LINGUIST List 14.2341

Fri Sep 5 2003

Review: Morphology/Syntax: Barss (2002)

Editor for this issue: Naomi Ogasawara <naomilinguistlist.org>


What follows is a review or discussion note contributed to our Book Discussion Forum. We expect discussions to be informal and interactive; and the author of the book discussed is cordially invited to join in.

If you are interested in leading a book discussion, look for books announced on LINGUIST as "available for review." Then contact Simin Karimi at siminlinguistlist.org.


Directory

  • Miguel Rodriguez-Mondonedo, Anaphora: A Reference Guide

    Message 1: Anaphora: A Reference Guide

    Date: Thu, 04 Sep 2003 14:06:52 +0000
    From: Miguel Rodriguez-Mondonedo <Miguel.Rodriguez-MondonedoHUSKYMAIL.UCONN.EDU>
    Subject: Anaphora: A Reference Guide


    Barss, Andrew, ed. (2002) Anaphora: A Reference Guide. Blackwell ISBN: 0631211179, Pages: 320, Price: $71.95

    Announced at http://linguistlist.org/issues/14/14-899.html http://linguistlist.org/issues/13/13-1454.html

    Miguel Rodr�guez-Mondo�edo, Department of Linguistics, University of Connecticut

    This book presents the cutting edge of anaphoric relations' studies, showing its syntactic and semantic properties, and also pointing out philosophical, psycholinguistic, and cognitive implications of our comprehension of anaphors. Eight outstanding articles not only draw a remarkable big picture of the issue, but also drive the reader deeply inside its most intriguing details.

    First I will summarize each article, and then I will make some brief comments about each one of them.

    The first one is ''Timing Puzzles in Anaphora and Interpretation,'' by Andrew Barss. He develops a model of the syntactic-semantics mapping without Logical Form (LF) as level of representation, according to which the semantic representation of a sentence is built up in parallel with the syntactic derivation, under the assumption that anaphoric dependences are formal constructs (akin chains) formed in the course of a derivation. Two principles rule this process:

    (i) Earliness of Semantic Interpretation: Compute the semantic interpretation of elements of a phrase marker as early as possible in a derivation (a version of Pesetsky's (1989) Earliness Principle)

    (ii) An anaphoric dependency D (A, B) cannot be formed (or filtered) until both A and B are morphologically licensed.

    In this case, ''as early as possible'' means as soon all the ingredients of the dependency are merged in the tree, and ''morphologically licensed'' includes lexical projection and feature checking. The author tests the empirical power of this model explaining anaphoric reconstructions effects related with weak crossover, adverbial scope, events and negation scope, and antecedent- contained deletion; and also deriving strict cyclicity for wh-movement.

    ''Two Types of Scrambling Constructions in Japanese,'' by Ayumi Ueyama, analyzes Object-Subject-type (OS-type) constructions in Japanese, which are considered cases of scrambling (as opposed to unmarked SO constructions) in this language. Two kinds of OS-type constructions are distinguished in the literature: Deep and Surface. Deep OS-type does not show Weak Crossover (WCO), and its Object shows wide-scope reading over its Subject; whereas Surface OS-type preserves WCO, and its Object shows narrow-scope reading over its Subject, in addition to exhibit reconstruction effects. The author adds to this characterization that long distance OS-type is necessarily a Surface type, and that in multiple OS constructions at least one instance is a Surface type (preferable, the second one). According to Ueyama, these last observations pose a serious challenge to Saito 1992 (among others), who assumes that a Deep OS-type is available only if the A-chain reanalysis is possible, and that the Deep OS-type undergoes movement. The author shows that Deep OS-type is base-generated, with an empty operator (Op) movement, whereas Surface OS-type involves a PF (Phonetic Form) movement of O: [''-k'' is my index]

    (i) Deep OS-type: O [Op-k [S t-k] (ii) Surface OS-type: O-k S t-k

    Since OS constructions show Subjacency effects, Ueyama assumes that PF is relevant for Subjacency.

