Date: Sun, 24 Jul 2005 22:53:59 -0700 (PDT) From: Sanjukta Ghosh Subject: Structuring Sense, Vol 1: In Name Only
AUTHOR: Borer, Hagit TITLE: Structuring Sense, Volume 1 SUBTITLE: In Name Only PUBLISHER: Oxford University Press YEAR: 2005
Sanjukta Ghosh, Research Scientist, Language Technologies Research Centre, International Institute of Information Technology, Hyderabad
[Structuring Sense, Volume 2: The Normal Course of Events, by Hagit Borer, was reviewed in http://linguistlist.org/issues/16/16-2184.html --Eds.]
[Angle brackets have been replaced by square brackets throughout. Material immediately following a closing square bracket is to be understood as subscripted; also the notations ''max'' and ''min''. -- Eds.]
This book is an excellent thought-provoking work by a great intellectual mind of current linguistic world written from a non-conventional perspective challenging generative as well as lexical frameworks. The book under review 'In Name Only' is the first of a three volume work entitled 'Structuring Sense', by Hagit Borer. Difference between word and structure is the subject matter of this series. The question addressed in these volumes is how words can mean many things but structures cannot. The significant shift proposed in these three volumes explains linguistic competence only from the syntactic computation, thus shifting the load from the lexicon of a language. Borer selected two most important areas of a language, viz. nominal structure and predication in the first two volumes (third volume under preparation) to establish the theory proposed. The reading of the first volume of the book leads to a desire for the reading of the second volume without which the theory proposed seems to be incomplete.
The volume I is divided into three parts. The first part sets the stage for the whole series by introducing the theoretical mechanism in the first two chapters. The second and third part covers the entire area of nominal structure covering topics like interpretation of Proper Nouns, mass-count distinction, quantifiers and measure phrases, classifiers, singular-plural distinction and definite as well as indefinite articles. The difference between these two parts lies in the choice of the language taken: while in the second part English nominal structure is focused with some relevant examples taken from Hebrew, Chinese etc, the third part exclusively discusses other than English data for accounting language parameters.
Borer develops a new model of linguistic analysis (similar to some extent to the constructionist and neoconstructionist approach) which she calls Exo-skeletal approach in opposition to endo-skeletal where properties of the higher structure is projected from the lexicon. This lexicon-driven view is opposed in this work where listeme is considered nothing more than a sound-meaning pairing giving an encyclopedic knowledge. Syntactic properties typically associated with listemes like argument structure and category type are properties of the structures in exo-skeletal approach. When listemes are selected to form a part of the conceptual array, they are not an ordered pair in an unmarked Lexical-Phrasal domain (L-D). Alongside the encyclopedia, the grammar has a functional lexicon comprising of the grammatical formatives in the forms of features like [+pl], [+pt] as well as independent grammatical formatives like [the [+def]]. Some grammatical formative merges with L-D and projects some functional projection.
Functional heads have open values which are assigned range by a variety of means like f-morphs and abstract head feature (direct range assignment) and other operators such as independent morphemes like quantifiers and discourse operators (responsible for indirect range assignment). Open values are denoted with a subscript which marks its category membership like the following [e]d. The system does not predict a one-to-one correspondence between these open values and range assigners because it is theoretically possible for an operator to bind more than one variable. However, the reverse is not possible as double marking of one open value by two operators is not allowed as natural languages do not allow a double marking. The derivation converges just in case the phonology dispenses a representation for the combination of (head + head feature).
There is a discussion on the inflection vs. derivation account within the framework in chapter 2 where Borer separates form and function in the domain of inflection. She assumes while more than one head feature can be realized on a single L-head, those features are neither ordered nor hierarchically organized. Some phonological regularities are associated with the realization of the feature combination but at morphological level there is no ordering. The argument is supported with data from Hebrew where form failing to predict function in the inflection domain and the order of inflection marking differs from one instantiation to another. So the function of inflection is clearly regular and syntactic but no prediction can be made about the forms.
Thus setting the stage of the main work for the remaining two more sections of the book, Borer takes up the issue of interpretation of proper name and common-name in the third chapter. She came to the conclusion that the distinction is neither the property of the lexicon nor can be explained by a type-shifter. Rather, this distinction emerges from two distinct syntactic structures in which range is assigned differently. Proper name interpretation emerges only when a noun is in D, and such a noun may move to D either by overt movement as in Italian or by covert movement as in English. In case the noun moves to D, it has a head-feature called [def-u] which assigns range to [e]d. She also argued that like definite articles, proper name behaves like a discourse anaphor taking its reference from discourse. On the other hand, in presence of an article in both the languages nouns are interpreted as common nouns occurring in NP and range-assignment to [e]d is done by the articles.
