Review of Case, Argument Structure, and Word Order
|AUTHOR: Shigeru Miyagawa
TITLE: Case, Argument Structure, and Word Order
SERIES TITLE: Routledge Leading Linguists
Stella Markantonatou, Institute for Language and Speech Processing, Research Center “Athena”, Athens, Greece
The book is a monograph comprising ten chapters that have been written or reworked for the book’s purposes. Five challenging topics are discussed, each one with a pair of chapters. The book can be read as an ongoing debate because the second chapter of each pair always presents work that takes into account the discussion on the topic as well as findings and developments of linguistic theory that have occurred since the publication of the first chapter.
Japanese provides most of the empirical material and it is compared with a number of languages (Dagur, English, Greek, Slavic, Turkish among others) in order to discuss important theoretical issues of universal validity. The book is organized as follows:
1. Chapters 1 & 2: Stranding of nominal quantifiers: empirical evidence for a-movement and for the hypothesis of a predicate internal subject position
2. Chapters 3 & 4 and Appendix: The ditransitive construction: universal validity of the dual goal analysis of ditransitive verbs; movement is feature driven
3. Chapters 5 & 6: Nominative-genitive alternation (in subject and object position): phase domains
4. Chapters 7 & 8: Causative verbs: syntax versus lexicon
5. Chapters 9 & 10: Evolution of accusative case marking in Japanese: the relation between universal grammar and language change.
The Introduction is very helpful both for the creation of the right expectations about what follows and for a quick summary.
Chapters 1 & 2
The study of the environments where Japanese nominal quantifiers (NQs) can be separated from their antecedents lends support to two central claims of the Minimalist Program: (i) direct passives and unaccusatives involve A-movement (from object to subject position) and, (ii) the existence of a predicate internal subject position (movement from predicate internal subject position to Spec,TP position).
Japanese NQs are formed by combining a numeral with a classifier that identifies the semantic class of the modified noun. Two classifiers which illustrate the phenomenon are ‘nin’: human beings, and ‘ko’: small objects of roughly equivalent extension in all three dimensions, such as fruits, candies and stones; also coins [general inanimate classifier, for some speakers].
Japanese NQs are normally adjacent to the modified noun phrase (NP) (“the NQ’s antecedent” in the author’s terminology) but, in certain environments they can be separated from it. The author argues that the patterns of NQ-NP occurrences are governed by the “mutual c-command condition” that dictates that ‘The NQ and the modified NP (or its trace) mutually c-command each other.’
The mutual c-command condition combined with a movement-out-of-the-VP-and-trace analysis explains the differences in the behavior of Japanese direct and indirect passives and of unergative and ergative verbs as regards NQs. The direct passive construction allows modification of the subject by an NQ that is within the verb phrase (VP) because the NP has moved (from object to subject position) and the NQ stands in a mutual c-command relation with the trace left behind (in the VP). No movement is involved in the indirect passive construction where an extra argument (the ‘experiencer’) is inserted. This explanation offers evidence for an A-movement analysis of passive voice and of unaccusative constructions.
Furthermore, a telic aspectual interpretation of a sentence enables NQ stranding with unergative intransitive predicates while a non telic interpretation does not. Environments that induce a telic interpretation of activities have been argued to affect the subcategorisation properties of activity predicates in a number of languages including English and Japanese (Levin & Rappaport Hovav 1995, Tsujimura 1989). The author introduces the “Telicity and the external argument (TEA)” hypothesis that accounts for (1) and (2) without a change of the subcategorisation properties of the verbs: ‘Once the external argument moves to Spec,TP, its lower copy in the predicate internal subject position is visible under telic interpretation.’ TEA is a stipulation. However, if it is combined with a grammaticalisation of aspect in the vein of Borer 2005, it explains (i) inverse scope phenomena that are observed with telic phrases, and (ii) why telic verb predicates of Japanese do not support resultatives (on the assumption that if a subcategorisation change had occurred, resultatives would be possible in principle).
Chapters 3 & 4 and the Appendix to Chapter 4
The author challenges the established view that the Japanese ditransitive construction is fundamentally different from the English one. He proposes that the two languages share an important similarity, namely that there are two dative positions, a high and a low one: the double-object-construction (DOC) structure chooses the high position (the ‘high goal’) while the ‘to’ dative chooses the low one (the ‘low goal’). He brings evidence that in Japanese the high and the low goal can occur in the same sentence. He builds on Marantz 1993, who argues that the English DOC involves a phonologically null applicative head V that explains the causation and possession effects observed with the DOC (‘John gave Mary a book’) but not with the ‘to’ dative structure (‘John gave a book to Mary’).