    ''The Psycholinguistics of Anaphora,'' by Janet Nicol and David Swinney, present the results of several experiments testing the processing of coreference during the comprehension. The general idea is that the appearance of an anaphoric element (a proform) triggers the search for antecedent and its selection among previously mentioned Nominal Phrases (the candidate set). Research questions include constrains over the candidate set, the selection process and the timing of pronoun resolution. The experiments show different results regarding this last issue: according to ones, the resolution begins immediately after a proform is presented; according to others, this happens during the subsequent words. The authors explain this discrepancy remarking that the experiments used different methods and materials; for instance, pronouns in different syntactic positions (subject/object, argument/adjunct), or different channels (auditory/visual). Their conclusion is that with auditory data the proform triggers the search immediately; the candidate set is constrained by Binding Conditions (Chomsky 1981) and for morphological, semantic and pragmatic conditions (such as agreement, topic or knowledge about the world); that means that not all NPs are initially considered as candidates. However, this is not so clear in the case of visual presentation, which seems to make coreference more difficult to compute.

    ''Two Pronominal Mysteries in the Acquisition of Binding and Control,'' by Dana McDaniel, analyzes children's failure to comply with Principle B (''PB lag''), and children's use of internal reference for pronouns, as if they were non-arbitrary PRO (Although previous studies consistently report PB lag, they do find some children that obey Principle B; in addition, children comply the condition when the antecedent is a quantifier (''Every girl is patting her''), and when they use spontaneous speech; and children who speak language with clitics (Italian or Spanish) obey Principle B. These facts are not well explained by other accounts for the phenomenon. McDaniel's solution is that children do not have sensitivity to emphatic stress; therefore ''Groves is patting him (non emphatic)'' can be interpreted as ''Grover is patting HIM (emphatic)''. Since contra-indexed Nominal Phrases can be coreferential when the pronoun is emphatic, children could interpret any NP as coreferential before mastering emphatic stress. The second topic is the Pronoun Coreference Requirement (PCR), according to which the pronoun in ''Grover kisses Big Bird before he jumps over the fence'' is interpreted in the same manner to PRO in ''Grover kisses Big Bird before PRO jumping over the fence''. No conclusive solution is provided, however McDaniel links the PCR to switch reference systems, suggesting that these could contain a grammaticized form of PCR.

    ''Reference Transfers and the Giorgione Problem,'' by Mario Montalbetti, address a problem that can be traced back to the twelfth century and whose modern form was expressed by Quine (1953) in sentences like (i-ii):

    (1) Giorgione was so-called because of his size (2) Barbarelli was so-called because of his size

    Assuming that (1) is true, and knowing that Giorgione=Barbarelli, (2) is false, and (1) means (3):

    (3) Giorgione-k was called Giorgione because of his-k size

    The problem is not only the failure of substitutivity in (1-2). In (1), whereas ''so'' refers to the name, ''his'' refers to the individual; furthermore, ''Giorgione'' refers to the individual. Then, we have an unusual situation: an anaphoric expression refers to something different than its antecedent. It is not possible to exclude Syntax from this problem (leaving Pragmatics alone), because it interacts with syntactic phenomena, for instance, only the sloppy-identity reading of (4) produces a failure of substitutivity:

    (4) Giorgione was so-called because of his size and Barbarelli was too

    Montalbetti's solution is that the predicate ''called'' coerces a shifted antecedent (''coercion'' is a notion from Pustejovsky 1995), from the individual to his name, that is picked up by the anaphoric element ''so''; ''his'' (a pronoun) is free to refer back to the unshifted antecedent. He suggests that his solution can be applied to other similar cases of reference transfer.

    ''Tense and Anaphora: Is There a Tense-Specific Theory of Coreference?,'' by Karen Zagona, argues against a tense-specific theory of tense licensing---like Tense Anchoring (En� 1997)---in favor of the Temporal Argument Structure hypothesis (TAS), according to which Tense Phrase (TP) is a transitive predicate with the Speech-Time (ST) as external argument and the Event-Time (ET) as internal argument (identified with the Verbal Phrase (VP) ). The visibility of ST and ET is ensured by a mood feature [+Indicative] and a [+/- Past] feature, respectively. These are Case-like features, purely formal; therefore we cannot derive temporal interpretations from them. Zagona proposes to derive temporal orders from the position of the arguments, considering the higher as more recent. Future tenses seem an objection because ET (the lower argument) is more recent than ST; therefore, in future tenses, ET must raise to a higher Phrase (maybe a Modal Phrase, headed by ''will''). Traditionally, Present tense is analyzed as the simultaneity of ET and ST, which also contradicts Zagona's view; she proposes that in present-moment interpretations ST (an specific time) is included in the times provided by VP (with ''mass'' interpretation); in this view, TP is ''the set of times such that VP'' (p. 154). According to Zagona, temporal dependencies between clauses cannot be understood as dependencies between tenses, since they hold even if there is not a c-commanding tense (as in clausal complements of nouns). Temporal dependences are the result of interactions between argumental structure, aspect and sub-event structure.

