This paper explores detransitivising processes that may apply to the verb in Irish Sign language (ISL). These allow the signer to diverge from the prototypical features of a transitive sentence (Hopper and Thompson 1980, Givón 1990): where an Agent/Actor argument is high on a scale of topicality or salience and a Patient/Undergoer argument is correspondingly low on the scale. This scalar positioning is reflected by representing these arguments as subject and object, respectively. We identify three processes which signal the downgrading of an Actor argument and one which allows the promotion of an Undergoer. Since these are elements of voice alternations in many languages, we compare our analyses with proposals by Janzen et al. (forthcoming) to identify a rule of passive in ASL. Finally, we consider the added value that corpus tagging can bring to further analyses of similar data in ISL.
The structure of our paper is:
This paper is a report from current research on valency in the ISL verb. The paper is concerned with de-transitivisation processes, that is, with strategies that allow the signer to move away from the prototypical transitive event-description coded in the basic active-transitive sentence type.
Several aspects of our grammatical description have implications for the transcription of ISL and the design of suitable tagging systems. In the course of the discussion we investigate several important syntactic structures and their key markers, including reference shift, eye gaze, etc. and in section 6 we briefly discuss the implications of our work for corpus tagging.
Following the characterisation of transitivity in the functional literature, in particular Hopper and Thompson (1980) and Givón (1990), we identify the following semantic features of a prototypical transitive sentence (PTS):
(a) Agent: the PTS involves an active volitional agent who causes the event;
(b) Patient: the PTS also involves an inactive non-volitional patient upon which/whom the event causes a change of state;
(c) Verb: the PTS involves a verb that describes a discrete, complete, real, single event.
At a pragmatic level the PTS codes for the signer viewing the situation from the point of view of the agent. This can be viewed in terms of the agent being high on scales of cognitive salience (Givón 1990), empathy (Kuno 1987) or topicality.
We assume that there are linking rules (for example, the mapping rules of Bresnan and Kanerva 1989) which reflect the mapping between these semantic and pragmatic roles such as subject and object. A prototypical subject is a combination of agent and viewpoint, while a prototypical object is patient and non-viewpoint. In the rest of this paper we concentrate on the relationship between semantic and grammatical roles.
Grammatical roles like subject and object are defined at the level of morpho-syntax. In the double agreement verbs that form the main focus of this paper, the subject argument corresponds to the onset of the verb, while the object corresponds to the offset of the verb. This can be seen in example 1 below.
Example 1:
c. +GIVE+f.
'I gave (it) to you'.1
In discussing the semantic roles that map into these grammatical relations we prefer to employ the more general macro-role labels of Actor and Undergoer (following Foley and Van Valin 1984), since the grammatical processes we describe operate on clusters of more specific roles. For example Actor includes agent, experiencer and instrument, and Undergoer includes patient and theme.
In this paper we are concerned with grammatical strategies which move away from this prototypical transitive sentence by either or both:
(a) downgrading the role of the Actor;
(b) upgrading the role of the Undergoer.
In descriptions of spoken languages a number of related grammatical processes serve this general function of detransitivisation. These include passive voice, middle voice, reflexives, and the use of impersonal subjects.
Before considering the strategies that are used, we briefly outline the verb types that are relevant to this discussion and their more typical appearance in transitive utterances in ISL.
We follow McDonnell (1996) in identifying the following types of verbs in ISL:
In this paper we focus solely on the ways in which person agreement verbs interact with detransitivisation strategies in ISL. As such, examples listed here only include person agreement verbs, of which McDonnell identified two types – double agreement verbs and single agreement verbs.
These mark only for agreement with subject and direct/indirect object relations. McDonnell's evidence suggests that such marking is obligatory in these verb types. He classifies ISL verbs like GET-ATTENTION-OF, IGNORE, ACCUSE and GIVE as double agreement verbs as these allow for two arguments. Onset of the verb marks for subject/agent argument while offset point marks for direct or indirect object/ patient, experiencer or recipient arguments.
Example 2: Double Agreement Verb in ISL
TV-fl BOYc+IGNORE+fl TV-fl
The boy ignored the television.
Subsets of this category include backward agreement verbs, reciprocal agreement verbs, single agreement verbs and person agreement verbs with variable handshapes (McDonnell 1996: 157-167). An example of each is listed below.
Example 3: Reciprocal verb in ISL
fr+CONSULT+fl
fl+CONSULT+fr
They consulted each other.
(McDonnell 1996: 164).
