Antw: Re: O-only agreement

Sonja Erlenkamp sonjaerlenkamp at HOTMAIL.COM
Mon Sep 21 12:30:30 UTC 2009

Since I am on my way to a conference I have not the time now to answer to everything in your message, Franz, but I am very curious about one thing:


You write:


Sonja Erlenkamp and Dan Slobin talk about their linguistic models. They should not mix up them with language or "reality" (e.g. by saying: "the language phenomenon X IS a subject or an agreement").


I cannot see that I wrote that. 


As far as I can tell do you present your linguistic model as well (which - as mine - is a cognitive approach, although obviously somewhat different from mine). And you argue through the filters of your model. Fair enough. But how are we able to talk about reality and language without a model? It is my believe that we never will be able to describe reality in other terms than through models. Whenever I give a description of what is going i a signed language it is meant as a model - hopefully one that matches reality as much as possible for the time being, but there will always room for improvement. In other words: I think the term "agreement" as used in many spoken languae models does not give a sufficient picture of what is going on in meaning construction in the signed languages I have seen. Of course we can change the notion of agreement or other notions (like S,A and O) something often done in (signed)  linguistics, but in many cases this leads only to more confusion since we end up comparing apples and pears. For me it seems to be more important to give a good model for the mapping between form and meaning in any signed language and thus develop a grammatical theory of signed languages in general, than adopting terms from spoken language research which do not fit properly - or as Müller Gotama 1994:1 puts it:

 “Any grammatical theory must ultimately be concerned with how it handles the mapping of meaning onto grammatical form if it is to be successful as an account of human knowledge of language.” 
Thus said: of course do sign languages mark semantic roles, everything else would be very inefficient for communication, wouldn't it. But the means used to do this are IMHO so much different from what is called agreement (following for example the model for spoken language agreement applied in this discussion) or even subject and object (following Comrie's model as a correlation of markings between S and A in opposition to O) that I'd rather not use these terms. But in the end it's of course all about the models. Otherwise we wouldn't have any discussions ;) 
All the best


Date: Mon, 21 Sep 2009 09:40:13 +0200
From: Franz.Dotter at UNI-KLU.AC.AT
Subject: Antw: Re: O-only agreement

Dear colleagues,
I am rather sad about the fact that the "language units have to be arbitrary"-stereotype gets revived in morphology now. That goes together with the fashion of the use of "blending" (following Liddells publications): If one finds an item which has some existence outside language (as, e.g. location), then they argue that we have "blending of language and non-language". 
I take a cognitive perspective which says that language is "a phenomenon of the whole body" and only asks whether there is "some coding for some meaning" (keeping the axiom that cognitive/meaning items can be a basic standard of comparison of languages). Under this respect it is more adequate for me to interprete what some people call "blending", is the human procedure to exploit the non-language area for items which can be used as languae units. 
>From that standpoint it is also clear that sign languages can transport what linguists call 'roles', especially by using location or sign movement direction. For me it is not adequate (neither for spoken nor for signed languages) to hypothesize that any language utterance may systematically consist of sequences like: "language unit(s) - non-language unit(s) - language unit(s) - non-language unit(s) - …".
In other words: From a phenomenological view, we have "coding for roles" in sign languages. Which formal status these coding may have (e.g. whether they can be derived from other language phenomena in a certain model by certain means), is a secondary question, interesting for anthropology, cognition, psychology, linguistics, etc. That you can find a model which exactly describes the "language unit(s)-non-language unit(s)"-sequences as underlying normal utterances (= communication in natural language), is no argument that there is any relation to "reality" behind such a model assumption.
There is no scientific possibility "to show" = to prove (as Sonja Erlenkamp) writes " there is nothing  of a linguistic structure that the verb and the noun phrases actually share". Neither you can show (as Dan Slobin writes) that "there is no formal motivation to label any of these participants in the syntactic terms of grammatical relations, grammatical cases". You can only make that rationally valid in a special model of language with special axioms. We all have learned that terms like subject, direct object, indirect object, nominative, accusative, dative or ergative, absolutive, oblique etc. are model dependent and are not definable uniquely for all languages.
>From the standard of comparison, meaning, I would hold the assumption that sign languages do show (= can code) semantic roles.
As you may notice, I separate "language" and "linguistic" in order to separate the object area (= language as a phenomenon) from the model area (= linguistic models about the functioning of language). (In brackets: No speaker or signer has ever used a "linguistic unit"; they only use "language units") From that perspective, the phenomenon "location" or "movement direction", are language units (following my axiom that a language utterance consists of coding elements used for language in this moment). Which different forms of status they can get in different linguistic models, depends on the model, not on the language.
Sonja Erlenkamp and Dan Slobin talk about their linguistic models. They should not mix up them with language or "reality" (e.g. by saying: "the language phenomenon X IS a subject or an agreement").
Best Regards
Franz Dotter
University of Klagenfurt
Center for Sign Language and Deaf Communication
Funded by: Provincial government of Carinthia, Bundessozialamt Kaernten, European Social Fund
Head: Franz Dotter (hearing)
Collaborators: Elisabeth Bergmeister (deaf), Silke Bornholdt (deaf), Jennifer Dörrschuck (hearing), Katja Hablich (hearing), Christian Hausch (deaf), Marlene Hilzensauer (hearing), Petra Käfer (hearing), Klaudia Krammer (hearing), Christine Kulterer (hearing), Andrea Lackner (hearing), Anita Pirker (deaf), Andrea Skant (hearing), Nathalie Slavicek (hard of hearing), Natalie Unterberger (deaf)
Deaf server (in German):
Fax: ++43 (0)463 2700 2899
Phone: ++43 (0)463 2700 /2821 (Franz Dotter), /2822 (Andrea Skant), /2823 (Marlene Hilzensauer), /2824 (Klaudia Krammer), /2829 (Christine Kulterer)
Email addresses: firstname.lastname at

