LINGUIST List 7.1035

Tue Jul 16 1996

Sum: Experiencer role

Editor for this issue: Ann Dizdar <>


  1. "David Harris", Experiencer Role

Message 1: Experiencer Role

Date: Sun, 14 Jul 1996 20:12:05 GMT
From: "David Harris" <>
Subject: Experiencer Role

For anyone that may have been looking for the replies to my question
about the experiencer role some time ago, here they are. I'm sorry it
has taken me so long to get them on the list. QN

Dear Quentin North,

I haven't touched this subject for years, but some of the references
in my 1988 LI article could stil be valid:
Brekke, Magnar, The Experiencer Constraint
Linguistic Inquiry vol 19 No 2, Spring 1988, pp 169-180.
Jackendoff and G.Horn are basic readings.

Hope this may be of some help.

Magnar Brekke

Look at "Notes on Psych-Verbs, Theta-Theory and Binding" by Adrian
Belleti and Luigi Rizzi in _Principles and Parameters in Comparative
Grammar_ edited by Robert Freidin, MIT Press, 1991 (Current Studies in
Linguistics No. 20).

Limbex Corporation
13160 Mindanao Way, Suite 234
Marina Del Rey, CA 90292 USA
(310) 309-4281 x4505 (office/vmail)
(310) 309 4282 (fax) (URL)

I am not an expert on the experiencer role per se, but I am something
of an expert on thematic roles generally, as I've been working in that
framework for machine translation applications for over 10 years now.

The set of "acknowledged thematic roles" is not all that well-defined,
and a literature search, while highly useful, is not likely to be

>Is the idea just that no change in state or location occurs in the
>observed element (theme)?

No. If anything, that would seem to characterize the PATIENT role,
but it may also be true of an EXPERIENCER under some circumstances.
For example, here's a sentence from the domain I currently work with
(heavy equipment):

"The pistons react to the high pressure oil."

This "reaction" might take the form of motion, and you might want to
call "piston" a THEME, but given the implications of "react" I would
call "piston" an EXPERIENCER. Or the piston might react by
disintegrating (in some presumed worst case :-) and that would be a
state change, but I would still call it an EXPERIENCER.

A more clear-cut set of examples of EXPERIENCER would be with humans
experiencing emotional states or perceptions, e.g. "I see the book"
(EXPERIENCER = "I") "The student feels confused" (EXPERIENCER =
"student") , etc.

Given the fuzzy nature of characterizations of thematic roles, it's
entirely possible somebody else will send you some examples that seem
to contradict these. I hope you will come out of your query attempt
at least knowing more than you did before.


- Marion Kee
- ---------------
Marion Kee | I don't speak for CMU,
Knowledge Engineer, Center for Machine Translation | and CMU returns the
Carnegie Mellon University Pittsburgh, PA, USA | All opinions are my own.

No doubt you are exclusively interested in the technicalities of theta
theory here, which is fair enough. But can I recommend you read
Michael Halliday on participant roles: Halliday MAK 1994 Introduction
to Functional Grammar (Arnold) Chapter 6 is fairly basic. Even more
basic is my own account in Bloor T & Bloor M 1995 The Functional
Analysis of English (Arnold) Chapter 5. I also think that Fillmore's
classic 1968 paper 'The case for case' still offers insights - in
fact, theta theory owes a seldom acknowledged debt to Fillmore. Best
Tom (Thomas Bloor) 

You're asking all kinds of semantic questions--sounds like you're
trying to learn in record time about a lot of the stuff I deal with in
terms of the informal ("naive") semantics of natural language.

You said:

>I was curious about how the element "experiencer" fits
>into the whole phrase, ie. what it is dominated by,

Generally speaking, it's dominated by the verb phrase (or INFL, in
GB-style X-bar theory.) This is a syntactic question. The VP (or
INFL) "licenses" a noun phrase (NP) to receive the thematic role
"experiencer". In most syntactic theories, it's assumed that in the
lexicon the verb is represented as "subcategorizing for" a certain
kind of NP to receive the "experiencer" thematic role that the verb is
specified as assigning. In most cases, the verb will assign this role
to the argument (NP) which fills the "subject" function with respect
to the verb. "subject" is usually defined in terms of syntactic
structure. In the case of verbs such as "feel" "see", etc., the NP
which receives the "experiencer" role has to represent something
animate (this is a semantically-based constraint.) There's a mix of
syntactic and semantic elements in this description.

>how the pair
>experiencer/theme compares structurally and otherwise with the pairs
>agent/theme or instrument/theme

That's a good question. In some ways, I don't think the kind of
"theme" that shows up in experiencer/theme is really the same kind of
"theme" that you find in agent/theme[/goal] constructions. The
experiencer role is a bit theme-like itself, because the (animate or
semi-animate) experiencer is, in a way, undergoing a change of state.
This is what I was talking about in my first email message to you,
where I pointed out that the verb "react" might well assign an
experiencer role to its subject, and that the NP which is the subject
might actually undergo some kind of movement or state change in
undergoing the process of "reacting". In the case of a human
experiencing a perception or emotion, I think there's an implied state
change in the human. But "theme" is a vague role. Characterizing it
in terms of a thing that moves or changes state doesn't capture it
perfectly. Another way to look at "theme" is to say that a theme is
something that gets "delivered" to somewhere. In that sense,
experiencer/theme is semantically more like agent/theme than it might
have appeared at first.

