyrirkomulagið SITTi verði rœtt.
he (dat) suits that arrangement himself become discussed
‘It suits him that his arrangement will be discussed.’
b. *HANN hentar að …
he (nom) suits that
‘He is suitable that …’
This is due to the fact that a main clause with the Nominative stimulus
as a subject doesn’t select for the type of subordinate clause that can
contain a re exive. The same is true for other psych-verbs in Icelandic
with a similar argument structure, as for instance tru a ‘bother’, which
is an ordinary nom-acc frighten/please verb, where only the Nominative
stimulus can behave as a syntactic subject:
(10) *HANN tru ar að …
he (nom) bothers that
The reluctancy of the Nominative stimulus to function as an antecedent
for a Long-distance anaphor is therefore expected and cannot disqualify
it as a subject.
2.3. Raising and control constructions
There are basically two kinds of “raising” constructions which have
been used to measure subjecthood, Subject-to-object raising (AcI in ni-
tivals) and Subject-to-Subject raising (D/NcI in nitivals). In the former,
the subject of a lower clause is “raised” to the object position of the
matrix clause, while in the latter the subject of the lower clause is
“raised” to the subject position of the matrix clause. Consider the fol-
lowing examples with our verb henta:
(11) a. Hann telur FYRIRKOMULAGIÐ henta se´r.
he assumes the arrangement (acc) please himself
‘He assumes that the arrangement will be pleasing for him.’
b. Hann telur SE´R henta fyrirkomulagið.
he assumes himself (dat) please arrangement
‘He assumes that he will be pleased with the arrangement.’
53
(12) a. FYRIRKOMULAGIÐ virðist henta HONUM a´gœtlega.
the arrangement (nom) seems to please him ne
‘The arrangement seems to please him.’
b. HONUM virðist henta FYRIRKOMULAGIÐ a´gœtlega.
him seems to please the arrangement (nom) ne
‘He seems to be pleased with the arrangement’.
Also we nd both arguments of henta as PRO, or the unexpressed argu-
ment, in Control clauses:
(13) a. Það er a´gœtt að henta ÞETTA.
it is good to [PRO] suit this (nom)
‘It is good to be pleased with this.’
b. Það er a´gœtt að henta HONUM.
it is good to [PRO] suit him (dat)
‘It is good to please him.’
Accordingly, both arguments of the verb henta, the Dative human argu-
ment and the Nominative stimulus, can occur as the syntactic subject in
Control Constructions.
To summarize, in this section I have demonstrated that some psych-
verbs in Icelandic vary in their syntactic structure in such a way that
either of their arguments, the Nominative stimulus or the Dative human
argument, can be realized as the syntactic subject and as the syntactic
object.3,4 As far as I know, for the Germanic languages this has only
been noticed in Icelandic and Faroese (see Berno´dusson 1982: 37–38 for
Icelandic and Barnes 1986: 33 ff. for Faroese). There are indications,
though, of similar behaviour of psych-verbs in previous stages of Ger-
manic (see Barðdal (1997: 44–45 and 1998) for Old and Middle Scandi-
navian, and Allen (1995) for Old and Middle English). Since such verbs
seem to be more common in (various stages of) Germanic5 than at rst
sight, and might be found in other languages of the world, this phenom-
enon is something that a theory of grammar has to be able to account
for. We will now turn to that.
3. THE ANALYSIS
3.1. The Data
The verbs exhibiting the dual nature, displayed in section 2, are not a
homogeneous group of verbs. After a thorough examination of a list
containing most predicates that select for Oblique subjects in Icelandic
(see Jo´nsson 1998), with the two rst subjecthood tests in (3) above as a
tool,6 I was able to discern the verbs presented in (14):
54
(14) berast ‘receive’, birtast ‘appear’, bragðast ‘taste’, duga ‘suf ce’,
dyljast ‘be not aware of sth’, endast ‘last’, falla vel ‘like, please’, fara
vel ‘suit’, fylgja ‘accompany’, gagnast ‘be of use to’, glatast ‘be lost to’,
greypast ‘stuck in sby’s mind’, henta ‘please, suit’, hverfa ‘be lost to
sby’, hœfa ‘suit’, ny´tast ‘be of use to’, nœgja ‘suf ce’, opinberast
‘appear in a vision’, passa ‘please, suit’, reynast ‘prove, turn out to’,
possibly sa´rna ‘get one’s feelings hurt’,7 smakkast ‘taste’, so´ma ‘be
proper, suit’, sœkjast vel ‘go well/badly’, sœma ‘be proper, suit’, vitrast
‘appear in vision’, þo´ knast ‘please, suit’.