Two word orders, theme-low goal & and low-goal theme (the dual base analysis) are established as base-generated drawing on idioms, -kata nominalization data and binding phenomena.
Lastly, the theme cannot scramble across the high goal. A two-fold explanation is offered: (i) the applicative head is not associated with an EPP feature, therefore it cannot attract any constituent, including the theme -- this explanation is supported by Greek data (Anagnostopoulou 2003), and (ii) a locality constraint. The conclusion is that movement is not entirely free but rather feature-driven.
Two challenges for the dual base analysis are discussed, drawing on recent research:
1. Experimental evidence shows that reaction times for the -ni-o (DAT-ACC) structure are shorter than for the -o-ni (ACC-DAT) one: the author explains that native speakers of Japanese overwhelmingly expect the verb to follow the accusative NP-o, an expectation that is not borne out by an NP-ni.
2. The ACC-DAT base order is reserved for idioms only: the author brings evidence that the DAT-ACC order can be observed in non-idiom data.
Chapters 5 & 6
Contra to a C-licensing approach, the author argues that the genitive subject of temporal clauses is licensed by a weak v in combination with dependent tense, while in relative clauses (RCs) it is D(eterminer)-licensed. In this framework, he argues that phase is defined by case rather than by uninterpretable agreement features.
An RC with a genitive subject is a defective Tense Phrase (TP) unable to assign nominative case, therefore D’ ‘sees through it’ and assigns genitive case. Evidence for a defective TP is brought by binding and movement facts. Evidence that RCs are TPs is offered by adverbial distribution: according to Cinque 1999, adverbial distribution distinguishes the CP region (adverbials such as ‘honestly’, ‘unfortunately’, ‘evidently’) from the lower regions (adverbials such as ‘probably’).
Next the author argues that the genitive subjects of argument temporal clauses are licensed by a weak v in combination with dependent tense (the ‘Genitive of dependent tense (GDT) constraint’). Dependent tense is a phenomenon that occurs with subordinate clauses where the semantic content of tense in the subordinate clause is determined “in relation to structurally higher tenses” (p. 158). He suggests that this genitive is similar to the Slavic genitive of negation that occurs on internal arguments only. Support for the GDT is offered by scope phenomena and the genitive objects of (Japanese) transitive stative predicates.
Chapters 7 & 8
The locus of -sase causative formation, i.e. lexicon or syntax, and the phenomenon of blocking are the main issues.
Earlier work: In Japanese, there are two causative morphemes that attach to a verb stem in the lexicon giving formations of the type V-sas and V-sase/V-sas where the second -sas is considered by the author to be an allomorph of -sase and not identical to the first -sas. He explains the phenomenon of blocking in Japanese causative verb formation as follows: Every verb that receives an entry in the permanent lexicon must enter a slot in what the author terms ‘Paradigmatic Structure (PDS)’ that has three slots, Intransitive, Transitive and Ditransitive. Each slot may accommodate only one verb. A verb stem, being the morphologically simplest form, automatically enters a PDS slot. A derivative verb is blocked from the PDS if a simple verb stem already occupies the corresponding slot. Crucially, such a derivative is not blocked from the language. A V-sase enters PDS unless blocked by a simple verb or by V-sas as shown by nominalization, idioms and adversity causatives data.
V-sase behaves both as a word and a complex structure (it allows for an ambiguous interpretation of ‘zibun’) a fact that undermines the hypothesis that causatives are formed in the lexicon. In response, the author assumes that V-sase is always associated with two subcategorization features, one with only NPs, the other having an S in addition. Both these features are projected into the syntax, one resulting in a simplex structure, the other in a complex one. Evidence for the complex structure of -sase is provided by honorification data.
Recent work: The author abandons the idea that causatives are formed in the lexicon and adopts the One-Component Hypothesis in which causatives (of all types) are formed in the syntax. He draws on work on Distributed Morphology (Halle & Marantz 1993). The overall approach is the following:
1. -sase is the default causative morpheme (an ‘elsewhere causative’): if there is no specific instruction for pronouncing CAUSE, -sase is inserted
2. -sase may be overridden by special formations (that is, when specific pronunciation instructions exist)
This approach explains the existence of irregular causatives, the phenomenon of blocking and the dual syntactic behaviour of V-sase constructions.
