Professional Documents
Culture Documents
003 - Cennamo (2003)
003 - Cennamo (2003)
issues*
Michela Cennamo
1. Introduction
by Rappaport Hovav and Levin (1998), Levin (1999). And section 5 sum-
marizes the conclusions.
Not all the semantic parameters pointed out in Hopper and Thompson
(1980) are equally relevant in determining the morphosyntactic encoding of
clauses cross-linguistically (Tsunoda 1985). As pointed out by Givón
(1984), (1990), Tsunoda (1985), (1994), Lazard (1998), (2002), among
others, the main semantic parameters affecting syntactic transitivity involve
a) the presence of two highly individuated participants, A and O (i.e., of
two distinct participants ranking high on the Animacy, Definiteness and
Referentiality hierarchies), b) the Agentivity and Control of the A partici-
pant over the situation described by the verb, c) completeness/perfectivity
and realis modality/affirmative polarity of the situation expressed by the
verb, d) high degree of Affectedness of the O participant, that undergoes a
change of state.2
The morphosyntactic correlates of the semantic parameters illustrated in
tables 1-2 involve a) the linguistic expression of the core participants of a
clause, b) their coding (case-marking/adpositions), c) the occurrence of re-
lated patterns (passives, antipassives, reflexives, reciprocals), d) the pres-
ence of cross-reference markers, e) verb morphology, f) word order
(Tsunoda 1994: 4671). Also among the morphosyntactic reflexes of seman-
tic transitivity some features are more relevant and/or more frequently at-
tested than others. For instance, whereas the number of participants and
their case-frames are relevant to the morphosyntactic encoding of clauses in
all languages, and constitute the nub of the traditional definition of Transi-
tivity, the occurrence of special verb morphology and cross-reference
markers are not relevant for all languages, and word order plays a marginal
role, being mainly relevant in languages with so-called ‘inverse’ voice,
such as the Algonquian languages (Givón 1994, Klaiman 1991, Tsunoda
1994: 4672). Also syntactic reflexes such as passivization, antipassivization
and reflexivization are not found in all languages. When they do occur,
however, they generally apply to highly transitive verbs/clauses and anti-
52 Michela Cennamo
The active (3a) is also preferred to the passive (3b) when both A and O are
high in Individuation (e.g., human, and proper names) (Comrie 1989: 81–
82):
Activity verbs, on the other hand, take a different passive form, the reflex-
ive passive. (For instance, the activity verb stroit’ ‘build’ occurs with the
reflexive passive (dača stroit-sja (rabočimi) –dacha-NOM build.3SG-RFL
worker- (INSTR-PL)- ‘The dacha is (being) built (by the workers’),
whereas the active accomplishment verb postroit’ ‘build’, takes the auxil-
iary byt’ ‘be’+perfective participle (dača by-l-a postro-en-a rabočimi –
dacha-NOM be-PAST-F.SG build-PPP-F.SG worker-INSTR-PL – ‘The
dacha was built by the workers’) (Van Valin and La Polla 1997: 657, note
4; see also Keenan 1985: 254).
As pointed out by Van Valin and La Polla (1997: 657, note 4), there
seems to be a continuum of acceptability with the passive of activity verbs.
A sentence such as (4a), with a generic subject, a specific passive agent and
a durational adverbial phrase, for instance, is ungrammatical for some Eng-
lish speakers, whereas it is possible when no time interval is specified (4b),
though only with a generic agent, that indeed seems to be the relevant pa-
rameter (4c) (example from Peter Austin, p.c.) (see also discussion in sec-
tion 3):
shall see in section 3.4 for the marking of O in dependent clauses (see also
discussion in Silverstein 1976, 1993, Austin 1992, 1998, int. al.).
