Professional Documents
Culture Documents
1. Introduction.
Ergativity has received quite a lot of attention in the last few years. Among the
reasons for such an increase in studies on ergativity is the attention paid to languages
which had not been studied before. But the term has been used to refer to a wide range
of different phenomena. As Dixon points out:
`Ergativity' is currently an `in' term in linguistics. It is used by a wide
variety of linguists, with a whole range of different meanings. As a result, much
confusion exists at present about what an 'ergative' language is, and about the
morphological, syntactic, and semantic consequences of such a characterization.
(Dixon, 1979: 59)
Intransitive subject: S
Transitive subject: A
Transitive object: O
Accusative Ergative
S=A S=O
O A
A further step in the study of ergativity has led linguists to believe that, strictly
speaking, there are no ergative languages: all those languages primarily thought of as
"ergative" also include accusative structures in their systems.1 And many accusative
languages have also been found to have some type of ergative organization.
2. Types of ergativity
The term ergativity has been progressively extended to any structure that treats S
in the same manner as O and differently from A. This phenomenon may show up in a
variety of different linguistic forms. In addition to morphological ergativity, new levels
of ergativity have been discovered: syntactic, discourse and lexical ergativity.
Morphological ergativity is found, among other languages, in Basque, which has a
different morpheme for S and A. S is marked as absolutive, like O, whereas A is
marked with the ergative morpheme {-k}:
Dyirbal, for instance, also shows ergativity at a syntactic level with the equi rule,
which allows deletion of an NP identical to a previous NP in a coordinate clause, if both
NPs are in S or O function. Thus in Dyirbal in a sentence like (2a) the S of the second
clause is taken to be identical to O, and not to A as would be the case in English. So the
whole sentence would be interpreted as (2b).
1 According to Verhaar (1990:153) "No known language is entirely ergative". Though it has been
pointed out that Basque is fully ergative at the morphological level. See Dixon (1994:2).
2 Examples from Moreno Cabrera (1991:423).
3 See Dixon (1987:4-5).
4 See Givón (1983).
3. Lexical ergativity.
As far as English and Spanish are concerned, we find some verbs used both
transitively and intransitively, the intransitive sentence having as subject the object of
the transitive counterpart. In such structures S and O behave in the same way, i.e. they
keep the same relation with the verb, and are different from A:
Other English and Spanish verbs relate their subjects and differentiate them from
O. Such is the case, for example, of the following sentences:
Thus we could say that both English and Spanish mix ergativity (3) and
accusativity (4) in their lexicons. But this type of ergativity which is restricted to the
lexical domain has to be clearly differentiated from the original sense of the term, which
makes reference to case-marking. English and Spanish are accusative languages at both
morphological and syntactic levels, but they have some lexemes which are ergative.5
Therefore, I will use the term ergative in its lexical sense, to refer to the semantic
parallelism between S and O,6 as exemplified in (3).
There are three types of structures where a verb may be used in both a transitive
and an intransitive pattern, where the O of the transitive pattern appears as S in the
intransitive counterpart.
Structures (5), (6) and (7) share this alternating organization of an identical
semantic role as S or O. Therefore, they may be grouped together under the term
ergativity.7 But this alternating relationship between semantic role and syntactic
function is not the only defining feature of these structures. Actually, they are not
always grouped together. A simple look at some of the terminology employed to refer
to them will give us a clear idea about the different perspectives under which they have
been analyzed. Notice that the only unifying term used to cover the three types is
ergative:
(5)
term source
pseudo-intransitive Kilby (1984:45-48)
derived-intransitive O'Grady, (1980)
middles Keyser and Roeper (1984) Fellbaum (1986)
activo-passives Jespersen (1927:347-355)
ergative Huddleston (1971: 65)
process-oriented clauses Halliday (1968: 47)
patient-subject construction Van Oosten, 1977
medio-passive Grady (1968)
undergoer verbs Verhaar (1990)
potential or hypothetical clauses Hatcher (1943)
promotion-to-subject construction Dixon (1991: 322-335)
passival verbs Sweet (1891:90)
(6)
term source
ergative Van Riemsdijk and Williams (1986), Van
Riemsdijk and Williams (1986)
alternating intransitives O'Grady (1980)
anticausatives Siewierska (1984: 77), Kholodovic (1969),
Masica (1976), Comrie (1977)
inchoative Fellbaum (1986)
unaccusative or absolutive Keyser and Roeper (1984)
theme verbs Wunderlich (1985)
unaccusative Fellbaum (1989)
(7)
term source
unergative verbs Ravin (1991: 107)
active intransitive verbs of locomotion Hale and Keyser (1986)
ergative Huddleston (1971:65); Halliday (1967:46);
Noguchi (1989:159) Dixon (1987:7)
This use of the term ergativity at a lexical level has been condemned by some
linguists.8 However, even those who prefer to discuss these structures under a different
rubric, take it as a starting point, since, whether acceptable or not, it is the term most
generally employed.9 I will subclassify these ergative pairs into three groups that I will
label derived intransitivity, alternating transitivity and derived transitivity ,
respectively.