Chapter 4 introduces two more functional open values [e]DIV and [e]# for accounting mass vs count distinction. Open value [e]DIV is responsible for classifying or dividing function and is assigned range by either the plural markers and the indefinite article in English. A very interesting proposal by the author here is all noun extensions are mass in all languages (here differing from Chierchia (1998) who proposed this only for classifier languages) and count as well as mass are grammaticality constructed notions corresponding to different pieces of syntactic structures. Borer argued that plural inflection is actually a classifier inflection and there is no language where they coexist in a single structure (though both plural and classifier may be present as two different strategies in a language). Chierchia's system was unable to account for the questions like why classifier languages cannot have plurals with classifiers or why the languages with grammatical plural cannot have classifiers. Borer answers this question by saying that classifiers and plural markers compete for the same functional projection or an open value [e]DIV can be assigned range by either a classifier which is an independent f-morph or a plural marker which is an abstract head feature.
She also argues that mass nouns are not inherently plural but unmarked for either mass or count and mass interpretation is defined from a default interpretation associated with the absence of a dividing structure. Bare plurals cannot be a function of singulars, but rather follow from the role of the plural as 'stuff divider'. Apples as a bare plural does not consist of singular apples, therefore individual cannot be created by a dividing function.
She argues further for a #P position dominating the Cl[assifier] Phrase which is responsible for the assignment of quantity to stuff or to divisions of it. Absence of Cl max gives rise to mass interpretation and absence of #P leads to non-quantity interpretation for bare plurals and determinerless mass nouns with generic interpretation. Bare plurals and mass nouns share the undetermined nature of the quantity involved and failure to show telicity of the event.
In the next chapter Borer tries to account for a null [e]d licensing. She assumes that strong quantifiers, unlike weak ones, merge a copy in D°, where they assign range over [e]d. Determiners which bind [e]d in this way are called D-determiners. Some of these are every, each, all, any, both and they do not allow the or this before them as these words also merge in the same D position.
Cardinal and weak quantifiers are ambiguous between a strong presuppositional reading with a movement to D and a weak reading with a null D. If they have a strong reading they bind [e]d, thus existential closure is blocked and the resulting expression behaves like a proper noun or a definite description.
If [e]d is bound with a D-determiner, there is no longer a variable in D and depending on the properties of the determiners (like if they are quantifiers such as each and every), the resulting expression is quantificational.
With examples from English and Hebrew, Borer concludes that any approach which tries to reduce the scopal properties of indefinites to a single structure is insufficient for analyzing certain specificity markers (the example given from Hebrew). The scope of the quantifiers as well as definiteness and indefiniteness reading emerge from a specific syntactic placement of determiners and the open values which they assign range to. So the properties of QP (#P) and of DP are responsible for derivation of strong vs. weak reading of quantifiers, cardinals and indefinites.
Chapter 6 investigates the nature of null #P and the properties of the definite article. Particularly the ungrammaticality of the construction where both definite and indefinite articles occur like * the a cat is discussed in the first section. It is conjectured that both these articles compete for a same structural position, which cannot be D° but must be either Cl° or #°or both. She concludes that both the and a compete for the same #P position resulting ungrammaticality of the aforesaid construction. The relation between telicity and definite descriptions is also attributed to the fact that in those constructions # is assigned range. This leads to the conclusion that determinerless mass nouns and bare plurals are not quantity expressions.
Comparing the ungrammaticality of *the a with the grammatical structures where definite article the co-occurs with weak quantifiers like few, many and cardinals, it is concluded that the and a are only heads of the quantifier phrase and they must project to #max. Whereas others can be either Fmin of Fmax, occupying either head/specifier position. She also explains in this context why 'all the' is an acceptable construction in spite of 'all' being a strong quantifier. 'All', according to this author, is a modifier of DP and occurs at the Spec of DP which is unique to it. All other strong quantifiers compete with the for merging into D slot and therefore are not permitted together with it.