It is important to note that reference shifting strategies may also be used by signers in expressing an agreement relationship between two non-first person arguments:
Example 4: Agreement between two non-first person arguments via reference shifting strategy
| (BOY) (SNOWMAN) | c+TALK+fl |
| fl+TALK+c |
In ISL, single agreement verbs mark only for direct or indirect object/ undergoer arguments, which are marked via the verb's offset locus. In this subset, there is not an explicit onset point which marks for the subject/actor argument. Where the onset of the verb occurs at 'c', the default actor /subject is first person:
Example 5: c. as default first person locus
Finally, person agreement verbs with variable handshapes incorporate information about direct object/ undergoer referents. The only feature distinguishing this category of agreement verbs from other agreement verbs is the hand configuration feature which indicates semantic variation in this verb subcategory:
Example 6: Person agreement verbs with variable handshapes
Cylindrical-entity-CL+c+GIVE+f
I handed (someone) a tumbler.
Example 7:
Tiny-0D-entity-CL+c+GIVE+f
I handed (someone) (a pill).
(McDonnell 1996: 167).
For further discussion on verb classification in ISL, we refer you to McDonnell (1996).
McDonnell follows Janis (1995) who argues that in American Sign Language (ASL), animacy interacts in specific ways with thematic roles that control agreement in the verb. Janis concludes that in ASL, inanimate referents may control agreement when these referents receive roles more typically associated with animate referents: the roles of agent, experiencer and recipient. However, the role of patient can be assigned to either animate or inanimate referents although only animate referents with the role of patient will control agreement. McDonnell argues that a similar pattern holds true for ISL, although it is unclear whether there are other contexts in ISL where animacy interacts with thematic roles in the determination of agreement. He notes three elements that relate to animacy:
Such reference to animacy hierarchies is not unusual in the literature as it relates to signed or spoken languages. For example, Kegl (1990) reports that in ASL, like many spoken languages (e.g. Navajo), a control hierarchy assigns role prominence in ranking nouns according to their ability to function as the agent in an utterance.
Kegl argues that this control hierarchy requires that the NP associated with role prominence (i.e. the NP that is co-indexed with the signer's body) rank higher on the hierarchy scale than any other argument in the utterance. She notes that when the control hierarchy is violated, 'verb-doubling' occurs. Kegl specifies the following control hierarchy for ASL:
ASL Control Hierarchy (highest to lowest) (Kegl: 1990)
First person
Second person/ third person
Animate
Inanimate
Moving
Nonmoving
Three-dimensional
Two-dimensional
Kegl argues that this hierarchy is violated when a third person is promoted to role prominence while the remaining argument, for example, a first person argument, is actually higher in terms of the hierarchy. 'Verb doubling' resolves the violation of the hierarchy in establishing a single argument taking construction: for example, a signer might establish a third person's role prominence in the first half of the 'verb-doubling construction' which leaves the object unspecified. Kegl argues that this is a passive construction in ASL.(See Kegl 1990 for examples).
The ISL literature has not considered the extent to which agreement constraints correlate with an animacy hierarchy to this degree but we do know from McDonnell's work that animacy features strongly in relation to the arguments that control agreement in agreement verbs (1990:168).
In this section, we will consider strategies that are used in detransitivising sentences in ISL.2 We can point out that several of the strategies we have noted in ISL have been reported for other sign languages, as we will discuss briefly in section 4 where we look at proposals for identifying passive voice in ASL.
For ISL, we are currently exploring the range of possible explanations available to us, but are avoiding higher level categorisation of the phenomena we have documented at this stage. We are not excluding the possibility that these strategies may in some ways operate in parallel ways to passive voice/ middle voice structures in spoken languages.
A primary characteristic of detransitivisation in ISL is the demotion of an actor. In this section we identify two important strategies for achieving this. The first type of demotion process (Demotion 1) involves a syntactically filled but semantically empty locus in double agreement verbs. The role of eye-gaze is a defining characteristic of this form. In the second type of demotion (Demotion 2) the signer's body functions as the locus for a non-specified actor. We will see that this contrasts with more typical situations where the signer's body functions as the locus for first person actions, or where reference shifting (role-shifting) occurs, for a specified non-first person actor.
Before we consider these two main types of demotion, let us look at another detransitivising strategy which involves an impersonal subject.
One feature of the prototypical transitive sentence (PTS) is that the actor argument
is higher in cognitive salience or topicality than the undergoer. Cognitive salience or topicality derives from such attributes as movement, intent, initiation, and control associated with this argument. Following Hopper and Thomson (1980), we may view this, like other features of transitivity, as occurring along a continuum, with sentences containing more or less typical actors.
In ISL we find one strategy for moving away from this feature of the PTS: sentences in which the actor is represented by an impersonal subject. These sentences represent a weakening of the cognitive salience of the actor, thus downgrading its role. They are also therefore lower in transitivity.
