Arguments and Agreements
Edited by Peter Ackema, Patrick Brandt, Maaike Schoorlemmer & Fred Weerman
Oxford Linguistics Series
Oxford: University Press, 2006.
Hardcover. vi + 349 p. ISBN 9780199285730. £100.00
Reviewed by Geneviève Girard-Gillet
Université Paris III-Sorbonne nouvelle
The introduction by Peter Ackema, Patrick Brandt, Maaike Schoorlemmer, and Fred Weerman [3-32] sets the scene for the nine articles collected and raises the question of the complex links that exist between a type of morphology and a type of language. The book explores the way the linguistic phenomenon called agreement functions cross-linguistically as regards the expression of arguments. The term “agreement” refers to the morphological link that exists between two elements in a clause, with one determining the morphological shape of the other in a controller/target relationship.
Agreement comes in many forms, and one can distinguish between languages having a rich agreement morphology and those which have a poor agreement morphology. Italian, for instance, belongs to the first category and English to the second. In Italian, the rich subject agreement paradigm allows the subject to remain unexpressed syntactically (credo, credi, crede, crediamo, credete, credono), whereas English requires the subject to be present (I believe, you believe, he believes, we believe, you believe, they believe). In the Government and Binding Theory [Chomsky 1981], languages such as Italian were called “pro-drop” languages as the subject could be dropped, and languages such as English were called non pro-drop languages, since the subject could not be dropped, at least in finite sentences.
Polysynthetic languages also show a relation between rich argument and lack of overt syntactic expression of arguments. The question is then whether pro-drop languages and polysynthetic languages display or not some kind of similarity. In generative syntax, language variations are accounted for in terms of parameters, which fill in a certain part of the common scheme underlying all grammars of natural languages (Universal Grammar). The parameters are all what one may call “macroparameters”, as they have two values, and languages are divided into two classes according to the two possible values. Such a parameter has been proposed both for pro-drop and polysynthesis. This macroparameter is divided into a positive and a negative value :
Macroparameter for polysynthesis:
Positive value Negative value
(Mohawk, Navajo) (Italian, English)
Macroparameter for pro-drop:
Positive value Negative value
The choice for one or the other value of a parameter is made by the language-learning child on the basis of particular cues in the input, so-called triggers. An important trigger for a positive setting seems to be the presence of rich inflection. The question is whether the assumed relation between empty arguments and inflection is borne out and the contributions in this volume centre around what determines whether one or more arguments need not be expressed syntactically, and to what extent is this related to the inflectional make-up of the language. In other words, the authors try here to understand why languages do not all follow the same pattern, and they argue differently as to what determines the category the language under study falls into. A great deal of languages are analysed or at least alluded to in comparison with others: Brazilian Portuguese, Colloquial Portuguese, German, Dutch, Yiddish, Standard Finnish, Hebrew, English, Spanish, Italian, Greek, French, Chinese, Japanese, Icelandic, Haitian, Dutch sign language, Mohawk, Nahuatl, etc.
Within the Government and Binding Theory, Rizzi [1982; 1986] suggested that pro is subject to two distinct types of licensing condition: the occurrence of an empty pronoun must be formally licensed, and the content must also be licensed. There is an arbitrary list of heads in a language that licenses the occurrence of pro within their governing domain. But if pro is formally licensed, its content must also be licensed, which means it must be recoverable, and this can be achieved by rich inflection (in Italian, for instance). Since formal licensing and licensing for content are kept distinct, this theory can account for languages that have a rich inflection but do not allow pro-drop, and for those where pro-drop is formally licensed but where there is no possibility of identifying its content.
His theory accounts for the basic data, but there are still problems that remain unexplained and in particular the following question: Is it possible to conclude that such and such a language is pro-drop by counting the number of different affixes it exhibits? The answer is no. German, French and other languages have a rich verbal morphology but they are not pro-drop. What can only be said is that “the chances of pro-drop increases significantly once the number of distinct affixes increases” [Olaf Koeneman 86]. And yet it is not clear why Finnish and Hebrew, for instance, are not fully pro-drop. Microvariations in pro-drop languages need then to be studied.