Chapters 9 & 10
The author examines the evolution of the assignment of the Japanese accusative case marking -o, argues that -o is a D-structure case marker, and concludes that language change is controlled by universal grammar principles.
He compares textual evidence covering a span from the 8th century to the 16th century. He concludes that in Old Japanese (OJ), the conclusive form of the verb assigned abstract case while the case assigning feature of the attributive form had to be manifested overtly as -o or by head incorporation with simple nouns. Adjacency was required for abstract case assignment while the overt case marker -o relaxed requirements on strict word order.
The conclusive form was gradually abandoned and the attributive form took over and imposed its case assignment requirements. This language change was mainly motivated by the widely used ‘kakarimusubi rule’ that converts a matrix conclusive form to an attributive one. The overall change fits well with Case Theory and the principles of Universal Grammar.
Thus, Japanese went from abstract case marking to overt case marking as regards accusative case. English followed the opposite path: the rich Germanic case marking system developed into today’s English that lacks morphological case marking almost completely.
An important merit of the book is that it is an excellent piece of scientific writing, and in addition the author presents the arguments very clearly, as if he were teaching the material. The presentation of the topics as a debate that follows transformational theory through the years makes the book an excellent source of information on the evolution of linguistic theory. Furthermore, each individual chapter can be read independently of the others and this enhances the readability and the usability of the book. Repetitions occur across chapters but are not redundant because they ensure chapter independence.
One might feel a bit skeptical about some of the data and the judgments about them, in the sense that they might be too marginal or too far-fetched. Some of the data are crucial for the approaches advocated in the book. A discussion about the method used to collect these data would be helpful.
Furthermore, the reader is left with open questions as regards the treatment of causatives. In Chapter 7 the author presents arguments that causatives are created in the lexicon and in Chapter 8 that they are created in the syntax. One would expect that the new approach presented in Chapter 8 would give answers to the questions raised in the detailed discussion presented in Chapter 7. For instance, an explanation would be welcome as regards the way that the new approach deals with the -sase/-sas allomorphy described in Chapter 7 or a more explicit treatment of the binding facts that were the reason for assigning a dual subcategorisation feature to -sase.
Overall, ‘Case, argument structure, and word order’ is a collection of insightful and data-rich analyses of challenging phenomena in Japanese (and in other languages) within the theoretical framework of Minimalism. It is an excellent specimen of well informed and meticulously built linguistic analysis in the generative tradition. The work goes beyond the limits of the study of an individual language being, in fact, a study about universal grammar. In this sense, the book is a source of inspiration for students and researchers of linguistics of every theoretical persuasion and linguistic background.
Anagnostopoulou, Elena. 2003. The syntax of ditransitives. Berlin: Mouton de Gruyter.
Borer, Hagit. 2005. The normal course of events. Oxford, New York: Oxford University Press.
Cinque, Guglielmo. 1999. Adverbs and functional heads. Oxford: Oxford University Press.
Halle, Morris & Marantz, Alec. 1993. Distributed morphology and the pieces of inflection. In Hale, Kenneth & Keyser, Samuel Jay (eds.). The view from Building 20: Essays in honor of Sylvain Bromberger, pp. 117-176. Cambridge, Mass: MIT Press.
Levin, Beth & Rappaport Hovav, Malka. 1995. Unaccusativity: At the syntax-lexical semantics interface. Cambridge, Mass: MIT Press.
Marantz, Alec. 1993. Implications of asymmetries in double object constructions. In Mchombo, Sam (ed.). Theoretical aspects of Bantu grammar, p. 113-150. Stanford, Calif.: CSLI Publications.
Tsujimura, Natsuko. 1989. Unaccusative mismatches in Japanese. In DeJong, Ken and Yongkyoon No (eds.). ESCOL ’89: Proceedings of the sixth annual meeting of the Eastern States Conference on Linguistics, pp. 264-276. Columbus: Ohio State University.
ABOUT THE REVIEWER
| ABOUT THE REVIEWER:
Dr. Stella Markantonatou is a Principal Researcher with the Institute for Language and Speech Processing/R.C. Athena, Greece. Her research interests lie in the study of the lexical semantics-syntax interface (mainly within unification based formalisms) and the development of lexical databases for usage by humans and machines, especially for the Greek language. She has also worked on the development of corpus based Machine Translation systems.