One can grade clauses, and locate them along a continuum, usually con-
ceived of as an axis with two poles, illustrated in table 3, that refers to the
type of syntactic construction encoding events and their participants:
In other languages verbs of impact such as hit occur in so-called object in-
corporation constructions (9), which signal that O is non-referential (Co-
oreman 1994, Foley and Van Valin 1985: 341, Van Valin and La Polla
1997: 123):
At the other pole of the transitivity continuum there occur canonical in-
transitive patterns, i.e., patterns with only one participant, S. Whereas it is
relatively straightforward to characterize prototypical transitive clauses, it
is not easy to give a unitary definition of prototypical intransitive patterns.
Even if we confine ourselves to only one of the functional domains of In-
transitivity, namely that of eventive predicates (Stassen 1997), it is difficult
to single out syntactic, semantic and pragmatic properties which character-
ize prototypical intransitive clauses. They comprise, in fact, different pat-
terns, such as meteorological verbs (Italian piove –rain.3SG – ‘it rains’),
that in some languages require the presence of a dummy subject/pivot
holder (English it rains, German es regnet ‘it rains’), or a reflexive marker
on the verb (Russian sverkaetsja –daybreak.3SG-RFL- ‘it daybreaks’, Ital-
ian si fa buio –RFL make.3SG dark- ‘it is getting dark’), as well as differ-
ent types of impersonal constructions The latter may comprise either only
one participant, S, occurring in the accusative and/or dative case (German
es freut mich –it please.3SG I.ACC- ‘it pleases me’, Icelandic mig/mer lan-
gar að fara –I. ACC/DAT wish.3SG.IMP to go- ‘I wish to go’, mig kelur –
I.ACC freeze.3SG.IMP- ‘I am freezing’, mér kónar- I.DAT get.3SG.IMP
cold- ‘I am getting cold’) (Smith 1996: 22), sometimes with the presence
of a dummy subject, as in German, or they may involve the presence of two
participants, A and O, occurring in different case-frames (Icelandic–mér
líkar þu –I.DAT like.3SG.IMP you.NOM- ‘I like you’ (example from J.
Barðdal) (see Smith 1996: 28–29, Barðdal forthc.), Russian, snegom pok-
rylo dvor -snow.-INSTR cover-3SG.NEUT ground-.ACC - ‘the snow cov-
ers the ground’), the verb reverting to the unmarked third person singular
(if finite), or occurring in the unmarked (‘neutral’) form, according to the
language, if non-finite (most typically expressed as a past/perfective parti-
ciple) (North Russian dialects, u otca svoix detej zabyto -at father.GEN. his
child.PL-ACC forget-PPP-NEUT ‘the father has forgotten his children’)
(Fici Giusti 1994: 69).
Intransitives also comprise various types of impersonal-passive pat-
terns, i.e., patterns with a passive morphology and different types and de-
grees of defocusing of A, O, S, which are either unexpressed (Italian fu
discusso a lungo –be.3SG.PAST discuss-PP.M.SG at long- ‘one debated
for a long time’; Sorbian bu spane-be.3SG.PAST sleep-PP -‘sleeping took
place’ (Fici Giusti 1994: 130)) or surface as a dummy subject (German es
wurde getanzt ’it was danced’). The defocused argument may also be ex-
pressed by the reflexive morpheme, as in Italian (cf. si è pagato molto-RFL
be.3SG pay-PP.M.SG much - ‘one has paid a lot’, si è pagati molto -RFL
58 Michela Cennamo
be.3SG pay.PP.M.PL much- ‘one is paid a lot’), where the difference be-
tween the A-SA/O-SO orientation of the pattern (according to whether the
verb is either divalent or monovalent) is signalled, respectively, by the lack
vs. presence of past participle agreement with the underlying argument sig-
nalled by the reflexive morpheme (e.g., (si è) pagato (A-orientation: [-
AGR]) vs. (si è) pagati (O-orientation: [+AGR]); si è camminato –RFL
be.3SG walk.PP.M.SG- ‘one has walked’ (SA orientation: [-AGR] ) vs. si è
partiti – RFL be.3SG leave.PP.M.PL- ‘one has left’ (SO orientation:
[+AGR]) (Cennamo 1984, 1993: 40, 1995, 1997).