(14) above lists the simple verbs in Icelandic that exhibit the dual nature,
while (15) below lists the complex predicates with the same property:
(15) berast õ´ hendur ‘receive’, falla õ´ geð ‘like, please’, falla e-ð õ´ skaut
‘receive’, falla verk u´r hendi ‘fail to do sth’, fara e-ð vel u´r hendi ‘do
sth well’, festast õ´ minni ‘stick in sby’s memory’, hrjo´ta af vo¨rum ‘let
words slip’, hverfa vero¨ldin ‘sleep for a while, t’, koma að gagni ‘be
of use to’, koma a´ o´vart ‘surprise’, koma við ‘be of sby’s business’,
koma spa´nskt fyrir sjo´nir ‘ nd sth strange’, koma õ´ koll ‘get in trouble’,
leika õ´ lyndi ‘go well’, lõ´ða ekki u´r hug/minni ‘be unable to forget’, ligg-
ja e-ð a´ hjarta ‘be anxious’, ratast a´ munn ‘accidentally speak’, renna
til rifja ‘cut to the quick’, renna kalt vatn milli skinns og ho¨runds ‘be
terri ed’, reynast er tt ‘be dif cult’, standa til boða ‘be offered sth’,
standa fyrir þrifum ‘hampered by sth’, stõ´ga til ho¨fuðs ‘go to sby’s
head’, svella õ´ munni ‘be exaggerated’, vaxa e-ð õ´ augum ‘ nd sth more
dif cult than it really is’, veitast auðvelt ‘ nd sth easy’, vera auðvelt ‘be
easy’, vera augljo´st ‘be obvious’, vera e-ð a´ mo´ti skapi ‘dislike’, vera
að kostnaðarlausu ‘be free’, vera allar bjargir bannaðar ‘be in a hope-
less situation’, vera allir vegir fœrir ‘be able to do anything’, vera dy´r-
keypt ‘suffer, pay dearly for sth’, vera eðlislœgt ‘sth comes naturally for
sby’, vera efst õ´ huga ‘think of sth more than anything else’, vera eigin-
legt ‘sth comes naturally for sby’, vera ekkert að vanbu´naði ‘be fully
prepared’, vera e-ð fjarri skapi ‘dislike’, vera e-ð fyrir bestu ‘be best for
sby’, vera e-ð fyrir mestu ‘be most important for sby’, vera er tt ‘be dif-
cult’, vera framandi ‘be alien to sby’, vera frja´lst ‘be free to’, vera
hollt ‘be healthy for sby’, vera hugleikið ‘be important to sby’, vera õ´
blo´ð borið ‘have a natural talent for sth’, vera õ´ fersku minni ‘remember
vividly’, vera (o´)gerlegt ‘be (im)possible’, vera glatað ‘be lost to sby’,
vera e-ð õ´ lo´fa lagið ‘be easy for sby’, vera õ´ sja´lfsvald sett ‘have per-
mission to decide for oneself’, vera kœrt ‘be dear’, vera e-ð kœrkomið
‘be welcome’, vera e-ð (mikil) kvo¨l ‘be painful for sby’, vera ljo´st ‘be
obvious’, vera lju´ft ‘be a pleasure’, vera mikilvœgt ‘be important for
sby’, vera minnisstœtt ‘remember vividly’, vera e-ð mo´tfallið ‘be against
55
sth, dislike’, vera e-ð ofraun ‘be too dif cult’, vera o´kunnur ‘be
unknown to sby’, vera ofvaxið ‘be beyond sby’s power’, vera ofviða ‘be
too dif cult’, vera o´heimilt ‘be prohibited’, vera o´kleift ‘be impossible’,
vera o´kunnugt ‘be unknowing about sth’, vera (o´)mo¨gulegt ‘be (im)pos-
sible for sby’, vera o´skiljanlegt ‘be incomprehensible’, vera e-m ra´ðga´ta
‘be a mystery to sby’, vera tamt ‘be natural for sby’, vera til ama ‘be
disturbing’, vera til efs ‘doubt sth’, vera um megn ‘be too dif cult for
sby’, vera uppo¨rvun ‘be an encouragement’, vera velkomið ‘be welcome
to’, vera þvert um geð ‘dislike’, vera þyrnir õ´ augum ‘be a thorn in
sby’s side/ esh’, verða að falli ‘cause a downfall’, verða að fo´take i ‘be
a hindrance’, verða að go´ðu ‘be good for sby’, verða til happs ‘be off
luck’, verða til lõ´fs ‘survive’, vera til lista lagt ‘have a talent’, vinnast e-
ð vel ‘make good progress’.
Note that the complex predicates are of two types: rstly, the same as
the simple verb but followed by an attribute, which is often a locative
(or a bodily) speci cation (16a–b). And secondly, a copula with an
adjective in the default form or an attributive complement of some kind
(17a–b):
(16) a. Me´r fellur þetta vel.
me (dat) falls this (nom) well
‘I like this.’
b. Me´r fellur þetta vel I´ GEÐ.
me (dat) falls this (nom) well in mind
‘I like this.’
(17) a. Me´r er þetta LJU´FT.
me (dat) is this (nom) dear
This is a pleasure for/to me.’
b. Me´r er það mjo¨g TIL EFS.
me (dat) is that (nom) very in doubt
‘I doubt that.’