This strategy is characteristic of informal or narrative discourse in ISL. Typical examples include use of SOMEONE within the signed utterance:
Example 8:
fr.+SOMEONE fr.+TELEPHONE+fl. AMBULANCE
'Someone phoned for an ambulance'.
(UCC Consecutive Interpreting 1 Video Material, 'Moving Hands' Archives.)
While we can see that an actor is specified in the syntax, it is underspecified in comparison to typically active constructions. In an active construction, the actor would be explicitly specified, as we can see in the example below:
Example 9:
CARMEL c.+TELEPHONE+fl. AMBULANCE
'Carmel phoned for an ambulance'
(elicited example).
In Example 9, we could say that verbal transitivity is increased as the slot for actor is explicitly filled. We can recall that in double person agreement verbs, two slots are available to be filled, allowing explicit marking for subject and direct object. Another example which places SOMEONE in the subject/actor slot is the following:
Example 10:
ME ARRIVE+a / JOB GONE// SOMEONE sr. +GO-TO +a. COMPLETION JOB
'When I arrived, the job was gone. Someone else had already been there and had gotten the job.'
This example includes a downgraded actor (SOMEONE) which co-occurs with a marker of perfective aspect COMPLETION. We can also note that the verb in this example, GO-TO is a locative agreement verb.
We can say then that it is possible to reduce the degree of transitivity of an ISL clause by replacing the specific actor with another NP, typically SOMEONE.
In the next section we will look at demotion as a feature of detransitivisation which instead of downgrading the actor, totally omits the actor.
In this section we can examine the first, and seemingly more frequent type of demotion strategy seen in ISL, the use of a syntactic subject slot which is not filled by a specific semantic reference. We can consider several examples, beginning with (11) below:
Example 11:
FRIENDS ALL-OF-US GO-TO RESTAURANT/EAT CHAT/FEEL fr. +STARE+c. / f.r.+STARE+c. +++ /LIKE (reference shift) DEAF 'f' handshape at cheek TALK WITH HAND w.o.w./ (reference shift)THINK NEVER SEE SIGN BEFORE
'(I was with) some friends and we went to a restaurant. As we were eating and chatting I became aware of someone staring at me. The staring continued. It was like they were saying 'Oh! They're deaf and they talk with their hands'. You'd think they'd never seen sign language before.'
Here we see several features that characterise demotion (1). Firstly, we see that the locus for fr.+STARE+c. has not been established in the way that is expected of active constructions. The ISL verb STARE implies an animate actor, and from this, we conclude from context that some non-specified human is staring at the signer and her friends in the restaurant.
Second, we can see that the signer shifts reference from c. locus towards fr. where the unspecified actor who can be inferred from context as a 'hearing person' is established for the purpose of considering the rationale for the staring. Reference shifting also changes point of view from that of the undergoer to that of the actor. A second reference shift, this time back to the c. locus, allows the signer to consider that the 'starer' must never have seen a sign language before.
Third, we can comment on the role of shifted or averted eyegaze. This is a feature that plays a particular role in detransitive constructions where demotion occurs. It is also a highly visible feature in this example. While using c. locus, the signer is the undergoer of the action, i.e. she is stared at. The undergoer's gaze is averted: the signer does not look towards either the actor's nor the interlocutor's locus. Instead, eyegaze is averted towards the left. This implies that the signer at c. locus was the undergoer and that she had not initiated the actor's action. If the signer at c. locus had met the stare of the actor, meaning would change to ' I looked at them while they stared me in the face'. In this example, once the signer has shifted reference to the actor's locus, eyegaze becomes firmly fixed towards c. locus, allowing agreement to take place in the usual way.
Another example where use of averted eyegaze is criterial in identifying the lack of intentionality on the part of the undergoer is the following:
Example 12:
NEWSPAPERc. {(L) 'B'handshape}
{(R ) 'Leaf through newspaper'} TAP-ON-LEFT-SHOULDER/ (reference shift) (eyes to left, gaze upwards) MY FATHER/ (reference shift) COME FISHING...
'I was reading the newspaper when I was tapped on the shoulder (or: when someone tapped me on the shoulder). It was my father. 'Come fishing', he said. (Moving Hands Video Library, ISL Stories).
In this example, we can see that viewpoint is that of undergoer: he is sitting reading the newspaper when an unspecified actor taps the signer on the shoulder to gain his attention. Note that the undergoer is in subject position, a point we shall return to later. In comparison, the actor is located at a remote point in signing space. At this point, the signer reacts to the act of being tapped by turning his head in the direction of the actor's locus. He then tells us that the actor is his father. Next, he reference shifts to 'become' his father. He then shifts reference back to undergoer locus, becoming 'himself' again.