The term includes split-S and fluid-S marking systems as well. One-
argument intransitive verbs, in fact, in many languages do not constitute a
homogeneous class, but subdivide into two major subclasses, labelled
SA/Unergative, SO/Unaccusative verbs, depending on the theoretical per-
spective adopted, and in particular on whether the phenomenon is regarded
as mainly semantic, determined by the interplay of semantic parameters
such as inherent lexical aspect, the Agent/Patient-like nature of the S argu-
ment, as well as its degree of Control and affectedness (Van Valin 1990,
Klaiman 1991: 124–132, Mithun 1991, inter al.), syntactic, reflecting the
syntactic nature of S in a level of representation of the clause (D-
subject/initial 1 with unergative/SA verbs, D-object/initial 2 with unaccusa-
tive/SO verbs (Hale and Kayser 1987 for English, Perlmutter 1989) or syn-
tactico-semantic, i.e., ‘syntactically represented but semantically deter-
mined’ (Perlmutter 1978, Levin and Rappaport 1989: 316, Levin and Rap-
paport Hovav 1995, among others).
The distinction between S arguments behaving morphologically and/or
syntactically, like the O argument of canonical transitive verbs, and S ar-
guments with morphological and/or syntactic properties of A arguments of
canonical transitive verbs may show up in different areas of the grammar,
both synchronically and diachronically: auxiliary selection, as in Italian —
where it also correlates with the presence vs. lack of past participle agree-
ment with S, so that SA/unergative verbs select the auxiliary avere ‘have’
and lack past participle agreement (Giovanna ha camminato molto –Jane
have.3SG walk.PP.M.SG much- ‘Jane has walked a lot’) whereas
SO/unaccusative verbs select the auxiliary essere ‘be’ and show past parti-
ciple agreement (Giovanna è partita –Jane have.3SG leave.PP.F.SG - ‘Jane
has left’) – and other European languages (Sorace 2000, inter al.), passivi-
zation, as in English (Perlmutter 1978), applicatives, as in some Australian
Aboriginal languages (Austin 2002a) as well as word order, case-marking,
crossreferencing, causatives (Dixon 1994: 70–83 and references therein).
(In)transitivity and object marking 59
Recent work on event structure and argument linking, carried out within
different frameworks, has pointed out that the varying encoding(s) of O
60 Michela Cennamo
The event structure templates accounting for the various event types com-
prise various subtypes of simple event structure templates and one complex
event structure template (14) (Rappaport Hovav and Levin 1998: 108;
Levin 1999: 229–230):
Verbs such as kill, break have a complex event structure template (14e).
They consist of two subevents, the causing event (an activity) and the re-
sulting change of state. Verbs such as sweep, eat have a basic simple event
structure (14b) in both their occurrences (with and without an object) (12a-
b) (13a-b). Of the two arguments associated with them, only the subject
realizes a structure participant; the object realizes instead a pure constant
participant (which is underlined). Their event structure template can be
augmented, leading to the complex event structure template illustrated in
(14e), as in their accomplishment uses (12c), (13c). Verbs of consumption
such as eat, in fact, behave like activities with an indefinite object (13a),
and like accomplishments with a definite one (13c).6
The different morphosyntactic behaviour of verbs such as kill, break and
sweep, hit in English, and the cross-linguistic variability in the encoding of
Os, therefore, is regarded as stemming from the different licensing of the O
argument. Structure participants are marked uniformly in English and
across languages in that their source is in the event structure template, and
therefore they fall within the universal linking rules. The x and y partici-
pants in causative complex event structure templates such as the one illus-
trated in (14e) above, are therefore realized by arguments encoded as sub-
ject and object (in languages where these relations obtain) (taking the
NOM-ACC or ERG-ABS case-frame, depending on the alignment and
case-marking system of a language or of patterns of a language). Pure
(In)transitivity and object marking 63
passive construction with the original O retaining its dative case is allowed
instead (15c):
These verbs are regarded as basically activity verbs entering the activ-
ity/accomplishment alternation, a phenomenon which is very widespread
cross-linguistically. Unlike causative accomplishments (kill, (trans.) break),
causative achievements (trans. crack) and causative activities (trans.
bounce), active accomplishments do not denote a relationship of causation
between two events, but the sequence of two events (see also note 6).