An investigation of the lexical meaning of our verbs reveals that many
of them are synonyms or near-synonyms:
(18) like: falla vel, falla õ´ geð
dislike: vera e-ð a´ mo´ti skapi, vera e-ð fjarri skapi, vera e-ð mo´t-
fallið, vera þvert um geð
please, suit, t: fara vel, henta, hæfa, passa, so´ma, sæma, þo´knast
be(come) hurt: sa´rna, renna til rifja
be terri ed: renna kalt vatn milli skinns og ho¨runds
56
be anxious: liggja e-ð a´ hjarta
doubt: vera til efs
suffer: vera dy´rkeypt
surprise: koma a´ o´vart
nd dif cult or painful: reynast er tt, vera er tt, vaxa e-ð
õ´ augum, vera allar bjargir bannaðar, standa fyrir þrifum, vera e-ð
ofraun, vera ofvaxið, vera ofviða, vera o´kleift, vera o´mo¨gulegt,
vera um megn, vera dy´rkeypt, vera e-ð (mikil) kvo¨l, vera til ama
og leiðinda
nd easy: leika õ´ lyndi, vera allir vegir færir, veitast auðvelt, vera
auðvelt, vera e-ð õ´ lo´fa lagið, vera ekkert að vanbu´naði
be natural for sby: vera eðlislægt, vera eiginlegt, vera õ´ blo´ð
borið, vera tamt, vera õ´ lo´fa lagið, vera til lista lagt
remember: greypast, festast õ´ minni, lõ´ða ekki u´r hug/minni, vera
hugleikið, vera õ´ fersku minni, vera minnisstætt
be (un)aware of sth: dyljast, vera framandi, vera o´kunnur, vera
o´kunnugt, vera augljo´st, vera ljo´st
be important/dear/a pleasure: vera e-ð fyrir mestu, vera kært,
vera lju´ft, vera mikilvægt, vera efst õ´ huga
be a problem: koma õ´ koll, vera dy´rkeypt, vera þyrnir õ´ augum,
verða að fo´take i, verða að falli
be strange/incomprehensible: koma spa´nskt fyrir sjo´nir, vera
o´skiljanlegt, vera e-m ra´ðga´ta
be good for sby: vera e-ð fyrir bestu, vera hollt, verða að go´ðu,
verða til happs, verða til lõ´fs, vera uppo¨rvun
appear (in a vision): birtast, opinberast, vitrast
sleep: hverfa vero¨ldin
taste: bragðast, smakkast
accidentally speak: hrjo´ta af vo¨rum, ratast a´ munn
receive: berast, berast õ´ hendur, falla e-ð õ´ skaut, vera að kost-
naðarlausu
be lost: glatast, vera glatað, hverfa e-ð,
be free to/offered to: vera frja´lst, vera heimilt, vera õ´ sja´lfsvald
sett, vera velkomið, vera e-ð kærkomið, standa til boða
be suf cient: duga, endast, nægja
accompany: fylgja
turn out to be: reynast
be of use to: gagnast, koma að gagni, ny´tast,
make a good/bad progress: sækjast vel, fara e-ð vel u´r hendi, fal-
la verk u´r hendi, vinnast e-ð vel
be of sby’s business: koma við
go to sby’s head: stõ´ga til ho¨fuðs
be exaggerated: svella õ´ munni
57
Obviously, our group of verbs is both a semantically de ned group and
a lexically de ned group. It is semantically de ned since many of the
verbs are synonymous; for instance, all simple verbs in Icelandic mean-
ing ‘please, suit’ and ‘be of use to’ seem to be included, and the group
is lexically de ned since its members cannot be predicted from any gen-
eral semantic or syntactic rule. It is possible, though, that at one point or
another in the history of Icelandic there was a productive syntactic/
semantic rule that generated these examples, but such a rule is certainly
not productive in today’s Icelandic, and their dual behaviour with
respect to choice of syntactic functions is not a general property of all
dative-nominative verbs in Icelandic (see example (25) below).8
Examining the lexical meanings of our verbs, we nd that some of
them are typical Experiencer verbs, like the ones with the meanings
‘like’, ‘dislike’, ‘please’, ‘be hurt’, ‘be terri ed’, ‘be anxious’ and ‘suf-
fer’, whereas others, contrary to what is usually assumed for verbs
selecting a dative subject, should rather be classi ed as Cognition verbs,
such as the verbs meaning ‘remember’, ‘be (un)aware of sth’, and
‘appear in a vision’. Verbs meaning ‘taste’ can be classi ed as Percep-
tion verbs. Yet others denote the attitude of the speaker; for instance,
verbs meaning ‘ nd dif cult’, ‘ nd easy’, ‘ nd strange’, ‘be important’,
‘be a problem’, ‘be natural for sby’. Also, some of the verbs should be
categorized as Benefactive verbs; for instance, the verbs meaning
‘receive’, ‘appear in a vision’, be offered to’, ‘be of use to’, ‘be suf -
cient to’ and ‘accompany’. On the basis of this I suggest a division of
our predicates into the following ve classes: Emotive verbs, Cognition
verbs, Perception verbs, Verbs of Attitude and Benefactive verbs:
(19) Emotive verbs: like, dislike, please, suit, t, be(come) hurt, be
terri ed, be anxious, doubt, suffer, surprise.
Cognition verbs: remember, be (un)aware of sth, be incompre-
hensible.
Perception verbs: appear in a vision, taste, disappear for sby’s
eyes.
Verbs of attitudes: nd dif cult or painful, nd easy, be natural
for sby, be important/dear/a pleasure, be a problem, be strange/
incomprehensible, be good for sby, accidentally speak, be lost to,
be free to, be suf cient, be of use to, make a good/bad progress,
be of sby’s business, go to sby’s head, be exaggerated.
Benefactive verbs: receive, be offered to, be suf cient, accom-
pany, be of use to.
Some of the lexical meanings listed in (18) and (19) above are derived
58
or non-literal meanings. Consider for instance hverfa vero¨ldin, which is
glossed here as ‘sleep’, and which intuitively does not fall into any of
the ve classes above. The dative human argument is hardly an Experi-
encer, possibly a Perceiver, but de nitely not a Bene ciary:
(20) ME´R hvarf VERO¨LDIN.
me (dat) disappeared the world (nom)
‘The world disappeared for my eyes.’ or ‘I slept for a while.’
The predicate hverfa vero¨ldin is probably a metaphorical extension of
other impersonal uses of hverfa such as e-ð hvarf e-m ‘disappear for
sby’s eyes’, where the dative human argument probably is a Perceiver:
(21) a. UMHVERFIÐ hvarf ME´R õ´ sma´stund.
the environment (nom) disappeared me (dat) for a while
‘Everything went black for me.’
b. SY´NIN hvarf ME´R eitt augnablik.
the vision (nom) disappeared me (dat) a moment
‘I couldn’t see anything for a moment.’
c. HANN hvarf ME´R sjo´num sma´m saman.
he (nom) disappeared me (dat) sight gradually
‘Gradually, I lost sight of him’.
To summarize so far, most of our verbs seem to be constructed with the
thematic roles of the Experiencer (Emotive verbs and Verbs of Attitude),
Perceiver (for Perception verbs), Cognizer (for Cognition verbs) and
Bene ciary (for Benefactive verbs). Further, the majority of our verbs
are psych-verbs and a small subset comprises Benefactive verbs.