We have already noted that eyegaze is important in this piece. Initially, the signer is an actor, reading his newspaper, oblivious to the fact that he is being approached. At this point, eyegaze is forward and downwards, towards the location where the newspaper has been established. Following the tap on the shoulder, the signer, now undergoer, turns to the left and his eyegaze moves upwards to the left.
As such, we can say that in this example, there are two major features that indicate detransitivisation: first, the actor is not specified within the discourse, point of view remains that of the undergoer and second, eyegaze is averted for the duration of the verb phrase TAP-ON-LEFT-SHOULDER.
Another example which utilises reference shifting to indicate a non-specified actor is the following:
Example 13:
[a]PICK-PICK[b] STRONG REGULAR/ NOTHING TRAINING 'REJECTED'/ THINK IMPOSSIBLE/ THERE TRAINING THURSDAY-THURSDAY THERE (+++) STAY SAME 'PERMANANCY'
'Those strong players who attended (training) regularly were selected while those who didn't were rejected. We thought it would be an impossible task. The team trained every Thursday without fail'.
(Leeson 1996: 103).
In this example, no actor is overtly specified and point of view is that of the undergoer- those who were training on a weekly basis in the hope of being chosen for the football team. In this instance, a non-canonical locus, which we might call 'a' is where the non-specified actor is located. The signer minimally shifts reference to actor locus. We should note that the context would have established players at the locus 'b'. The actor, in this case, the team selector, is not specified. Shifts in eyegaze are minimal in this example.
In all of the examples above, the verbs in question have been double person agreement verbs which allow for one slot in the verb's syntactic structure to be unspecified semantically in these circumstances. We have also found an example where a double locative agreement verb, REMOVE-A-TUMOUR, occurs:
Example 14:
TUMOUR (be-located-at-left-side-of-chest)REMOVE-FROM(f)
'The tumour was removed'.
Here, the signer shifts reference to the locus that has been established for a sick woman suffering from cancer. The cancerous tumour is identified by the signer as being located in the undergoer's left chest area. A non-specified actor removes the tumour from the undergoer in a hospital situation. As with many of the examples we examined earlier, the undergoer's eyegaze is averted for the duration of the actor's action.
One of the interesting features of this example is that the signer reference shifts into the role of the undergoer. Pragmatically, this lends empathy to the experience of the undergoer and increases the degree of attention that is focused on this role. It would equally have been possible to reference shift to the locus of an established actor, the surgeon, and have discussed the operation/ procedure from that point of view.
At this stage, we do not know to what extent other locative agreement verbs appear in detransitivised utterances like this. We hope to develop a clearer understanding of the constraints that operate in this kind of demotion.
In addition to demotion (1), we have found that signers also use a second detransitivising strategy in ISL, that which we are currently referring to as demotion (2).
In situations where Demotion (2) occur, the locus that is typically reserved for first person reference is instead used to represent an unspecified actor. One example is the following where a signer is discussing the development of linguistic research in ISL:
Example 15:
SAME KNOW JAR [hold left hand]
[right hand]: PUT-IN/ CL-C 'close lid' NAME LABEL CL.G ['be located on jar'] CL.G [be located at f-hi.] [R] locus a, [L] locus b, [R] locus c, [L] locus d.
___________q
KNOW NAME LABEL – [be located at pre-established loci a,b,c,d].//RESEARCH PROCEED/JAR SAME AGAIN c.+TAKE+a. [move to c.] AGAIN/ CL.C. ['take off lid'] CL.G. c.+TAKE-OUT-OF+ a. /CL.4 [be-located at hi., c., loc.] ['classify']...
'Its like taking a specimen jar and placing a specimen in it, labelling it and placing it on a shelf along with many other similarly labelled specimens. As research progresses, those same jars can be taken down again and the specimens examined. (or: Someone can take those jars down again). The specimens can be classified'.
(IDS Video Library).
It has been argued that in typical passive constructions, the actor is often derivable from context or is obvious. In this instance of detransitivisation, it is possible to suggest that the actor is derivable from context: the subject of linguistic comparison is the general topic for this narrative and the issue of research is broached in an earlier segment of this text. We see that the actor is derivable from context: some animate actor, but not the signer will re-examine some previously collected linguistic samples (from ISL) and categorise them.
We have seen that there is a tendency in detransitivised examples for point of view to be that of the undergoer. Here, it is possible that this is not feasible grammatically without creating a semantically marked utterance. It seems that inanimate objects are not as central as animate objects in terms of encoding point of view / perspective, so for example, a jar does not typically function as an actor. However, in this example, the identity of the undergoer is evident while the identity of the actor is not.