Structurally, agent/theme usually manifests as subject/direct object,
which is like experiencer/theme. Agent/theme can also take a "goal"
(optionally) which is an oblique argument, usually expressed in
English as a prepositional phrase, that tells you "where" the "theme"
"ended up". (Sorry about all these quotes; I don't want you to think
that I'm really using technical terms with invariant meanings here!)
Verbs of motion provide the classic examples: in "I sent the package
to Pheonix." "I" is the subject and the agent; "the package" is the
direct object and the theme; and "Phoenix" is an oblique locative
prepositional phrase and the goal. If you say "I feel confused.",
though, the destination of "confusion" is "I"--the experiencer. So
the parallel between agent/theme and experiencer/theme is not total;
they handle questions of "where" the "theme" "goes" in different ways.

I'm not entirely sure what you mean by an "instrument/theme" pair (do
you mean a case where they are both direct arguments of some verb? as
in "the golf club hits the ball"?) so can't hold forth on it. :-)

>and how the verb fits in with it all.

See above. The verb is usually assumed to be the core of the argument
structure, with various configurations of NP's related to it as
arguments. In a way, the verb is like the motor that makes the
sentence go.

Re: Adjectives:

>It finally occured to me that there may be some basic
>listing put together in a study years and years back that every
>linguist knew about and that it may just be a matter of asking on a
>list or news group with a language-related theme.

Well, I wish there were! For any given human language, it's possible
somebody has done this. For English, I suggest looking in Quirk,
Greenbaum, et al.'s _A Comprehensive Grammar of the English Language_.
I haven't researched adjectives in that work, but I think you might
find something there.

"adjective" is really a syntactic term; an "adjectival" thing is
something (a word or phrase) that modifies a "nominal" (noun-like)
thing. The questions you're asking about adjectives are semantic
ones: how may we characterize the meanings which are associated with
"adjectives?" Thus, "How may we characterize the meanings which are
used to modify the meanings of things usually expressed as nominals?"
You see it gets fuzzier as you abstract away from syntax--and syntax
can be amgibuous enough!

>I am also interested in discovering
>what structural differences there might be between the members of such

syntactic structure? It's likely to vary a bit across languages.
Also, just how far do you want to push the question of what counts as
an "adjective"? If I say "the red book", "red" is clearly an
adjective. If I say "the book with the torn cover", "torn" is an
adjective (or at least a past participle acting like an adjective),
but is the modifying PP "with the torn cover" an "adjective"? How
about a relative clause, e.g. "the book which I gave to George"?
(There are some languages where you can say the equivalent of "the
which-I-gave-to-George book".)

semantic structure? One hopes to discover universals, but that's
always an exceedingly tricky undertaking. I've been working on this
problem as it relates to the semantic classification of meanings which
are expressed via prepositional phrases in English, and after 3 years
I think I've made progress, but there is so much to be considered....

Anyway, once again, good luck and if a "magic listing" turns up
someplace, I'll be pretty thrilled.

- Marion Kee
- -------------------------
Marion Kee | I don't speak for CMU,
Knowledge Engineer, Center for Machine Translation | and CMU returns the
Carnegie Mellon University Pittsburgh, PA, USA | All opinions are my own.

Dear Silent Q,
 The Experiencer role is treated explicitly in case grammar systems
such as Fillmore, Chafe, and implicitly in Anderson and Jackendoff.
For a review of all these see Cook, Case Grammar Theory, 1989.
 Nonlocalists such as Fillmore and Chafe treat Experiencer as an
explicit role in a case system with a minimum of 5 cases: Agent,
Experiencer, Benefactive, Object (Theme) and Locative. The Experiencer
is the person undergoing sensation, emotion, cognition, communication,
and occurs in State verbs (like, want, know), Process verbs (enjoy),
including psych movement verbs as object (amuse, bore) and Action
verbs (teach) including communication (say) with E-case as hearer.
 Localists such as Anderson and Jackendoff usually deal with Agent,
Theme, Location, Source, and Goal. In these systems Experiencer is
treated as an abstract Location, Source, or Goal. In Fillmore know is
classed as E,Os; in Anderson as L,Os. According to Jackendoff
(1990:47) "Experiencer presumably is an argument of an as yet
unexplored State-function having to do with mental states."(See also
1990:140-141 with Experiencer as a kind of Patient). Hope this helps.

>Date: Thu, 11 Jan 1996 04:41:37 GMT
>From: ("David Harris")
>Subject: ?? on Experiencer Role
>Can anyone offer a brief explanation and/or point me to a fairly basic
>reading on the subject?
>Quentin North

Good question. Unfortunately, I have no help
to offer, since I've never quite understood
the experiencer role myself.

Would you please do me the favor of forwarding to me
any particularly illuminating examples and/or
explanations you might receive? Or if you're
planning on posting a summary of the responses to
the Linguist list, could you let me know that?

I'd greatly appreciate it.

Equally in the dark about experiencer roles,


;Kathleen M. O'Neill ... Language Laboratory Technician I ;
; ... ;
;University of Illinois at Chicago ... Language Laboratory ;
;703 South Morgan Street (M/C 042) ... Grant Hall, Room 311 ;
;Chicago, IL 60607-7025 ;
;312.996.8838 or 8836 ... 312.996.5501 FAX ;
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue