LINGUIST List 6.1784

Sat Dec 23 1995

Sum: Self-/Centre-Embedding

Editor for this issue: Ljuba Veselinova <lveselinemunix.emich.edu>

Directory

1. Richard Hudson, Sum: self-/centre-embedding

Testing nesting: A bibliography on the effect of self-embedding and
centre- embedding on ease of processing.

This bibliography reflects the joint efforts of various members of the
Linguist network (listed below) and arises out of a query that I
posted in early December 95.

CONTRIBUTORS:
Sue Blackwell
Annabel Cormack
Jennifer Ganger
Ted Gibson
Chris Golston

MY QUERY!Steve Harlow
Caroline Liberg
John Limber
Bruce Nevin
!Neal Pearlmutter
Colin Phillips
Karin Stromswold
David Wharton.
I asked for information about *empirical* work on centre-embedding
examples like (1).

(1) The dog the stick the fire burned beat bit the cat.
 N1 N2 N3 V3 V2 V1

TERMINOLOGY

I deliberately used the term `centre-embedding' as the term which
other people have applied, but I should of course have called it
`self-embedding'. In the following I shall try to distinguish the
two, as CE for `centre- embedded sentence' and SE for `self-embedded
sentence', though I don't actually think either concept is at all
clear. Some people use `nesting' instead of `centre-embedding'. The
clearest examples of unprocessable SEs are of one very specific type,
like (1): object-relative inside object-relative modifying the first
noun, i.e. abstractly `N1 N2 N3 V3 V2 V1'.

CHOICE EXAMPLES

Though I asked for publications, some people very kindly supplied
relevant raw data as well:

 English: "The claim that the link between convection heating and the
time and energy which can be saved by baking biscuits in a convection
oven rather than a conventional oven is not obvious at first sight is
undoubtedly true." (Annabel Cormack)

 Classical Greek: "Ancient Greek did possessor constructions with
 center-embedding as the unmarked case:
 [teen [tou prosoopou] phusin]
 the the face nature
 'the nature of the face' (Plato, Politicus 257d)

 This happened even with two possessors (!!):

 [to [tees [tou ksainontos] tekhnees] ergon]
 the the the wool-carder art work
 'the work of the art of the wool-carder' (Plato, Politicus 281a)

 [ta [tees [toon polloon] psykhees] ommata]
 the the the many souls eyes
 'the eys of the souls of the many' (Plato, Sophist 254a)

 The data is to appear in my article `Syntax outranks phonology'
 (Phonology 12.3), which, however, has little to do with
 center-embedding. The data given here are just cannon-fodder in a
 paper on the syntax-phonology interface." (Chris Golston)


BIBLIOGRAPHY

This is presented chronologically, in the hope of giving some
impression of the lean years of the 70s and 80s. I haven't seen all
these references yet, so the comments should all be taken with a pinch
of salt.

Miller, G. & Isard, S. 1964. Some perceptual consequences of
linguistic rules. Jnl of Verbal Learning & Verbal Behavior 2: 217-28.
 Ss tried to repeat SEs which they heard. Affected by degree of embedding.
 Also quotes evidence that problems set in with the stacked verbs, shown
 by eye-movements.

Blumenthal, A. 1966. Observations with self-embedded sentences.
Psychonomic Science 6, 453-4.

 According to Gibson (1991:171): people paraphrase SEs as though the
 nesting was coordination.

Fodor, J. & Garrett, T. 1967. Some syntactic determinants of
sentential complexity. Perception and Psychophysics 2, 289-96.
 The presence of a relative pronoun made the SEs easier to paraphrase.

Blumenthal, A. 1967. Prompted recall of sentences. Jnl of Verbal
Learning and Verbal Behavior 6, 674-6.
 According to Bever (1970: 295) this tells much the same story as
 Blumenthal 1966.

Schlesinger, I. M. 1968. Sentence Structure and the Reading Process.
Mouton.
 Various experiments in which Ss read a SE and judged its
grammaticality, paraphrased it, or repeated it. Results showed that
embedding as such causes difficulty *only* where the meaning is
unhelpful, suggesting that we may not use a push-down store to keep
track of word order, but just follow meaning in linking verbs and
nouns.

Fodor, J.; Garrett, M. & Bever, T. 1968. Some syntactic determinants
of sentential complexity, II: Verb structure. Perception and
Psycho-physics 3, 453-61.
 CEs are harder with SEE or LIKE as their embedded verb than with
 e.g. HIT or SLAP, because of the ambiguous subcategorisation.

Bever, T. G. 1970. The cognitive basis for linguistic structures. In J
R Hayes (ed) Cognition and the Development of Language. Wiley,
286-341.
 General survey, with some discussion of SEs. 338: Maybe (2) is more
 comprehensible than (3)?
 (2) The dog the destruction the wild fox produced was scaring will run
 away fast.
 (3) The dog the cat the fox was chasing was scratching was yelping.