We can also note that as with examples considered in demotion (1), we see a syntactically filled but semantically unspecified locus for the actor's role at c. space: 'JAR SAME AGAIN 1 +TAKE +a [move to c.] (' those same jars can be taken down again') and 'CL.C. [take off lid] CL.G. 1 +TAKE-OUT-OF+c. ('and the specimens examined') are two agreement verbs where the actor is not specified. No NP has been established but the locative agreement verbs still behave as they would in an agentive construction with an overt subject/agent. Syntactically, use of canonical space establishes the syntactic argument structure necessary for a grammatical utterance.
This use of c. locus is the characteristically defining feature of demotion (2). In a typical active construction in ISL, c. locus is used as a default first person reference locus. However, in this context it is clear that the signer was not referring to themselves as the researcher but to the process of research and hypothetical future researchers.
Another example which uses c. as locus for a non-specified actor is the following:
Example 16:
DEAF HAVE NOTHING HERE IRELAND/ FOR EXAMPLE NOW ONE SINCE (hold) ONE RECENT ASSESSMENT/ PUBLIC SERVICE/ ONE PUBLIC SERVICE sr.+GRAB+sl. FROM DEAF (sl.) AND MOVE-TO +s.r. CHARITY ORGANISATION
'Deaf people have nothing here in Ireland. For example, recently there was an assessment of public services. A public service was taken from the Deaf (community) and relocated with a charitable organisation'. (IDS 2nd Congress. November 1993. 'Human Rights of Irish Deaf People')
This example sees the signer establish the Deaf community at a location to the left side of signing space while the charitable organisation is located at the right hand side of signing space. Although we might expect a signer to establish a referent in space before they articulate an utterance that encodes a motion event, this does not happen here. Instead, the signer aims to background the referents and highlight the act of a public service being taken from the Deaf community. This is effected through use of a number of strategies:
1. Demotion of actor: we do not know who is responsible for removing the public service from the Deaf community and relocating it under the auspices of another charity organisation. As such we can say that both syntactically and semantically the actor is demoted. For us, context does not regenerate a possible actor. We might imagine however, that for members of the Deaf community at the time of the utterance, the loss of a public service would be highly topical and as such, veiled reference would serve only a pragmatic/ politeness function.
2. Promotion of undergoer: in this example, we are focusing our attention on the undergoer, the public service.
3. Animacy hierarchy controls agreement: while this event is not described from the point of view of the undergoer, we know that typically in ISL, while inanimate entities can function as undergoers, only animate referents with the role of patient (in macrorole terms, undergoer) will control agreement (McDonnell 1996: 168). It is possible that if this utterance was part of a parody, the signer might well have shifted reference and 'become' the public service and walked to another location. However, in typical ISL discourse, this does not happenIt is also worth noting that the handshape used in the verb MOVE-TO is suggestive of an animate actor, so while the actor is not explicitly identified, their role is implicitly marked in the morpho-syntactic properties of the verb.
4. Use of canonical space (c.) as non-specified actor locus: here,the actor is not assigned a non-central locus within signing space per se. In the examples we considered in relation to demotion (1), we noted that syntactic locations were established within signing space which became reference points for actor activity. In comparison, in this example, as with the last example we looked at, the signer's c. location seems to serve as a default location for agent-type action: it is clear that the signer is not the actor.
5. Pragmatic constraints on interpretation of utterance: we can conclude that the signer in this example is not the person responsible for removing a public service form the Deaf community. As a Deaf person addressing an international Deaf audience, including representatives from the European Union of the Deaf and the World Federation of the Deaf, it is impossible to argue that this is a first person as actor construction. It seems then that we must account for the occurrence of demoted actor using c. locus in another way.
One way of examining the data is in light of mental spaces theory (Fauconnier 1997), drawing on the work of Karen van Hoek (1996) who applies Fauconnier's theory to ASL's use of referential loci. Van Hoek's work aims to establish a relationship between mental spaces and referential loci on the basis of evidential support for the view that referents in a discourse event are accessed even when other loci have previously been used for the same pronominal referent. She suggests that the principles of locus selection in ASL seem congruent with the general principles of accessibility developed by Givón (1989) and Ariel (1988, 1990). She says that:
'Accessibility Theory holds that a particular nominal form is selected for reference in a given context to reflect the degree of accessibility (roughly 'retrievability') of the referent in that context. Cross-linguistically, full nominals (names and descriptive phrases) are markers of relatively low accessibility, used where the referent is not highly active in the addressee's awareness. Pronouns are markers of relatively high accessibility and null anaphora (i.e.) no phonological marking of co-reference) marks still higher accessibility (1996: 337).'