Unlike the other causative classes, in fact, this type of alternation is never
marked with a causative suffix cross-linguistically (Van Valin and La Polla
1997: 101ff).
The O participant of multiple argument activity verbs qualifies the ver-
bal action rather than identifying a particular referent in the Universe of
discourse. In English and in other languages it is an inherent argument, in
that it can be freely omitted. In languages with noun incorporation and an-
tipassives, it is incorporated into the verb ((17b), (17d)), and/or deleted,
with the verb occurring in the antipassive voice if it is non-referential. If it
is referential instead, with some verbs it may be marked as an oblique core
(In)transitivity and object marking 65
argument (17f) (Van Valin and La Polla 1997: 123–124; 150 and discus-
sion in section 3):
Croft (1998), starting from a model of argument linking based on the no-
tions of force dynamic relations among participants (Talmy 1985b) and of
verbal profile (i.e., the segment(s) of the causal (force-dynamic) chain de-
noted (=profiled) by the verb), hypothesizes that argument linking is con-
sistent cross-linguistically if “there is a straightforward force-dynamic rela-
tionship between two participants in an event and events are clearly indi-
viduated” (Croft 1998: 25). Cross-linguistic and intralinguistic variation in
the coding of O reflecting semantic parameters such as Control of the A
participant or degree of affectedness of O is regarded as reflecting the de-
gree of “self-containedness of the event” (Croft 1998: 47). If the event is
clearly self-contained (i.e., if the whole event is profiled by the verb), the
participants are encoded as canonical core arguments. If the event is not
clearly self-contained (i.e., if only some segment(s) of the event is (are)
profiled), O is coded as an oblique. In this case, in fact, O is not the “end-
point of the verbal profile” (Croft 1998: 46). This would account for the di-
rect-oblique case alternation in English (e.g. Engl. John hit the wall/hit at
the ball), and in several other languages, where it may be expressed not
only by an oblique case-marking on O, but by the antipassive morphology
as well (see discussion in section 3). Four universal linking rules are pos-
ited (18), based on the antecedence relationship between participants (X >
Y) in the force-dynamic chain (see Croft 1998: 24, 2000: 53):
(18) 1. The verbal profile is delimited by Subject and Object (if any)
2. Subject > Object
3. Antecedent Oblique > Object > Subsequent Oblique9
4. Subject > Incorporated Noun > Object (if any)
Notation of extended causation: arrows (↑↑) at the beginning/end of the time inter-
val
The zig-zag line in figure 3 represents the profiled phase of the verb eat,
which is an undirected activity (Croft 2000: 9; 61). Achievements, accom-
plishments and activities differ in the phases profiled. With achievements
(e.g. break) the result state is not profiled (cf. figure 1); with accomplish-
ments (e.g. eat in (20)) all three phases of the situation are profiled (its in-
ception, the directed process, and its completion, i.e., its result state, repre-
sented by solid lines) (figure 2); activities involve an “unprofiled incep-
tion” (represented as a dotted line (…) and an “unspecified extension on
(In)transitivity and object marking 69
3.1. Individuation of O
In Icelandic with some verbs animate ([± Human]) Os occur in the dative
(24a), unlike inanimate Os, which take the accusative case (24b) (Barðdal
1993; 2001a: 148–149). Although dative marking on O is not usually re-
ferred to in the literature on Icelandic as a case of ‘differential object mark-
ing’ (see Svenonius 2001, Maling 2002: 27–29 for recent discussion), the