It is not surprising that the dative can also be realized as a Bene ciary
dative. That is parallel to passives of ditransitive verbs, of which the
morphological dative is also a Bene ciary. Recall from section 1 above
that passives of ditransitive verbs also occur with the same case pattern
as our group of alternating verbs (see Zaenen, Maling & Thra´insson
(1985) for an illustration of the subject properties of the dative and the
object properties of the nominative of passives of ditransitive verbs in
Icelandic):
(22) a. JO
´ NI var ge n BO´KIN.
Jo´n (dat-subj) was given the book (nom-obj)
‘John was given the book.’
b. BO´KIN var ge n JO´NI.
the book (nom-subj) was given Jo´n (dat-obj)
‘The book was given to John.’
59
Another verb in (14) above, fylgja ‘accompany’, poses a problem since
its dative does not immediately suggest itself as any of our four thematic
roles above:
(23) a. VE´LINNI hafa alltaf fylgt LEIÐBEININGAR.
the machine have always followed instructions
‘The machine has always been accompanied by instructions’.
b. LEIÐBEININGARNAR hafa alltaf fylgt VE´LINNI.
the instructions have always followed the machine
‘The instructions have always accompanied the machine.’
However, examples of fylgja ‘accompany’ with the dative being more
Bene ciary-like can be found in Icelandic:
(24) a. HONUM ho¨fðu alltaf fylgt ÞRI´R VASKIR FYLGDARSVEINAR.
him had always followed three robust followers
‘He had always been accompanied by three robust followers.’
b. ÞRI´R VASKIR FYLGDARSVEINAR ho¨fðu alltaf fylgt HONUM.
‘three robust followers had always followed him
‘Three robust followers had always accompanied him.’
And not all verbs selecting for a Dative human argument and a Nomina-
tive stimulus submit to the dual assigning class:
(25) a. ME´R hefur alltaf lõ´kað GUÐMUNDUR.
me has always liked Guðmundur
‘I have always liked Guðmundur.’
b. GUÐMUNDUR hefur ME´R alltaf lõ´kað.
Guðmundur have me always liked
‘Guðmundur I have always liked.’
c. *GUÐMUNDUR hefur alltaf lõ´kað ME´R.
Guðmundur has always liked me
d. ?ME´R hefur GUÐMUNDUR alltaf lõ´kað.
me has Guðmundur always liked
Note that the Dative human argument me´r occurs between the nite and
non- nite verb in (25b), which is a subject position (criteria 2 in (3)
above), and it cannot occur in the object position following the in nite
verb in (25c) (see also footnote 6 above), nor can the Nominative stimu-
lus occur between the two verbs. Thereby we know that me´r has to be
the syntactic subject of lõ´ka at all times and that Guðmundur has to be
the syntactic object at all times. Compare this to the following examples
of lõ´ka ‘like’ from Old Icelandic:
60
(26) a.
[…] þa´ lõ´kar HON ME´R yfir allar þœr er ek hefi fyrr se´t ok
heyrt
then likes she (nom) me (dat) over all those which I have earlier
seen and heard
‘her I like best of all those I have seen or heard’
(Barlaams ok Josaphats saga 1981: 68. 12–13)
b.
[…] og lõ´kaði ÞAÐ O¨LLUM vel.
and liked it (nom) everyone (dat) well
‘[…] and everyone was at ease with that.’
(Brennu-Nja´lssaga 1987: 132)
In (26a) the Nominative stimulus behaves like a subject, since it inverts
with the verb when þa´ ‘then’ is situated in the rst position (criteria 2 in
(3) above). The same goes for það in (26b), it follows the verb in V1
clauses (so-called Narrative Inversion (see Sigurðsson 1983)). The
Dative human argument behaves like an object, in both examples, either
situated in front of the in nite verb, which was an object position in Old
Icelandic (OV word order within the VP (see Ro¨gnvaldsson 1996b)), or
following the nite verb, located after the Nominative stimulus. This
can be taken as an indication of the verb lõ´ka ‘like’ not behaving in
Modern Icelandic as in Old Icelandic, with respect to syntactic realiza-
tion of arguments (see Barðdal (1997), and the references cited there,
for a discussion on subject properties in Old Scandinavian, and Barðdal
(2000) for a methodological discussion of the problem). We might there-
fore suspect that our group of psych-verbs alternating their syntactic
structure was larger in Old Icelandic than in Modern Icelandic.9
3.2. Construction-based Approach
It is not obvious how contemporary syntactic theories would explain the
existence of psych-verbs taking either argument as the subject or the
object. I can think of two possible ways. The rst one is to say that there
are two verbs henta, with different speci cations for grammatical func-
tions. A notational variant would be that there are two senses, or uses,
of the verb henta, with subsequent differences in grammatical functions.
The strongest argument against such an analysis is also the most obvious
one. It is strange to assume two verbs henta with the same phonological
and morphological form and the same semantic meaning. It is also
strange to assume two different uses since the uses in example (1) above
are not notably different. Also, assuming two verbs henta with the same
form and the same meaning but a difference in syntactic structure seems
highly unmotivated. The only motivation for such an analysis is to
61
explain the kind of data that generate the analysis, and as such it is cir-
cular.
Another possible way to account for our data would be to argue that
one of the examples in (1) above is derived from the other by some sort
of a transformation or derivation. Such an analysis would, however,
have to invoke an invisible transformation/derivation, to my mind a
clear disadvantage of the analysis.