It is possible that this plays some role in the fact that fluent signers can distinguish between c. locus as first person and c. locus as non-specified actor in discourse. A second point that Van Hoek's work raises is that in the detransitivised examples we have examined, the undergoer is typically the most highly activated referent in both the signer's and the addressee's awareness which, in the signer's case has provoked choice of demoted actor as subject. As such, we might concur with Van Hoek that:
'Accessibility is essentially a matter of how 'active' a referent is in a conceptualiser's awareness. It is influenced by salience, both perceptual and conceptual (1996: 338)'.
This seems to offer one possible interpretation for the overriding of the default where c. typically represents first person or, after a reference shift, a non-first person actor.
In this section we have looked at three strategies that are used in detransitivising utterances in ISL: the use of an impersonal subject and two strategies for demoting an actor. Demotion (1) involves the use of a syntactic subject slot that is not filled by a specific semantic reference, while demotion (2) typically uses canonical space as a non-specified actor locus. In the next section we will look at the role played by promotion of undergoer.
In many languages the basic passive construction combines the promotion of undergoer with demotion of the actor. In contrast, we have seen that demotion of an actor can take place without a concomitant promotion of an undergoer argument. However, we have found examples where demotion (1) co-occurs with promotion of undergoer.
The 'fishing story' example (12) that we looked at earlier provides a good example of promotion of undergoer co-occurring with demotion of actor. When we looked at this example, we made reference to the fact that 'point of view' is that of undergoer. Here we can say that this arises because the undergoer has been promoted to subject position where he is 'acted on' by the demoted actor.
Other examples that illustrate promotion of undergoer include use of double person agreement verbs: BEAT-UP and RAPE:
Example 17:
Example 18:
There are several features to note in the articulation of these verbs. In the (a) examples, we find that the articulation of RAPE and BEAT-UP occurs in citation form: the palms are oriented away from the signer's body.
In the (b) examples, we see that the actor is explicitly referred to while the undergoer is implied by the change in orientation of the palms in the articulation of the verbs. When this occurs, the signer does not explicitly refer to the undergoer as ME. Instead, the singer's body functions as the undergoer, which we can argue is an instance of undergoer promotion. We can see that this has occurred in the (b) sentences where two arguments are present.
This is also a feature of the (c ) sentences. Again, we see that orientation of the palm differs from citation form in that the palms are towards the signer. Location is that of the signer's body. In contrast with the (b) sentences, the (c ) sentences are marked by the fact that the subject/actor argument is demoted. We can say that at an explicit level, the actor is omitted, but given the nature of these verb forms, animate actors are implied.
Another feature that we have commented on before is that of eyegaze. In the (a) sentences, we would not expect to find averted eyegaze. However, in the (b) sentences, averted eyegaze is possible, but not necessarily obligatory. In contrast, it seems that the (c) sentences demand averted eyegaze to mark for lack of intension/co-operation on the part of the undergoer. This correlates with the obligitory marking that we saw in demotion (1).
It seems that this is the common thread that runs between the two types of demotion we have considered. As yet it seems unclear what the exact conditions for choosing demotion (2) over demotion (1) are, though there seems to be a bias towards using demotion (2) when referring to first person as undergoer. This is an area that requires further analysis.
Returning to the comparison between the (b) and (c) sentences above, we can note that the features we have identified in the (c) utterances cannot be described as underspecification (where the subject or another argument is not mentioned) rather than detransitivisation. Unlike the prototypical transitive sentence the actor argument is omitted. Note that no actor referent has been previously established in the discourse. Even where the verb implies an animate (probably human) actor, the role of actor has not been explicitly identified within the clause.
Similar features have been noted for ASL by Terry Janzen and colleagues in New Mexico (forthcoming). They argue for the existence of a passive construction in ASL which can be characterised by the following key features:
We can look briefly at each of these criteria:
Demotion of the agent: Janzen et al note that in prototypical ASL examples, the agent is assumed or obvious. In less prototypical examples, the agent may be mentioned, but not as the subject of the verb. Another alternative in such situations is that the agent appears in an NP that is lower in transitivity which would be glossed as SOMEONE or WHO. This relates to the fact that signers can choose to be vague in their utterances which is marked by constructions that use an indefinite actor and are lexicalised by SOMEONE, PERSON or SOMEWHERE. We have seen that similar strategies are used by ISL signers.