    ''Surface and Deep Anaphora, Sloppy Identity, and Experiments in Syntax,'' by Hajime Hoji, presents a revision of a classic problem posed by Hankamer and Sag (1976), in relation with sloppy/strict identity readings. A surface anaphora (SA, 5b) needs a linguistics antecedent, while deep anaphora (DA, 5c) can be licensed pragmatically:

    (5) a. You will be able to stuff this ball through this hoop b. You will be able to [VP ] : SA- VP Ellipsis (VPE) c. You will be able to [VP do it] : DA

    As noted by Dalrymple (1991), both SA and DA can lead to sloppy/strict identity ambiguity (SSIA), which modified the standard assumption that only SA can do that:

    (6) a. John washed his car on that rainy day. b. Bill did THE SAME THING. [DA]

    (6b) can mean (i) Bill washed John's car on that rainy day (strict) or (ii) Bill washed his own car on that rainy day (sloppy). Dalrymple thinks that this fact challenges the idea that we can explain SSIA using the difference between their syntactic representations. However, following Lasnik 1976 and Reinhart 1983, Hoji notes that the sloppy identity in SA has the same constrains than bound variable anaphora (BVA); BVA (and therefore, sloppy SA) requires what Hoji calls Formal Dependency (FD). We have FD (A, B) if (i) B is [+beta] (in terms of Fiengo and May 1994), (ii) A c-commands B, and (iii) A is not in the local domain of B. No coindexation is required. Hoji presents a set of tests in English and Japanese where he modifies the requirements for FD and finds that this causes the loss of sloppy identity reading in SA but not in DA. The conclusion is that DA does not need a FD for its sloppy reading. In other words, (5b) has the same LF representation than (5a), but (5c) has not. Sloppy readings in SA have a syntactic explanation.

    ''The logic of Reflexivity and Reciprocity,'' by Terence Langendoen and J�el Magloire, presents an account for the logical properties of reflexive (REF) and reciprocal (REC) sentences. REC/REF's properties result from (i) the interaction between plural properties of predicates (Goodman 1951); (ii) the core meaning of REC/REF; and (iii) the way REC/REF is expressed: overtly (with anaphora, as in (7)) or covertly (without anaphora, as in (8)). In (7) the predicate has two places, whereas in (8) it has only one place (the REF/REC property is incorporated into the predicate).

    (7) a. Ana and Bob are in love with themselves [REF/REC] b. Ana and Bob are looking at each other [REC]

    (8) a. Ana and Bob are shaving [REF] b. Ana and Bob disagree [REC]

    Langendoen and Magloire explore all the plural properties of REC/REF in one- place and two-places predicates, extending Goodman (1951)'s original insight. They also deal with reciproreflexive two-place predicates, where REC/REF distinction is neutralized, as in Spanish sentences with ''se'' (9); and cases of hyporecyprocality (Fiengo and Lasnik 1973), that exhibit a logical or pragmatic asymmetry (10)

    (9) Ana y Pepe se quieren [Ana and Pepe like themselves/each other]

    (10) The plates are stacked on the top of each other

    Langendoen and Magloire establish the entailments among REC/REF sentences, and propose a criterion for reflexivity and for reciprocity (the ''core'' meaning of REC/REF).

    COMMENTS

    As a whole, an outstanding property of this book is that it shows how deeply the study of anaphoric relations is connected with almost every subfield of Linguistics. They are relevant not only for Syntax, Semantics or Pragmatics, but also for Language Acquisition, Intonation, and Cognitive Sciences. Additionally, it evidences the extraordinary vitality of the issue inside Generative Grammar. Also, each article is outstanding by its own. I will comment very briefly each one of them.

    Barss' article presents a convincing way to do Syntax and Semantics at the same time. This allows the system to get rid of indexes (following Chomsky 1995's advice). In addition to empirical success, he is able to enforce cyclicity for wh-movement (what makes the system very appealing). It is not clear, however, until which extent his Earliness of Semantic Interpretation creates different levels of representation. If it does, it will be no different than an old condition on Deep or Surface Structure. Although he tries to avoid that possibility by comparing his system with chain formation, the issue remains unclear.