However, an earlier version of the present paper has inspired Christer
Platzack to try to account for the behaviour of our alternating verbs, for-
mulating it within the Minimalism program (Platzack 1999). That analy-
sis assumes different internal VP structures for verbs like lõ´ka as
opposed to verbs like henta. An LFG analysis has also been proposed by
Zaenen, Maling & Thra´insson (1985) for the alternation between Dat-
Nom and Nom-Dat passives.
The possibility of the verbs in (14) and (15) occurring with either the
Dative argument or the Nominative argument as the subject can easily
be captured within Construction Grammar, a recent theory, advocated by
Goldberg (1995), Fillmore & Kay (1999), Croft (2000, 2001) and others.
Such a theory assumes that the construction is a basic unit of language
and grammar, consisting of syntactic form and semantic (or pragmatic)
meaning. Given a de nition of the construction as a form and meaning
correspondence, the syntactic frames of our dual assigning psych-verbs
would qualify as two different constructions:
(27) a. Impersonal SubjDat V ObjNom Me´r hentar þetta
me pleases this
‘I’m pleased by this’
b. Transitive SubjNom V ObjDat þetta hentar me´r
‘This pleases me’
The Impersonal construction is different from the Transitive construction
since it has a Dative subject and a Nominative object, while the Transi-
tive construction has a Nominative subject and a Dative object. There-
fore these have to be considered to be two different constructions.
Furthermore, within Construction Grammar all linguistic knowledge is
represented as constructions and a network of constructions (Goldberg
1995, Cruse & Croft, in prep.). This is a crucial difference in the de ni-
tion of constructions between Construction Grammar and other theories,
such as traditional grammar and Generative grammar. While in tradi-
tional grammar and Generative grammar the label construction is used
to denote deviations from ordinary argument linking constructions, such
as the passive construction or the Way construction, Construction Gram-
62
mar views everything as a construction, i.e. every piece of form and
meaning correspondence constitutes a construction of its own. This is an
important point to be made for the reader’s understanding of the discus-
sion of the present paper, and in fact all discussions within Construction
Grammar. Furthermore, it is important to stress that the meaning of con-
structions can either be general (derivable from the parts of the construc-
tion) or speci c (not derivable from the constructional parts) (see Cruse
& Croft in prep.: ch. 10), but does not necessarily always have to be
speci c. In other words, within Construction Grammar it is justi able to
invoke an analysis that is dependent on the concept of construction also
for data that are not idiomatic.
Within Construction Grammar, Goldberg (1995) has advocated a the-
ory of argument linking which runs like the following: Certain semantic
relations exist between constructions and groups of verbs, resulting in
certain groups of verbs being associated with certain constructions. The
lexical entry of each verb only contains encyclopaedic information about
the meaning of that particular verb. The construction functions as the
tool for mapping information from the lexicon to the syntax, i.e. as the
syntax-semantic interface. From the semantics of the verb it follows in
which constructions it can occur. Consider for instance the Caused
Motion construction (X causes Y to move Zloc). On a Constructional
approach we expect verbs expressing motion to appear in the Caused
Motion construction. We assume, only when it is not predictable from
the semantics of the verb, that it is listed per se which verbs are asso-
ciated with which construction.
More speci cally on psych-verbs, their argument linking has been dis-
cussed by Croft in a number of papers (see at least Croft 1993 and
1998), where he suggests that argument linking in general and of psych-
verbs in particular can be derived from conceptual causal structure. The
important issue here is the relation of the participants of an event to each
other. The participant that acts upon another participant is linked to rst
position, to subject (and nominative), while the participant acted upon is
linked to the second position, to object (and accusative), in a transitive
construction of a nominative-accusative language. This is a direct conse-
quence of the unidirectional causal structure of the event.
However, certain events, prototypically involving mental experience,
do not have a unidirectional causal structure. Such events can therefore
be encoded in two ways, i.e. either with the human participant as the
subject and the stimulus as an object, or with the stimulus as a subject
and the human participant as an object. This is true both cross-linguisti-
cally and within the same language. These two conceptualizations of the
same event highlight differences in perspective, in that when the human
participant is the subject the event is conceptualized as if the human par-
63
ticipant directs his/her attention to the stimulus, while when the stimulus
is linked to subject the event is conceptualized as if the stimulus is caus-
ing an effect on the human participant. This is illustrated in the follow-
ing example:
(28) a. E´g hrœðist hunda.
I (nom) fear dogs (acc)
b. Hundar hrœða mig.
Dogs (nom) frighten me (acc)
The subject is in nominative in both examples and the object is in accu-
sative. Thereby the nominative is used to encode the initiator of the
transmitted force, while the accusative is used to encode the endpoint of
the transmitted force. Croft (1993) furthermore points out that case lan-
guages often have a special way of encoding bidirectional stative events,
either by both arguments being encoded with the same (neutral) morpho-
logical case, or by using a third case, i.e. neither the nominative nor the
accusative, to encode human participants of stative mental events. Both
of these are represented in Icelandic:
(29) a. MIG dreymdi O¨MMU.
I (acc) dreamt grandma (acc)
‘I dreamt of grandma.’
b. ME´R lõ´kar vel við Guðmund.
I (dat) like well with Guðmundur (acc)
‘I like Guðmundur.’
In (29a), both arguments are encoded with the morphological accusative,
while the human argument of (29b) is encoded with the morphological
dative, thereby morphological case is used to signal lack of the proto-
typical causal relation of one entity transmitting force onto another
entity. A clear majority of the dual assigning verbs discussed in this
paper are stative predicates and some are inchoatives. However, since
the human argument of our group of verbs is always encoded with the
morphological dative case in Icelandic, it does not come as a surprise
that both arguments can behave as subjects and both can behave as
objects, neither argument is the initiator of transmitted force.