Events viewed from perspective of the patient rather than the agent: In typical active utterances, point of view is that of the agent. Janzen et al point out that for ASL, many verbs are articulated with onset beginning at a point in space in front of the signer while offset occurs at a second point in signing space. They argue that certain features surrounding use of signing space are associated with the agent, for example, the spatial locus at verb onset. They note that while in active constructions the signer's shoulders lean slightly in the direction of the agent positioned in signing space, while eyegaze is in the direction of the patient, passives are established differently:
'In the passive, the patient is more accessible, or given information than is the agent. The patient is marked in the verbal agreement system as the final locus of the verb movement, but the event coded by the verb is viewed clearly form the perspective of patient rather than of the agent. The identity of the patient is evident, whereas that of the agent is not. The signer may move his or her shoulders and torso slightly away from the spatially positioned agent, with eye-gaze directed towards the agent of the action' (ibids.:6).
While the ASL passive may be formally marked as the topic constituent, it is presumed that this is not the major function of topic marking. They also argue that while a patient may be marked as a topic, the order of the verb and its argument(s) may satisfy the definition of an active clause.
Agent specified in syntax with empty locus in agreement/directional verbs: Janzen et al argue that with an agreement (or directional) verb, prototypical defocusing means that instead of an agent being specified in the syntax by associating an agent NP with a particular locus, the locus is empty. However, they note that verb movement must still have an onset at some locus in signing space.
They comment on the widely accepted view that a locus must be semantically designated before it is used: that is, a point in space is typically established as being coreferential with a specified referent prior to pronominal reference being established. They argue that while this is the common understanding of how loci work, with a passive construction, no agent is specified and such 'movement of the agreement verb begins at a syntactic, but semantically empty locus' (ibid:7). In the prototypical passive in ASL then, the agent is not identified.
This kind of occurrence is said to differ from so called null arguments where lexical items are not signed in local discourse having already been established in an accessible way, as such allocating a semantically valid reference framework for the ongoing discourse (ibid.: 7). Similar arguments have been made regarding spoken 'Pro-Drop' languages, like Italian and Spanish.
We have looked at several strategies that are used by ISL signers to diverge from the prototypical transitive sentence. We have described several strategies which downgrade or delete the actor argument: the use of impersonal subjects, the use of an empty subject locus with double agreement verbs; and the 'unusual' use of the c. locus. We have seen that there are consistent strategies used across various types of double agreement verbs (i.e. person and locative agreement verb classes both allow for these strategies to occur). We noted that eyegaze is a criterial feature that seems to mark for lack of intention or awareness on the part of the undergoer in utterances where detransitivisation strategies are used.
While we cannot categorically state what the differences are between demotion (1) and (2), we have identified some patterns emerging: there seems to be a preference for using demotion (2) when referring to first person as undergoer.
We have also looked at promotion of the undergoer. We noted that this can co-occur with demotion strategies, although it does not seem to be the case that such promotion must co-occur with demotion. It will be interesting to examine further data to determine the extent to which these features co-occur to from a strategy close to the prototypical passives found in many languages.
We also aim to conduct a more in depth analysis of the theta roles that occur in detransitivised situations, which should lead us to a clearer understanding of the semantic constraints that operate. It may also be the case that other factors operate at the pragmatic level, for example, signer use of empathy or distancing strategies.
Given the relative lack of research that has been carried out at the morphosyntactic level in ISL, no established approach to transcribing and/ or tagging of data has been established. However, we are sure that the development of tagging systems will be of great assistance in the development and identification of other morpho-syntactic patterns in ISL. While we have not used a tagging system in the course of analysing data described in this paper, we recognise the value of such an approach and in this section will outline how we envisage tagging could be used to assist further analysis of the kind of utterances we have described.
In the course of this paper, we have identified a number of key parameters that identify specific types of detransitived utterances in ISL. The features we have identified could be highlighted in a tagging system to enable a more rapid identification of such utterance types which in turn would provide a more realistic awareness of the degree of frequency of such utterances in ISL.
Particularly, we have noted that a number of features tend to cluster in the marking of detransitived utterances which we have referred to as 'Demotion (1)'. We have seen that averted eye-gaze tends to co-occur with an agreement verb that has its on-set at a locus that has not previously been established. This specific clustering is found in all of the demotion (1) examples and seems to be obligatory.
We also found that with several verbs like STARE (Example 9), shifted reference occurs to allow the signer to change point of view from that of undergoer to that of actor. It is as yet unclear as to whether this is a pragmatic function or a mandatory feature in the establishment of demotion (1). Evaluating the actual co-occurrence of these features could be easily conducted with the aid of a tagging system and could lead to clearer identification of the patterns relevant to the location of unspecified actor loci in signing space in this kind of detransitivised utterance.
Tagging also has a potential role to play in the identification of utterances like those described as demotion (2). We noted that the characteristic defining feature of demotion (2) is the way in which the locus that is usually reserved for first person reference (the 'c.' locus) is instead used to represent an unspecified actor (E.g. Example 13). As yet, it is not possible to state categorically the differences between demotion (1) and (2), but the general tendency seems to be that signers prefer to use demotion (2) when referring to first person as undergoer. Through use of tagging it may be possible to identify frequency of such occurrence and identify the thematic roles that occur most frequently in this position.