Coming back to proper nouns again, in this chapter Borer proposed that they must assign range to both [e]# and [e]DIV because they are neither mass nor plural and no other quantifier is allowed with them to assign range over them. With the presence of a quantifier, a proper noun becomes a common noun.
In the last section of the chapter discussing Chinese data Borer proposed that Classifiers can assign range to both [e]DIV and [e]#. There is an interesting proposal to merge cardinal and classifier node coming from Simpson (forthcoming) mentioned by the author which she also can accommodate in her theory. In that case, the specifier position will be taken by the number and the head of the phrase and the range assigner will be the classifier. Singular interpretation emerges in this system with the classifier not because that is specified for singular but because of the structure identifies [e]DIV and [e]# and no other value for [e]# is provided. The plural classifiers are inherently quantificational and assign range to [e]#. For a strong interpretation of Num-Cl-N construction, Num movement to D is proposed and cardinals assign range to [e]d . Weak interpretation emerges [e]d when is bound by an existential operator. Strong interpretation of bare Classifier and Noun in some Chinese dialects suggests that Cl can assign range to [e]d also.
Part III of the book analyzes extensively Hebrew nominal system particularly definite articles, structure of singulars, the distribution of cardinals (chapter 7) and the classifiers as well as the measure and container phrases (chapter 8) using the same tool. The data presented in these two chapters elaborates the interaction between the range assignments by various means and the emerging syntactic and semantic properties of the nominals.
The last chapter concludes the book by discussing the language variation in relation to universal properties of language. It is proposed in this system that ultimately all variations are properties of range assigners to functional open values and these properties are morpho-phonological properties as f-morphs, abstract head feature or potential phrasal projections. The direct impact of this proposal is claimed to simplify the task of language acquisition. It leads to the point where syntactic acquisition becomes the task of matching the properties of phonological forms of the grammatical formatives with the set of otherwise unchanged grammatical computational operations. In this theory language parameters are the properties of the morpho-phonology only.
In this connection, I am tempted to share certain observations from Bangla (popularly known as Bengali), a classifier language of Indo-Aryan family. While analyzing the classifier system of Bangla, I found out that they essentially do two types of job: one is quantification/counting and another is of individuation, a cover term for me for denoting specificity- definiteness-focusing effect. The classifiers can be graded based on these two properties. A default classifier which can be used with categories such as N (both common noun and proper noun with certain pragmatic effect), Q, Num, Dem, V is /Ta:/ and I proposed in absence of any overt determiner in this language the classifier takes some of the jobs of overt D in certain construction. This can be translated in Borer's term they can assign range not only to [e]DIV and [e]# but also to [e]d.
For people like me who work in Syntax-Pragmatics interface, Borer's theory looks to be extremely positive. If a grammatical description were to make only projection and compositionality the relevant factors determining the interpretation of utterances, then one would be forced to leave interpretation entirely, without residue, to the social conventions of the language. That type of picture would leave no role for the speaker, whose freedom to anchor an utterance in its context would become vacuous. On the other hand, If the speaker constructs a syntactic template before filling it with the lexical items she has the choice to opt for a certain template for a specific reason (say specificity or focusing) from two allowed by grammar. This work, in this way, does not snatch the speaker's freedom away.
The Functional open values of Borer resemble the predicates of the predicate logic and the functional elements, which assign range to them, are the arguments. The predicate as we know is a constant, therefore, is the property of the language universals whereas the parameters are set with the argument-like functional elements which are variables. The recent research in neurobiology also supports the existence of a distinct predicate-argument structure in the human brain and the information processing happens in two different ways based on this distinction. Borer's work is probably guiding us towards a right path which can explain the intra as well as the interlinguistic variations more satisfactorily.
Chierchia, Gennaro. 1998. 'Reference to kinds across languages'. Natural Language Semantics 6:339-405.
ABOUT THE REVIEWER:
ABOUT THE REVIEWER
Currently I am working as a research scientist in Language Technologies Research Centre(LTRC) of International Institute of Information Technology, Hyderabad, India. In LTRC, my primary responsibility is to give linguistic input for English-Bangla as well as English-Hindi Machine Translation system. I was awarded Ph.D. degree in 2004 in Syntax- Pragmatics Interface studies of Bangla by University of Hyderabad. My research interests include syntax, pragmatics, philosophy of language and cognitive linguistics. I also taught in University of Hyderabad as a guest lecturer from July 2002 to December 2004.