How would we formally represent our linguistic knowledge of the
dual assigning verbs discussed in this paper in a Construction Grammar
framework? I opt for the solution that constructions are assumed to exist
at different abstract levels, like Croft (2000) and Cruse and Croft (in
prep.) assume, following Langacker (1988). On such an account we
would assume a more general construction on an abstract level, consist-
64
ing only of syntactic slots.10 At a lower level of schematicity we would
assume a verb-class speci c construction, consisting of syntactic slots
and the semantic content (corresponding to (19) above). One level below
we would expect to nd the verb-subclass speci c construction (corre-
sponding to (18) above). At yet a lower level of schematicity we nd
the verb-speci c construction, which is more concrete and lexically
lled (corresponding to (14) and (15) above). This can be represented as
follows (with eventually more intermediate levels not spelled out):
Figure 1. Different levels of schematicity for Dat-Nom verbs.
A similar gure can be drawn for the Nom-Dat construction:
Figure 2. Different levels of schematicity for Nom-Dat verbs.
The relation between the two constructions, i.e. the Nom-Dat and Dat-
Nom, or the Transitive and the Impersonal construction, can be graphi-
cally illustrated in the following way:
65
Figure 3. The relation between the Nom-Dat and Dat-Nom constructions.
The difference between the constructions in Figures 1 and 2 is clear. In
contexts where the Dative human argument is topical, the Dat-Nom con-
struction is chosen by the speaker, while in contexts where the Nomina-
tive argument is topical, the Nom-Dat construction would be used.
When uttering a sentence with the Dative argument rst the speaker is
making a proposition about the human participant while uttering a sen-
tence with the Nominative argument rst a proposition is being made
about the stimulus. It is therefore extra sentential factors that are crucial
and decide upon the grammatical functions of the arguments.11 The
relation between the Impersonal and the Transitive construction, when
our alternating verbs occur in them, is the same as the relation between
an argument linking construction and its topicalization construction.
Hence I have named the network link in Figure 3 between the two con-
structions IT, where I stands for instance and T stands for Topicalization.
This can be illustrated by the fact that when one of the arguments is in
rst position of the clause the other has to follow the in nite verb and is
not comfortable with being situated in the intermediate position between
the two verbs:
(30) a. GUÐMUNDI hefði o¨rugglega hentað FYRIRKOMULAGIÐ vel.
G (dat) had de nitely suited the arrangement (nom) ne
b. ?FYRIRKOMULAGIÐ hefði GUÐMUNDI o¨rugglega hentað vel.
the arrangement (nom) had G (dat) de nitely suited ne
c. FYRIRKOMULAGIÐ hefði o¨rugglega hentað GUÐMUNDI vel.
the arrangement (nom) had de nitely suited G (dat) ne
d. ?GUÐMUNDI hefði FYRIRKOMULAGIÐ o¨rugglega hentað vel.
G (dat) had the arrangement (nom) de nitely suited ne12
Had the arguments been comfortable in the intermediate position
between the verbs in examples (30b and d), that would have been an
indication of that argument inverting with the verb when something else
is topicalized, a clear subject criteria (recall (3) above). These data
therefore show that when one of the arguments is the topic, and subse-
quently the subject, the other argument prefers to be linked to the object
position, which means that topic and subject coincide for this group of
verbs in Icelandic. Put differently, as soon as the “object” argument of
one of the two constructions is topicalized by the speaker the other con-
66
struction is activated and not the “ordinary” topicalization construction.
Let us compare these data to the corresponding data with the verb lõ´ka
‘like’. Consider again (25) above, with the verb lõ´ka, repeated here for
the sake of convenience:
(25) a. ME´R hefur alltaf lõ´kað GUÐMUNDUR.
me has always liked Guðmundur
‘I have always liked Guðmundur.’
b. GUÐMUNDUR hefur ME´R alltaf lõ´kað.
Guðmundur have me always liked
‘Guðmundur I have always liked.’
c. *GUÐMUNDUR hefur alltaf lõ´kað ME´R.
Guðmundur has always liked me
d. ?ME´R hefur GUÐMUNDUR alltaf lõ´kað.
me has Guðmundur always liked
When the Nominative argument is topicalized to
rst position, the
Dative argument occupies the position between the verbs, hence the
Dative argument is the subject and the Nominative argument in rst
position is a topicalized object, contrary to our group of alternating
verbs.
Recall now that some of our predicates are simple verbs and some are
complex predicates. Within Construction Grammar, the simple verbs
simply occur in the lexically empty Transitive and Impersonal construc-
tions, while the complex predicates would be regarded as lexically lled
instances of the basic construction (see Goldberg 1995: 79–81).
In my view, one reason for choosing a Construction-based analysis
like the one presented in this paper is that it is simple, in epistemological
terms. Construction Grammar was rst developed to account for idioms.
It turned out that the machinery needed for that, i.e. the construction,
also took care of less lexicalized expressions and even simple sentences,
without adding any extra machinery to the theory (Goldberg 1995: 6–7;
Cruse & Croft in prep.: ch. 10). That machinery, i.e. the construction,
can now be used to account for the double mapping of certain Dat-Nom
verbs in Icelandic (and other languages). Thereby, the major tools of
Construction Grammar, that is the construction, can be added to account
for phenomena the theory was not designed to account for in the rst
place. In that way, Construction Grammar is a good scienti c theory.
Further, in Lakatos’ terminology (Couvalis 1997: 70), in which a dis-
tinction is made between the core claim and the auxiliary assumptions
of a research program, Construction Grammar accounts for the double
mapping of the verbs in question with its core claim and does not need
to make use of auxiliary mechanisms.
67
4. SUMMARY
In this paper I have shown that a certain subgroup of Dat-Nom verbs in
Icelandic, with two arguments, a Dative human argument and a Nomina-
tive stimulus, shows a variation regarding the realization of syntactic
functions. This means that both arguments of these verbs can be realized
as the syntactic subject and both arguments can be realized as the syn-
tactic object.