No Spanish Grammar as far as I know, has used the term ergative as it has been
used to refer to any of the English structures exemplified by (5), (6) and (7) above.
However, we also find ergative verbs in Spanish, though most verbs take the clitic se in
the intransitive pattern.
8 See Dixon (1987:7): "this usage must be deplored, since it can only serve to confuse unwary
readers, and make it harder for them to appreciate the nature of morphological, syntactic and discourse
ergativity"; Dixon (1991:323); or Comrie (1978:391) "In much recent work on ergativity, or at least in
much recent work using the term ergativity, the term has been extended to cover sets of sentences like the
following English pair John broke the window/the window broke (...) This use of the term can be very
misleading". Zwolanek (1987:59) also condemns the use of the term ergative to refer to John in a
sentence like John broke the window and absolutive for the window in either the transitve or intransitive
pattern: "Diese Verwendung des Begriffs 'Ergativität' ist sehr irreführend".
9 See Lyons (1968:352): "(...) the term that is generally employed by linguists for the syntactic
relationship that holds between (1) The stone moved and (3) John moved the stone is "ergative"". Lyons
(1968:342) uses the term to refer to structures like (9). For Halliday (1967:46 ff) the following subjects
are ergative: i. She washed the clothes. (actor in effective) ii. He marched the soldiers. (initiator in
descriptive). Huddleston (1971:65) uses the term in a wider sense: "It will be convenient to have a simple
term to refer to the verbs entering into such contrasts: I shall speak of them, informally and somewhat
loosely, as 'ergative' verbs, since they suggest an ergative organization of the clause (one where a one-
place verb, i.e. a verb combining with a single NP, takes an 'affected' element as subject, while a two-
place verb takes a 'causer' as subject and an affected element as object) rather than a transitive
organization (one where a one-place verb takes an 'actor' as subject and a two-place verb an actor as
subject and a 'goal' as object)."
b. El perro paseó por el jardín.
5. Derived intransitiviy.10
In (11), the adverb implies that the result of the action is due to a positive property
of the patient. In this sense these derived intransitives are patient-focused as Fellbaum
(1986:21) labels them or process-oriented as Halliday (1967:47) terms them, in
opposition to passives which are agent-oriented.12
If, on the contrary, we are dealing with a negative property of the subject, then
negation may replace the adverbial:
10 For a more detailed study of these structures see L. González Romero (this volume).
11 Fellbaum (1986:9) calls them facility adverbs.
12 The agent in these structures is always implied since these verbs always involve two
participants, but it cannot be made explicit.
13 Example from Kilby (1984).
Sometimes an emphasis on affirmation replaces the adverbial:
14 Notice that in these examples we find an epistemic modality: the speaker somehow grants the
content of the proposition. As regards the verb sell in Spanish, María Moliner and DRAE describe it only
as transitive or pronominal. However, I have recently found it in an intransitive pattern: La nostalgia
vende bien. (El País, 3/1/96, p.53).
(32) María se asusta fácilmente.
(33) Mary gets scared easily.
Another restriction of these derived intransitive patterns has to do with their tense-
aspect reference. These sentences show a generic reference,15since they state properties
that are held to be generally true, and they don't refer to specific events in time16. Hence
the unacceptability of (34-35).
They are generally found in a simple present tense and making reference to
timeless situations. However examples like (36) are not rare:
6. Alternating transitivity.18
We find a different type of ergative structure with verbs like move, open, close or
break, which alternate the transitive and the intransitive patterns. Some linguists like
Anderson (1968), Lyons (1968) or Halliday (1970) claim that they are basically
intransitive and that their transitive use is derived. Others like Comrie (1977)
Kholodovic (1969) or Masica (1976) call these verbs anticausative, suggesting an
intransitive origin, thus the transitive being derived. In Spanish we find verbs like
15 See Keyser and Roeper (1984:384), O'Grady (1980), and Verhaar (1990:132) on this point.
Fellbaum (1986:4-5) points out that it is not the action that is generic in these structures: it is the agent
that is nonspecific. I find it confusing to talk about a non-specific agent. This would associate these
derived intransitives with passives since as Stein (1979) points out between an 80% and 85% of English
passive clauses do not present the by + agent phrase.
16 The Spanish verb colgar, originally transitive, has an intransitive use which is clearly stative: La
ropa cuelga del balcón.