Part 1: The Agreement-Pro-Drop Connection
In “Economy, Agreement, and the Representation of Null Arguments”, Margaret Speas proposes that there is no licensing per se, and hence no “pro-drop” parameter per se, and that null arguments occur whenever general principles of economy allows them. She intends to explain Jaeggli and Safir’s generalisation , which may sound strange indeed: “null subjects occur in the context of either rich agreement or no agreement at all”. Their generalisation accounts, for instance, for the fact that Chinese [Hang 1984, 1989] allows subjects and objects to remain unexpressed though it has no agreement morphology at all, unlike Italian. Taking into consideration the fact that agreement, unlike Tense or Aspect does not receive an independent interpretation, and then ought to be absent at LF, she hypothesises that in languages with strong agreement, a morpheme AGR heads the AGR projection, whereas in languages with weak agreement, the AGR morphology is just part of an inflectional paradigm. She discusses examples from German, Yiddish, Swedish, French, Italian, among others. Her proposal could explain more easily how children learn their mother tongue, and how they can go from thinking that their language allows null subjects to learning that it does not.
In “Deriving the Difference between Full and Partial Pro-Drop”, Olaf Koeneman tackles the problem of languages which do not entirely conform to the Identification condition: “An overt subject can be left out from a declarative clause if verbal agreement is rich enough to identify it”. He focuses on Hebrew and Standard Finnish, which are partial pro-drop languages, as thematic subjects can be dropped only in first and second contexts. This phenomenon can be related to the morphological correlation between pronouns and agreement that exists only for first and second persons (singular and plural). In Hebrew, this constraint is further restricted to the past and future tense. They also differ as regards word order. Olaf Koeneman contends that it is the diachronic evolution of the languages that is responsible for the morphological correspondence which he assumes explains the partial pro-drop. He distinguishes three stages of evolution: one in which a particular item X is listed as a pronoun or as a clitic, one transitional stage in which X functions either as a pronoun or as an affix, and the final stage in which X has been reanalysed as an affix. He argues that in Finnish and Hebrew at stage 2 not all pronouns started to appear as affixes, only the first and second persons did so. This suggests that third-person endings must have a different origin. He acknowledges that further research is necessary, in particular to see if the facts presented for Hebrew and Finnish really form a cluster.
In “Agreement, Pro, and Imperatives”, Hans Bennis adopts a different approach to the minimalist version of the pro-drop parameter advocated by Speas in this volume. He thinks that the availability of pro is not language-specific, but construction-specific, and to make his point analyses imperatives in Dutch. In this language it is possible to leave out the subject if the imperative verb is uninflected; but if the imperative has t-inflection or en-inflection, the subject U (you + polite) or jullie (you + plural) has to be present. In that case the subject must be the strong pronoun, and it is stressed. Hans Bennis considers then that the imperative subject behaves like a subject in pro-drop languages, such as Italian and Spanish, where the strong lexical pronoun has the empty pronoun as its weak correlate. The presence of pro in Dutch imperative is discussed in relation with the presence of the force feature in the C-domain of root clauses.
Part 2: Microvariation in Pro-drop Languages
In “Uniform and Non-Uniform Aspects of Pro-Drop Languages”, Artemis Alexiadou addresses the similarities and the differences between Greek/Spanish and Italian and Hebrew/Finnish in terms of the general framework introduced by Alexiadou & Anagnostopoulou . This framework takes the view that languages split into two groups on the basis of the mode of EPP (Extended Projection Principle) checking. EPP is universal and is understood as a D feature on I. Pro-drop languages satisfy this feature via V-movement to I owing to the pronominal nature of their verbal agreement. Non-pro-drop languages move or merge a DP in Spec,IP to check EPP. Alexiadou attempts to identify the structural properties that are the source of the differences existing between the languages under study. The variations depend on how their morphology checks EPP. Alexiadou agrees then with Speas that morphology is a necessary but not a sufficient condition for the emergence of null subjects.