When examining the semantics of this group of verbs it turns out that
they can roughly be divided into Emotive verbs, Perception verbs, Cog-
nition verbs, Verbs of attitude and Benefactive verbs. The dual assign-
ment property of those verbs is a result of their conceptual causal
structure. Since these verbs denote stative/inchoative events which have
a bidirectional causal structure it is possible to view them either as if the
human argument is directing his/her attention to the stimulus, or as if
the stimulus is causing an effect in the human argument. Case languages
often signal the lack of the prototypical causal relation, i.e. of one entity
transmitting force onto another entity, with a neutral case, i.e. neither
with nominative nor accusative case. This case is often the dative case.
This is also true of Icelandic.
A construction Grammar analysis has been proposed where the Dat-
Nom and the Nom-Dat categorization frames are viewed as separate but
related constructions, existing at different levels of schematicity. The
relation between the two constructions seems to be the same or similar
to the relation between an ordinary argument structure construction and
its topicalization construction, since for these verbs the topicalization
construction is not as readily available as the other construction is. For
this group of verbs subject and topic seem to coincide.
ACKNOWLEDGEMENTS
This paper is a revised and expanded version of a working paper in WPSS (see Barðdal
1999). I thank everyone who contributed to the WPSS version, and for a discussion of and
comments on this version I thank Bill Croft, Tho´rhallur Eytho´rsson, Jo´hannes G. Jo´nsson,
the editor, an anonymous reviewer of this journal, and the audience at the Linguistics
Seminar in Manchester, 12 December 2000, where this work was presented. This research
is supported in part by a grant from the Icelandic Research Fund for Graduate Students.
NOTES
1 Note, though, that ordinary fear- and frighten-verbs in Icelandic select nominative and
accusative case, while the verbs under discussion select for nominative and dative case.
2 When translating these examples into English I have consistently used the English
active versus passive forms. That is only to maintain the word order between the
arguments and should not be taken to imply any diathetic differences.
3 In the examples above I have only used pronouns, but full NP’s do not differ from
pronouns with respect to the subject properties in Icelandic.
68
4 The fact that such double mapping verbs exist has some interesting theoretical
consequences. It implies that the relation between the subject and the predicate is not
necessarily as “inherent” as it is often assumed to be. It implies that the subject is in
some sense a “derived” category. Surely it doesn’t mean that subjects don’t behave
differently from objects, they do, of course, but these data suggest that our assumptions
on why they behave differently needs to be reconsidered.
5 Modern High German has a verb gefallen ‘please, suit’, which behaves very much like
Icelandic henta. The Dative human argument of gefallen, however, is not considered a
subject in German. That is partly due to different criteria being used as subject criteria
in German and Icelandic. In German the subject is considered to be the nominative
argument, thereby position is excluded as a subject criterion. It is a fact, though, that
impersonal verbs in German pass some of the tests traditionally associated with
subjecthood (see Seefranz-Montag (1983) and a discussion thereof in Barðdal (1997)
and Barðdal in prep.). The literature on subjecthood in German usually does not
mention this fact.
6 To illustrate that those two tests are reliable subject tests, consider the following
Icelandic examples:
i. E´g hef keypt hana/bo´kina.
I have bought it/the book
ii. Bo´kina/Hana hef e´g keypt.
the book/it have I bought
iii. *Bo´kina/*Hana hef keypt e´g.
the book/it have bought I
iv. ?E´g hef bo´kina/hana keypt.
I have the book/it bought
6 These examples show that position is a reliable subject test in Icelandic.
7 Speakers vary in their judgements on sa´rna. Berno´dusson (1982: 38) gives sa´rna as an
example of such verbs with varying syntactic structure, but in my language it can only
have the Dative human argument as a subject.
8 However, Jo´nsson (1997–98: 35) has noticed that the copula construction together with
an adjective might still exhibit a mild degree of productivity in Modern Icelandic, and I
agree with him.
9 A note of warning though: Example (26a) is taken from Barlaams ok Josaphats saga
(1981), a saga belonging to the genre “Late Medieval Romances”. Firstly, the saga is
Old Norwegian and not Old Icelandic, and secondly, and more importantly, Barlaams
ok Josaphats saga is a translation from Latin. It is certainly a free translation, but
nevertheless it cannot be excluded that the construction is due to Latin in uence. The
order of the constituents in (26b) can be due to “quanti er oating”. Note that the
Dative human argument is a quanti er. These examples are therefore not conclusive
examples of the subjecthood of the Nominative stimulus. Furthermore, example (26b) is
not altogether bad in Modern Icelandic.
10 Psycholinguistic research (Barðdal, in press) has con rmed the existence of Nom-Acc
construction at this highest level, but corresponding evidence did not emerge for the
Dat-Nom construction. It is therefore not clear whether this highest abstract level of the
Dat-Nom and the Nom-Dat constructions is psychologically real in the mind of
Icelandic speakers.
11 About subject choice, Croft says (1991: 151):
(i) Most discourse analysts agree that, when a choice for subject is involved, topical-
ity governs the choice, and that, when a choice is not involved, the NP that is
grammatically required to ll the subject slot is a “natural topic”.
12 There seems to be some variation between speakers as to the (non-)grammaticality of
these and similar examples. This needs to be further investigated.
69
REFERENCES
Allen, C. L. 1995. Case Marking and Reanalysis. Grammatical Relations from Old to Early
Modern English. Oxford: Clarendon Press.
Barðdal, J. 1997. Oblique Subjects in Old Scandinavian. Working Papers in Scandinavian
Syntax 60: 25–50 [also in NOWELE 37: 25–51, 2000].