17 Notice that they do not appear in imperative or progressive constructions (as Keyser and
Roeper, 1984:385 point out):
i. *Bribe easily, bureaucrat!
ii.*Bureaucrats are bribing easily.
This will lead Verhaar (1990:133) to conclude that they are stative verbs.
18 I use the term transitivity in this case to refer to both the transitive and intransitive structures.
See Verhaar (1990:93): "'transitivity', a term which now increasingly functions as a superordinate term
for what is 'intransitive' and 'transitive'."
aumentar, disminuir, subir, etc. The class of verbs is enlarged by many others that
appear as intransitives with se. As in English, in Spanish there is not a unanimous
opinion as to which structure is the original. Thus in the DRAE aumentar is considered
as a transitive verb which may derived into an intransitive. María Moliner, on the other
hand, presents it as originally intransitive. Other Spanish verbs in this group are hervir,
cocer, colgar, disminuir, mejorar, empeorar, subir, bajar, pasar, variar, cambiar, etc.
The situation denoted by the verb in the intransitive pattern of these alternating
structures is not necessarily dependent on the intervention of an agent. The subject,
typically inanimate, is again a patient subject, but the event is self-originated, the verb
behaving almost as inherently reflexive. Thus, for instance, the verb open cannot be
associated to an accidental action if it appears with the NP can. (42); the verb empezar
is only possible in the intransitive pattern with eventive subjects, (44), since a non-
eventive subject, (46), would make the reflexive interpretation impossible.
The action is brought about accidentally; there is no agent explicit nor implicit:
It is interesting to remark that some verbs may appear both as derived intransitives
or as alternating transitives, depending on the situation implied. If it is interpreted as
self-originated it will be built as an alternating transitive structure (51). If an agent is
implicit, then it will have to be built as a derived intransitive (52), with all its constraints
(i.e. addition of an adverbial, potential or hypothetical sense, etc.). It would be
ungrammatical otherwise (53).
Also interesting to note is the fact that the formation of these structures is
restricted to the lexical domain. Thus semantically equivalent verbs are not necessarily
built in the same way. In Spanish, empezar or comenzar are ergative with eventive NPs,
but iniciar is not.
7. Derived transitivity.
The third group of structures with this diathetic alternation that links S and O,
which we have loosely called ergative, is formed by active intransitive verbs like walk,
march, swim, etc. which admit the addition of an NP subject to form a causative
structure.
The original structure involves one participant in a high intransitive structure, (61,
63). By adding an instigator we form the transitive structure, (62, 64). The subject of the
intransitive moves then to the position of object. The resulting structure has two active
participants: the subject or instigator and the object or agent.
All these structures need an animate agent, capable of performing the action. This
agent is the subject of the intransitive sentence. In the transitive counterpart the subject
is always instigator.
In English these causative constructions are said to be "exceptional", formed only
by a small group of verbs like gallop, march, walk, run, fly, jump and swim.20 I don't
think that this class is restricted to these motion verbs. The verbs work, dance, exercise,
and train also build similar ergative structures:
In Spanish they are less productive, but nevertheless the grammar permits them.
19 "Pasear es un verbo básicamente intransitivo que, no obstante, puede emplearse con valor
transitivo cuando adquiere un sentido factitivo: pasear el niño." Roca Pons (1970:233).
20 Noguchi (1989:159 fn 6): "We will not go into the dicussion of agentive objects exemplified in
(ib) below:
(i) a. The horse walked.
b. John walked the horse.
We simply note that this type of alternation is truly exceptional. Indeed, there are only a few verbs
which are involved in this type of alternation; gallop, march, walk, run, fly, jump, swim are the only
possible candidates that I have actually encountered. Rather, we take the existence of this type of sporadic
causativization to be indicative of the generality of the EA (ergative alternation)." See also Olszewska
(1974:254).
(81) Resucitaron a los muertos.
(82) María ascendió.
(83) Ascendieron a María.
8. Concluding remarks.
We may sum up by stating that lexical ergativity is also found in Spanish with
some verbs that alternate in transitive and intransitive patterns. The defining feature of
these structures is the presence of an NP which may appear either as O or S. We have
analysed three different groups of clauses which involve different syntactic behaviours.
In a first group we faced a phenomenon of derived intransitivity: the subject is
clearly a patient as in passives, but no agent can be expressed, even though there is
always one implicit. They emphasise the process and they involve a certain type of
modality.
In the second group we find verbs which alternate the transitive and the
intransitive patterns with the same range of productivity. This group of verbs forms a
larger class both in English and in Spanish. The intransitive pattern is seen as an
accidental process, therefore no agent is implied.
The third class is formed by intransitive verbs which may appear as transitives by
introducing a subject as instigator of the action. Unlike the two previous structures, in
these clauses the object of the intransitive is agentive.
References.