similarity with other languages which present this phenomenon is striking
(this interpretation of the occurrence of the dative in (24a) is apparently
supported by the sensitivity of case-marking to the Animacy hierarchy in
other areas of the grammar (J. Barðdal, p.c.):
In Hindi the –ko suffix differentiates definite Os ([± Animate] from in-
definite ones (Magier 1987: 192–193):
are activity verbs (Foley and Van Valin 1985: 342). In languages with
cross-reference marking of S, A and O on the verb, in the antipassive voice
there only occurs one cross-reference marker, that of the A participant:
ive’, A occurs in the absolutive case and the (optionally expressed) non-
referential O is in the dative (33b):
(38) a. z? tā win-m
I-DIR you.OBL see-1SG
78 Michela Cennamo
‘I see you’
b. z? da saray winm
I-DIR DEM man.DIR see-1SG
‘I see a man’
In Dyirbal first and second person pronouns (both A and O) are inflected on
a nominative-accusative basis, whereas third person participants pattern on
an ergative-absolutive basis (Comrie 1989: 189). In some Amerindian lan-
guages (e.g. Algonquian and Tanoan languages), if O is higher in Indi-
viduation than A, (i.e., if A=3rd person/noun and O=1st, 2nd person) a dif-
ferent verbal voice must be used, the inverse (39) (see Klaiman 1991: 161–
226, for discussion of the controversial inverse/passive status of these
structures in some Tanoan languages, e.g., Picurís):
(39) a. ni- sēkih -ā -nān atim (Plains Cree; Klaiman 1991: 162)
1 scare DIR 1PL dog
‘We scare the dog’
b. ni- sēkih -iko -nān atim
1 scare INV 1PL dog
‘The dog scares us’
With activity and mental process verbs in many Australian languages there
obtains the ABS-DAT case-frame (Blake 2001: 145–146) (see (41a), from
Bayangu (Austin 1982: 41 and references therein) and (41b), from Warlpiri
(Simpson 1991: 96; 317–20)):
The same type of marking obtains in Icelandic, where the dative case on O
appears to correlate with the affectedness/lack of Control of A (43), the pat-
tern being analogous to the Chukchi examples discussed in section 1 (note
2, ex. (vi)), where the degree of control of A over the verbal activity is re-
flected in the Instrumental marking of O (though also involving a different
voice, the antipassive):
In Icelandic, with these verb classes (i.e., with non-agentive verbs) O may
occur in the nominative, and the verb may revert to the impersonal form
(45b) (Andrews 1990: 210):
Also the internal (‘cognate’) object in some languages may occur in non-
canonical transitive case-frames. In Diyari (and in many other Australian
Aboriginal languages), for instance, cognate objects occur in the absolutive
case, in the case-frame ABS-ABS/NOM-ABS (Austin 1982: 39–40, Lazard
1994: 153):
ACC-kill-STEM-3SG.NOM.PRES
‘The boy will kill the dog’
Whereas in (50a), (50c) and (50e) the situation described by the verbs is
portrayed as attained, in (50b), (50d), so-called conative constructions, the
presence of a prepositional object in English, and of the adverbial case in
84 Michela Cennamo
The interplay of the aspectual nature of the predicate and the degree of af-
fectedness of O may be regarded as the basis for the dative coding of O
with some resultative verbs entering the anticausative alternation (e.g.,
fœkka ‘decrease’, fjölga ‘increase’, sökkva ‘sink’) (52a-b). The verbs fœkka
and sökkva, in fact, are not prototypical transitive verbs. They are ‘gradual
completion verbs’, i.e., they imply the gradual approach to a telos that may
be unspecified and not necessarily reached (Bertinetto and Squartini 1995:
12–13) and therefore they are less telic than core resultatives such as drepa
‘kill’, which take an accusative O. The dative therefore might reflect the
low degree of affectedness of O resulting from the aspectual nature of