Barðdal, J. 1998. Argument Structure, Syntactic Structure and Morphological Case of the
Impersonal Construction in the History of Scandinavian. Scripta Islandica 49: 21–33.
Barðdal, J. 1999. The Dual Nature of Icelandic Psych-Verbs. Working Papers in
Scandinavian Syntax 64: 78–101.
Barðdal, J. 2000. The Subject is Nominative. On Obsolete Axioms and their Deep-
rootedness. In Lindberg, C.-E. & Nordahl Lund, S. (eds), 17th Scandinavian Conference
of Linguistics. Odense: Institute of Language and Communication, pp. 93–117.
Barðdal, J. (in press) Case Assignment of Nonce Verbs in Icelandic. SKY Journal of
Linguistics 13.
Barðdal, J. (in prep.) The Dative Experiencer First Construction in Icelandic and German.
Ms. Lund University/University of Manchester.
Barlaams ok Josaphats saga. 1981. Magnus Rindal (ed.) for the Kjeldeskriftfondet, Oslo.
Barnes, M. 1986. Subject, Nominative and Oblique Case in Faroese. Scripta Islandica 37:
13–46.
Berno´dusson, H. 1982. O´perso´nulegar setningar. [Impersonal Clauses] Master’s thesis.
University of Iceland, Reykjavõ´k.
Brennu-Nja´lssaga. 1987. In Halldo´rsson, B., Torfason, J., To´masson, S. & Tho´rsson, O¨.
(eds), I´slendingaso¨gur og þœttir . Fyrsta bindi. Reykjavõ´k: Svart a´ hvõ´tu.
Couvalis, G. 1997. The Philosophy of Science. Science and Objectivity . London: Sage
Publications.
Croft, W. 1991. Syntactic Categories and Grammatical Relations. The Cognitive
Organization of Information. Chicago and London: University of Chicago Press.
Croft, W. 1993. Case Marking and the Semantics of Mental Verbs. In Pustejovsky, J. (ed.),
Semantics and the Lexicon. Dordrecht: Kluwer Academic Publishers, pp. 55–72.
Croft, W. 1998. Event Structure in Argument Linking. In Butt, M. & Geuder, W. (eds), The
Projection of Arguments: Lexical and Compositional Factor. Stanford: CSLI, pp. 21–63.
Croft, W. 2000. Lexical Rules vs. Constructions: A False Dichotomy. In Cuyckens, H.,
Berg, Th., Dirven, R. & Panther, Kl.-U. (eds), Motivation in Language: Studies in
Honour of Gu¨nter Radden. Amsterdam: John Benjamins.
Croft, W. 2001. Radical Construction Grammar: Syntactic Theory in Typological
Perspective . Oxford: Oxford University Press.
Cruse, A. & Croft, W. (In prep.) Cognitive Linguistics [to appear in the Cambridge
Textbooks in Linguistics series].
Goldberg, A. E. 1995. Constructions. A Construction Grammar Approach to Argument
Structure. Chicago and London: University of Chicago Press.
Jo´nsson, J. G. 1997–98. Sagnir með aukafallsfrumlagi. [Verbs selecting for Oblique
Subjects]. I´slenskt ma´l og almenn ma´lfrœði 19–20: 11–43.
Jo´nsson, J. G. 1998. A List of Predicates That Take a Quirky Subject in Icelandic. Ms.
Reykjavõ´k: University of Iceland.
Kay, P. & Fillmore, Ch. 1999. Grammatical Constructions and Linguistic Generalizations:
the What’s X Doing Y? Construction. Language 75: 1–34.
Langacker, R. W. 1988. A Usage-Based Model. In Rudzka-Ostyn, B. (ed.), Topics in
Cognitive Linguistics. Amsterdam/Philadelphia: John Benjamins Publishing Company,
pp. 127–161.
Maling, J. & Jo´nsson, J. G. 1995. On Nominative Objects in Icelandic and the Feature
[‡Human]. Working Papers in Scandinavian Syntax 56: 71–79.
Platzack, Ch. 1999. The Subject of Icelandic Psych-Verbs: a Minimalistic Account.
Working Papers in Scandinavian Syntax 64: 103–115.
Ro¨gnvaldsson, E. 1996a. Frumlag og fall að fornu. [Subject and Case in Icelandic Middle
Ages]. I´slenskt ma´l og almenn ma´lfrœði 18: 37–69.
Ro¨gnvaldsson, E. 1996b. Word Order Variation in the VP in Old Icelandic. Working
Papers in Scandinavian Syntax 58: 55–86.
70
Seefranz-Montag, A. v. 1983. Syntaktische Funktionen und Wortstellungsvera¨nderung. Die
Entwicklung ‘subjektloser’ Konstruktionen in einigen Sprachen. Studien zur Theore-
tischen Linguistik 3. Mu¨nchen: Wilhelm Fink Verlag.
Sigurðsson, H. A´. 1983. Um fra´sagnarumro¨ðun õ´ fornõ´slensku [On Narrative Inversion in
Old Icelandic]. Master’s thesis. University of Iceland.
Sigurðsson, H. A´. 1989. Verbal Syntax and Case in Icelandic. Doctoral dissertation,
University of Lund.
Sigurðsson, H. A´. 1990–91. Beygingarsamræmi [Agreement]. I´slenskt ma´l og almenn
ma´lfrœði 12–13: 31–77.
Sigurðsson, H. A´. 1992. The Case of Quirky Subjects. Working Papers in Scandinavian
Syntax 49: 1–26.
Thra´insson, H. 1979. On Complementation in Icelandic. New York: Garland Publishing
Zaenen, A., Maling, J. & Thra´insson, H. 1985. Case and Grammatical Functions: The
Icelandic Passive. Natural Language & Linguistic Theory 3: 441–483.