    Ueyama's article about Scrambling sets the stage for further inquiry on this intriguing phenomenon of Japanese (and other languages). It also fuels the discussion about A/A-bar dependencies, and presents a rigorous method to use anaphoric dependencies to uncover syntactic structures.

    Nicol & Swinney look anaphora from the point of view of the hearer, and their findings confirm the complex nature of coreference, which is sensitive even to the material channel of communication. This could be an experimental confound, especially in comprehension tests about anaphora.

    In the same line, McDaniel's article do find a confound in previous experimental tests about acquisition of anaphora, that were insensitive to emphatic stress of the pronoun. This opens an important line of reasoning and calls for the revision of other tests with similar properties.

    Montalbetti's article shows an intricate connection between lexical semantics and syntax, regarding interpretation of anaphors, which seem to be sensitive to some lexical shifts produced during the syntactic process. A shifted antecedent makes its unshifted version unavailable for anaphors but not for pronouns: this is a remarkable finding that poses new questions about the nature of the lexicon-syntax interplay (being the simplest one, why only anaphors do that)---- Montalbetti, however, does not address this kind of questions.

    Zagona's article successfully eliminates temporal anaphors as explanation for temporal orders and dependencies. But she makes a stronger claim: that each clause (main or embedded) is autonomous from the point of view of tense. It seems that some phenomena, considered ''sequence of tenses'' for a long time, turn out to be aspectual in nature. If true, this claims for a huge revision of most of the literature on the issue.

    Hoji's article about deep and surface anaphora is the one that makes the hardest effort to link his findings with conceptual and cognitive process, as well as with some assumptions in the generative view of language. He succeeds in showing that the sloppy reading in surface anaphora is obtained purely on the basis of linguistic resources, whereas in deep anaphora that reading must relay on other cognitive faculties. This is a strong case in favor of the so-called ''autonomy of language.''

    After reading Langendoen & Magloire's article, you will be so amazed by the detailed machinery they display to catch all the logical possibilities of reflexivity and reciprocity, that you will come to the conclusion that some times is better to see the trees instead of the forest. However, they have managed to do both. A logical definition of reflexives and reciprocals is provided, valid across a huge (and ordered) set of cases.

    As final remark, I must say that it has been a great pleasure to read this wonderful book. I have no doubts that it will be a starting point for future research.

    REFERENCES

    Chomsky, Noam. 1995. The minimalist program: Cambridge, Mass.: MIT Press.

    Chomsky, Noam. Lectures on Government and Binding, Studies in Generative Grammar ; 9. Dordrecht, the Netherlands: Foris Publications, 1981.

    Dalrymple, Mary. 1991. Against reconstruction in ellipsis. Xerox Technical Report, Xerox-Park, Palo Alto, CA.

    En�, M. 1987. Anchoring conditions for tense. Linguistic Inquiry 18:633-657.

    Fiengo, R., and Lasnik, Howard. 1973. The logical structure of reciprocal sentences in English. Foundations of Language 9:447-468.

    Fiengo, R., and May, R. 1994. Indices and Indentity. Cambridge, MA: MIT Press.

    Goodman, Nelson. 1951. The Structure of Appearance. Cambridge, MA: Harvard University Press.

    Hankamer, J., and Sag, I. 1976. Deep and surface anaphora. Linguistic Inquiry 7:391-428.

    Lasnik, Howard. 1976. Remarks on coreference. Linguistic Inquiry 2:1-22.

    Pesetsky, D. (1989). The earliness principle, Ms. MIT.

    Pustejovsky, J. 1995. The Generative Lexicon. Cambridge, MA: MIT Press.

    Quine, W.V.O. 1953. Reference and modality. In W.V.O. Quine From a logical point of view, ed. Cambridge, MA: Harvard University Press.

    Reinhart, T. 1983. Anaphora and Semantic Interpretation. Chicago: University of Chicago Press.

    Saito, M. ''Long Distance Scrambling in Japanese.'' Journal of East Asian Linguistics I (1992): 69-118.

    ABOUT THE REVIEWER

    Miguel Rodr�guez-Mondo�edo is a PhD student in the Department of Linguistics, in the University of Connecticut. He has done research in Binding Theory (in particular, Romance obviation), but he also has strong interest in other aspects of Syntax (like DP structure and clitic-doubling), Morphology (nominalizations) and Philosophy of Language. His web page is http://rodriguezmondonedo.virtualave.net