Labov, W. 1973. The place of linguistic research in American
society. In E. Hamp (ed.) Themes in Linguistics: the 1970s. Mouton.
 Quoted (p. 101-2) in De Roeck et al (1982) - see below - as giving
 evidence of a `large-scale' experiment showing the acceptability of
 multiple CEs.

Nevin, B. c. 1975. Unpublished research.
 "If I provided context in which the constituent propositions are
separately stated antecedents, as in a story, then some naive
listeners understood such sentences [as (1)] at least some of the
time. (Some listeners seemed to balk on them no matter what.)

 <Prior story context omitted>
 Then the fire began to burn the stick.
 The stick jumped out of the fire and beat the dog.
 The dog turned and bit the cat.

 The dog the stick the fire burned beat bit the cat.

 My difficulty was how to produce the twice-embedded sentence with
 appropriate intonation. And this appears to me to provide a simple
 explanation for the difficulty understanding it.
 [.... examples omitted]
 One generally singles out one thing at a time--that is indeed the
 nature of the act of singling something out.

 The fact that some [SEs] are easier to understand, and the things
 that make them easier to understand, put that presumption [that the
 syntax is unprocessable] in serious question."

De Roeck, A; Johnson, R; King, M.; Rosner, M.; Sampson, G. & Varile,
N. 1982. A myth about center-embedding. Lingua 58, 327-40.
 A wonderful collection of attested examples of incredibly complex
 (and unreadable) syntax, full of centre-embedding - but no examples of
 object-relative inside object-relative.

Frazier, L. 1985. Syntactic complexity. In D. Dowty; L. Karttunen & A.
Zwicky (eds.) Natural Language Processing: .... CUP
 Quoted in Gibson (1991:169): People accept sentences of the form `N1
 N2 N3 V3 V', although one V is missing. She concludes that the missing
 V must be V1. [Why not V2?]

Bach, E.; Brown, C. & Marslen-Wilson, W. 1986. Crossed and nested
dependencies in German and Dutch: a psycholinguistic study. Language
and Cognitive Processes 1, 249-62.
 German nested dependencies (= CE) are harder to process than Dutch
 serial dependencies.

Gibson, E. 1991. A Computational Theory of Human Linguistic
Processing: Memory limitations and processing breakdown. Carnegie
Mellon PhD.
 p. 169 quotes a pilot experiment by Howard Kurtzman which showed that
 our understanding of SEs crashes on V2.

Thomas, J. 199?. [title not known]. MIT Masters thesis.
 Three studies on English directed by Ted Gibson.

Gibson, E.; Thomas, J. & Babyonyshev, M. 1995. Processing center-
embedded and self-embedded structures in English and Japanese. Handout
for NELS presentation, Oct 30 95.

Babyonyshev, M. & Gibson, E. 1995. Processing overload in Japanese. MIT
Working Papers in Linguistics.
 "Each of these studies has a number of conditions testing a bunch of
 possible theories. We have found some quite interesting things:
 different in a number of ways from what was predicted by every theory
 that I know (including what was in my thesis)." (Ted Gibson)

Stromswold, K; Caplan, D; Alpert, N & Rausch, S. in press. Localisation of
syntactic comprehension useing PET. Brain and Language.
 "I've done some psycholinguistic and PET studies investigating the
 processing of single center-embedded sentences. The upshot of these
 studies are that subjects find even single center embedded sentences harder
 to process than right-branching sentences." NB CE, not SE, but still
 relevant! Stromswold is karinruccs.rutgers.edu. The abstract follows:

 Positron Emission Tomography (PET) was used to determine regional
 cerebral blood flow (rCBF) when 8 normal right-handed males read and
 made acceptability judgments about sentences. rCBF was greater in
 Broca's area (particularly in the pars opercularis) when subjects
 judged the semantic plausibility of syntactically more-complex
 sentences as compared to syntactically less-complex sentences. rCBF
 was greater in left perisylvian language areas when subjects had to
 decide whether sentences were semantically plausible than when
 subjects had to decide whether syntactically identical sentences
 contained a nonsense word. The results of this experiment suggest
 that overall sentence processing occurs in regions of left perisylvian
 association cortex. The results also provide evidence that one
 particular aspect of sentence processing (the process that corresponds
 to the greater difficulty of comprehending center-embedded than
 right-branching relative clause sentences) is centered in the pars
 opercularis of Broca's area. Although it is impossible to ascertain
 with certainty what this process is, it is likely to be related to the
 greater memory load associated with processing center-embedded
 sentences.


Prof Richard Hudson Tel: +44 171 387 7050 ext 3152
 E-mail: r.hudsonling.ucl.ac.uk
Dept. of Phonetics and Linguistics Tel: +44 171 380 7172
 Fax: +44 171 383 4108
UCL
Gower Street
London WC1E 6BT
UK