these verbs:
The habitual nuance associated with the antipassive is also apparent in (55),
from Tzutujil (a Mayan language) (Dayley 1985: 346, in Cooreman 1994:
58), where the O participant is deleted and A occurs in the absolutive case:
clause when the A of the governing verb is not overtly expressed, as in im-
peratives (56b) or impersonal constructions (56c) (the Finnish passive is in
fact an impersonal verb form) (Sands and Campbell 2001: 280–284)11:
Ngurtirti-ACC
‘I got another man Ngurtirti’
b. parampu-wu ngunh a kumpa-ja purra-rnu
acacia-DAT that sit-PAST grind-IMPFVSS
marrkara nganaju parampu- wu pirruwa-wu
younger brother I-DAT acacia-DAT husk-DAT
‘My younger brother sat grinding acacia husks’
5. Conclusions
Notes
In some languages the degree of Control of A over the verbal process affects
the coding of O and relates to a different voice. In Chukchi (Polinskaja/Ned-
jalkov 1987: 263, in Lazard 1994: 186), if A has no Control over the verbal
process, the verb occurs in the antipassive voice and O is in the instrumental
case:
vi. ’ətt’-n ine-nlp’et-g’i gutilg-e
dog-ABS ANTI-break-3SG leash -INSTR
‘The dog breaks the leash (and runs away)’
vii. ’ tt’- rlp’en-nin gutilg-n
dog-ERG break-3SG/3SG leash-ABS
‘The dog breaks the leash’
In (vii) speakers regard the ergative pattern as odd, in that it implies a higher
Control of A over the verbal process (Lazard 1994: 186).
3. The term ‘oblique’ is used with a variety of meanings in the literature (see
Blake 2001 for an overview). In this paper it is used as a cover term, referring
to the non-canonical encoding of core arguments (e.g., da-
tive/instrumental/adpositional O) as well as to non-core arguments (which
may be marked either by a case-marker or an adposition). In some languages
the same marker used for a non-core argument (e.g., the dative/instrumental
case) may also occur for core arguments (e.g., non-canonical Os) (Blake
2001: ch.5 and references therein).
4. For some English speakers, however, kill, unlike break, can occur without an
object, to refer to a habitual activity (M. Parry, p.c.):
(i) John killed/has killed, this is why he is in prison
(ii) Stalin’s police killed indiscriminately/for many years
This suggests that more fine-grained distinctions are to be made within the
class of core transitives.
5. Not all verbs of surface contact through motion, however, allow unspecified
object deletion. Wipe and rub, for instance, can be used intransitively only if
their object is contextually recoverable ((i)-(ii)) (Rappaport Hovav and Levin
1998: 115):
(i) John washed and Mark wiped (sc. the dishes)
(ii) John applied the polish onto the brass handles and Mark rubbed
6. Although accomplishments such as kill, (trans.) break, and the accomplish-
ment uses of sweep, eat, as in (12c) Jane swept the floor clean and in (13c)
Jane ate two buns are represented as having the same event structure template
(14e), they instantiate two different subclasses of accomplishments. Unlike
with kill, break, the two subevents unfold together, i.e., they are temporally
(In)transitivity and object marking 95
11. For a discussion of an exception to this rule, the so-called ‘Missing person
construction’, see Sands and Campbell 2001: 302, note 22).
12. Although Levin’s theory only deals with O as a grammatical relation, and not
with the morphosyntactic encoding of O (B. Levin, p.c.), her approach offers
an interesting starting point for the study of the interface of the event structure
status of participants with their morphosyntactic realization.
References
Abbreviations
Comrie, Bernard
1978 Ergativity. In: Winfred Lehmann (ed.), Syntactic Typology: Studies
in the Phenomenology of Language, 329–394. Austin: University of
Texas Press.