As a matter of convention, we have used the glossing framework of Patrick McDonnell (1996), who in many instances adopts the principles used by Engberg-Pedersen (1993). A locus is described as a morpheme that is expressed through the way in which it influences the position and / or orientation of the hand(s) in the production of a sign: because of this, Engberg-Pedersen uses directional expressions to symbolise the most commonly used locus markers. We adopt McDonnell's (1996: 37-38) 'plan' and 'elevation'views of locus markers:
|
|
||
| . | ||
|
f(orward) l(eft) .
|
.f(orward) r(ight) | |
|
|
||
|
s(ideward) l(eft).
|
|
.s(ideward) r(ight) |
Figure (A1): A 'plan' view of locus markers (McDonnell 1996: 37)
| .hi | |
| Signer c. | .f |
| .lo |
Figure (A2): An 'elevation' view of locus markers (McDonnell 1996: 38)
McDonnell (1996: 38) discusses the relevance of locus marking for person in Irish Sign Language. In particular, he notes that ISL has a category of person which distinguishes between first and non-first person. He bases this distinction on the pattern of articulation found in deictic signs. Regarding use of the c-locus and of the f. locus, McDonnell says that:
"Articulation at the c-locus means that the sign is articulated in contact with, or proximal to, the signer's body; articulation at a locus such as 'f' means that the sign is articulated away from the signer, meaning at a distance that is medial or distal from the signer's body (Liddell and Johnson 1989). In Irish Sign Language the deictic signs translated in English as 'I', 'me' and 'my' are all articulated at the c-locus. The deictic signs translated as 'you', 'him'/'her/ 'it', 'your'/ 'his'/ 'hers', 'them' and 'theirs' are all articulated at non c-loci" (McDonnell 1996: 38).
This is particularly relevant to our description of non-typical use of the c-locus in the section that deals with what we call Demotion (2).
Given that the initial examples of verb categories in Irish Sign Language are taken from McDonnell's work, we decided to gloss our own examples in the same manner to allow for consistency across the examples. It is anticipated that this will be helpful for the reader.
Bresnan, Joan and Jonni. M. Kanerva 1989. 'Locative Inversion in Chichewa: A Case Study of Factorization in Grammar', Linguistic Inquiry 20:1: 1-50.
Fauconnier, Giles. 1997. Mappings in Thought and Language. Cambridge: Cambridge University Press.
Foley, William and Robert Van Valin. 1984. Functional Syntax and Universal Grammar. Cambridge: Cambridge University Press.
Givón, Talmy. 1990. Syntax; a Functional-Typological Introduction. Amsterdam and Philadelphia: John Benjamins.
Hopper, Paul J. and Sandra A. Thompson. 1980. 'Transitivity in Grammar and Discourse,' Language 56: 251-399.
Janzen, Terry; Barbara O'Dea and Barbara Shaffer (forthcoming): The Constural of Events: Passives in American Sign Language. In Sherman Wilcox (ed.): No title available. University of New Mexico.
Kegl, Judy. 1990. Predicate Argument Structure and Verb-Class Organisation in the ASL Lexicon. In Ceil Lucas (ed.) Sign Language Research: Theoretical Issues. Washington DC: Gallaudet University Press.
Kuno, Susumu. 1987. Functional Syntax: Anaphora, Discourse and Empathy. Chicago: University of Chicago Press.
Liddell, S. and R. Johnson. 1989. American Sign Language: The Phonological Base, Sign Language Studies, 64: 195-277.
Leeson, Lorraine.1996. The Marking of Time in Sign Languages with Special Reference to Irish Sign Language. Unpublished M.Phil. in Linguistics dissertation. Centre for Language and Communication Studies (CLCS), University of Dublin, Trinity College.
McDonnell, Patrick. 1996. Verb Categories in Irish Sign Language. Unpublished doctoral dissertation. Centre for Language and Communication Studies (CLCS), University of Dublin, Trinity College.
Van Hoek, Karen. 1996: Conceptual Location for Reference in American Sign Language. In Gilles Fauconnier and Eve Sweetser (eds.). Spaces, Worlds and Grammar. London and Chicago: The University of Chicago Press.
1 For details of the transcription used in examples, please see ‘Appendix: a note on transcription’ at the end of this paper. (back to text)
2 It should be noted that all ISL examples referred to in this section are drawn from either Irish Deaf Society or ‘Moving Hands’ archive materials, all of which have been recorded by deaf ISL users.(back to text)
Posted:9.5.2000