>>> "Dan I. Slobin" <slobin at BERKELEY.EDU> 9/20/2009 8:29 >>>
I agree entirely with Sonja Erlenkamp's response, on the basis of linguistic work on several sign languages.  The directionality of a sign (often along with gaze direction) encodes the relationship between the participants, whose identity has already been established by various means.  There is no formal motivation to label any of these participants in the syntactic terms of grammatical relations (subject, direct object, indirect object), grammatical cases (nominative, accusative, dative or ergative, absolutive, oblique), or semantic roles (agent, patient, recipient).  All of the necessary information for clause interpretation is present in the meanings of spatial locations, handshapes, and motion; and the interpretation itself does not seem to need anything but some set of semantic roles.  It follows that there is no such thing as “grammatical agreement” in these languages.  

I am presently working on a paper on sign languages and typology, in which I suggest that sign languages constitute a possibly unique linguistic type, with no alignment pattern at all.  This is because there are no arbitrary grammatical categories to align with formal markers­-no nominative/accusative or ergative/absolutive or active/stative or agent/patient, and no relevant formal markers.  In fact, alignment may be a peculiarity of the auditory modality, and not a necessary linguistic universal.  

Dan Slobin
Psychology & Linguistics
University of California, Berkeley

At 08:18 AM 9/20/2009, Sonja Erlenkamp wrote:

Well, I have to say, I don't agree with Ulrike on agreement (no pun intended). Most signed languages do - as far as I can tell - show no agreement patterns at all. There are constructions which are called "agreement" verbs by some researchers, but as for example Scott Liddell has shown for ASL (American Sign Language) in several of his publications from 1998 to this day, these constructions are not agreement patterns, since there is nothing  of a linguistic structure that the verb and the noun phrases actually share. They both make use of spatial locations to create reference to participants, but spatial locations are not morphemes in themselves. Signs can be placed at spatial locations, as well as verbs directed to, but space in itself cannot be a morpheme.  There are several other reasons why the application of the term "agreement" on this construction is misleading, but I won't go into detail on that one here. The number of researchers who agree with Liddell on this issue for different signed languages has been increasing since he started the debate.
My own research on Norwegian Sign Languages (and German Sign Language), shows that these two signed languages do use different markings of grammatical relations in different construction types withou any clear S/A or S/O correspondence. The directionality you mention (where movements are directed in space to mark relations between participants by means of prompting mental connections between spatial locations and referents) is only one type of construction, which is not even very frequent in signed language utterances due to the fact that the verbs involved are often ditransitive verbs. It seems as if this verb class consists mostly of verbs conceptualizing some kind of either concrete or metaphorical transfer, where the movement direction resembles the path of the transfer. In fact the directional movement in these verbs moves from the location related to the A towards the location related to the indirect "object", not the O. I have never seen a directional verb in the signed languages I have looked at that had a movement towards the O exclusively, but there are some verbs that can only be moved away from the spatial location related to the A, something which often involves another construction: surrogate blends. I wouldn't call that agreement though. 
All the best
Prof. Sonja Erlenkamp
University College of Sør-Trøndelag
Department of teacher - and sign language education
2004 Trondheim
> Date: Sun, 20 Sep 2009 10:23:57 +0100
> From: uzeshan at UCLAN.AC.UK
> Subject: Re: O-only agreement
> Hi, though this is only partially what you are looking for, the majority of sign languages have the following agreement patterns:
> - no agreement with intransitive S
> - agreement with both A and O for some transitive verbs
> - agreement with O only for some other transitive verbs
> Interestingly, agreement with A only in transitive verbs does not occur.
> Verb agreement with transitive verbs is also known as "directionality" in sign linguistics (due to the agreement being shown by the direction of the hand movement during production of the verb).
> Ulrike
> Prof. Ulrike Zeshan
> Director, International Centre for Sign Languages and Deaf Studies
> Faculty of Arts, Humanities and Social Sciences
> Livesey House, LH212
> University of Central Lancashire
> Preston PR12HE, UK
> uzeshan at
> Ph. +44-1772-893104
> >>> peterarkadiev <peterarkadiev at YANDEX.RU> 19/09/09 2:51 PM >>>
> Dear typologists,
> while Paul Hopper has come up with an example of A-only agreement in Malay (see references below), I ask a follow-up question concerning the mirror-image situation: are there any languages where the verb would agree exclusively with the transitive O (patient, undergoer, direct object), but neither with the transitive A nor with the intransitive S?
> Many thanks and best wishes,
> Peter Arkadiev
> Paul J. Hopper, 1987 Stability and change in VN/NV Alternating Languages:
> A study in pragmatics and linguistic typology. In M. Bertuccelli Papi and
> J.Verscheuren, eds., The Pragmatic Perspective, 455-476. Amsterdam: John
> Benjamins.
> Paul J. Hopper, 1983 Ergative, passive, and active in Malay narrative
> discourse. In F. Klein-Andreu, ed., Discourse Perspectives on Syntax,
> 64-87. New York: Academic Press.

Share your memories online with anyone you want anyone you want. 
Dan I. Slobin
Professor Emeritus of Psychology and Linguistics

Department of Psychology           email: slobin at
3210 Tolman #1650                    phone (Dept):  1-510-642-5292
University of California                phone (home): 1-510-848-1769
Berkeley, CA 94720-1650, USA   fax: 1-510-642-5293

More than messages–check out the rest of the Windows Live™.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Lingtyp mailing list