(In)transitivity and object marking 99
Comrie, Bernard
1979 Degrees of ergativity: some Chukchee evidence. In: Frans Plank
(ed.), 219–240.
Comrie, Bernard
1989 Language Universals and Linguistic Typology. London: Blackwell.
Cooreman, Ann
1994 A functional typology of antipassives. In: Barbara Fox and Paul J.
Hopper (eds.), Voice in Form and Function, 49–87. Amster-
dam/Philadelphia: Benjamins.
Croft, William
1991 Syntactic Categories and grammatical Relations: The Cognitive Or-
ganization of Information. Chicago: University of Chicago Press.
Croft, William
1998 Event structure and argument linking. In: Miriam Butt and Wilhelm
Geuder (eds.), 21–63.
Croft, William
2000 Verbs and Argument Structure. MS: University of Manchester.
Dixon, Robert M.W.
1979 Ergativity. Language 55: 59–138.
Dixon, Robert M.W.
1994 Ergativity. Cambridge: Cambridge University Press.
Feuillet, Jack (ed.)
1998 Actance et Valence dans les Langues de l’Europe. Berlin: Mouton
de Gruyter.
Fici Giusti, Francesca
1994 Il Passivo nelle Lingue Slave. Roma: Franco Angeli.
Fici Giusti, Francesca
1998 Diathèse et voix marquée. In: Jack Feuillet (ed.), 347–389.
Foley, William A. and Robert D. Van Valin Jr.
1985 Information packaging in the clause. In: Timothy Shopen (ed.), vol.
1, 282–364.
Givón, Talmy
1984 Syntax: a Functional-Typological Introduction, vol. 1. Amster-
dam/Philadelphia: Benjamins.
Givón, Talmy
1990 Syntax: a Functional-Typological Introduction, vol. 2. Amster-
dam/Philadelphia: Benjamins.
Lazard, Gilbert
1998b Définition des actants dans les langues européennes. In: Jack
Feuillet (ed.), 11–146.
Lazard, Gilbert
(In)transitivity and object marking 101
Shibatani, M. Masayoshi
2001 Non-canonical constructions in Japanese. In: Alexandra Y. Aikhen-
vald et al. (eds.), 307–354.
Shopen, Timothy (ed.)
1985 Language Typology and Syntactic Description, vol. 1, Cambridge:
Cambridge University Press.
Silverstein, Michael
1976 Hierarchy of features and ergativity. In: Robert M.W. Dixon (ed.),
Grammatical Categories in Australian Languages, 112–171. Can-
berra: Australian Institute of Aboriginal Studies.
Silverstein, Michael
1981 Case marking and the nature of language. Australian Journal of Lin-
guistics 1: 227–246.
Silverstein, Michael
1993 Of nominatives and datives: universal grammar from the bottom up.
In: Robert D. Van Valin (ed.), 465–498.
Simpson, Jane
1991 Warlpiri Morphosyntax. A Lexicalist Approach. Dordrecht: Kluwer.
Smith, Henry
1996 Restrictiveness in Case Theory. Cambridge: Cambridge University
Press.
Sorace, Antonella
2000 Gradients in auxiliary selection with intransitive verbs. Language
76: 859–890.
Stassen, Leon
1997 Intransitive Predication. Oxford: Clarendon Press.
Svenonius, Peter
2001 Case and event structure. In: Niina Zhang (ed.), Syntax of Predica-
tion, ZASPIL (Zentrum für allgemeine Sprachwissenschaft Papers
in Linguistics) 26: 1–21 [available at http://www.zas.gwz-
berlin.de/papers/zaspil/infos/].
Talmy, Leonard
1985a Lexicalization patterns: semantic structure in lexical forms. In:
Timothy Shopen (ed.), vol. 3, 57–149.
Talmy, Leonard
1985b Force dynamics in language and thought. CLS 21: 293–337.
Timberlake, Alan
104 Michela Cennamo