In “Asymmetrical Pro-Drop in Northern Italian Dialects”, Cecilia Poletto aims at determining to what extent the presence of overt morphology influences syntactic processes. She compares Rizzi’s theory with what a minimalist approach can suggest, in the terms of Alexiadou & Anagnostopoulou . Drawing from Benveniste , she considers important to take into account the distinction between the first/the second persons of the verbs, which are [+deitic] persons and the third, which is [-deictic], but suggests that a more refined distinction within the domain of [+deictic] persons is necessary to account for the split between pro-drop and non pro-drop persons within this domain. She first looks at the diachronic evolution of Romance languages (French and Northern Italian dialects) and proposes a reformulation of Haiman’s generalisation about the loss of V2 construction and the development of subject clitics. As French is now a non pro-drop language, and as its verbal morphology did not change between Middle and modern French, she concludes that morphology is not the immediate trigger for null subjects. She then studies Venetian and Paduan dialects which display the same evolution.
Part 3: Interpreting Empty Arguments
In “Agreement Phenomena in Sign Languages of the Netherlands”, Inge Zwitserlood and Ingeborg Van Gijn are concerned with the treatment of arguments in sign languages, as space is used for the expression of pronouns, verb agreement and spatial relations. They first explain the agreement system in NGT (sign languages of the Netherlands) by taking some verbs as examples: wait, visit, fall, put down, give, send, find, meddle, love. These verbs show that there are two kinds of agreement in NGT: location agreement when the verb is made near the location of a referent in signing space (wait, visit) and gender agreement when the hand configuration is determined by characteristics of the referent such as its shape or animacy (put down a book / a pen / a glass). Combinations of location and gender are possible, as in the case of give, as the type of agreement marking that a verb takes for its arguments is predicted from its semantic roles. This mode of functioning, which makes use of space, lead the authors to consider that in sign languages the elements that can be considered as adjuncts in spoken languages are actually arguments: Source, Goal, Location. The two types of agreement are then: locus agreement (Agent, Patient, Recipient, Source, Goal) and gender agreement (Theme). Then they discuss the relevant phi-features and argue that the formal category “person” does not exist in signed languages. The system makes use of the grammatical category “location”. Turning to the question of null arguments, they adopt an identification procedure similar to the discourse-pragmatic solution given by Huang . If a null argument cannot be identified within the sentence, it will be revovered thanks to an appropriate sentential topic.
In “ ‘Arbitrary’ Pro and the Theory of Pro-Drop”, Patricia Cabredo Hofherr discusses evidence that the possibility of pro-drop is not directly dependent on the theta-role borne by the pronoun. She proposes a modification of Rizzi’s theory, which distinguished three types of pro-drop: referential pro, quasi argumental pro, expletive pro. She tackles the split among the referential third person pronouns, in Modern Hebrew, Finnish, Russian and Icelandic (partial pro-drop), a split which has been less analysed than the first/second person split. Her modification is based on the distinction between deictic, anaphoric and non-anaphoric null pronouns, and the differences are derived from their content identification: an anaphoric third person null pronoun has a full set of phi-features (person, number and gender), whereas a non-anaphoric third person null pronoun only identifies a subset of phi-features. Her analysis of third person plural pronouns shows that the distribution of non-anaphoric third person plural is closer to that of quasi-argumental subjects than to that of anaphoric argumental third person pro. Certain non-anaphoric third person plural interpretations are indefinite (vague readings) and others are definite (universal or locative readings).
Part 4: Non-configurationality
In “The Pronominal Argument Parameter”, Eloise Jelinek analyses a class of Pronominal Argument languages (“polysynthetic” for M.C. Baker 1995, 1996), and in particular Navajo (Southern Athabaskan) and Lummi (Straits Salish) in order to show how they differ from the class of pro-drop languages. These languages are called by Jelinek “Pronominal Argument” languages because pronouns are obligatory for the sentence while the DPs are not. The subject and object pronouns are included in inflectional strings, and they are all backgrounded and discourse-anaphoric. The morphosyntactic status directly reflects information status: pronouns are topical, unstressed anaphors referring back to a referent earlier in the clause or the discourse; new information is given by predicates or DPs, which carry inherent focus and stress. To place focus on a pronominal argument, one must add a freestanding focus pronoun in an A-bar position preceding the verb sentence. Unlike Lexical Argument languages, where both pronouns and DPs serve as arguments, Pronominal Languages do not show the agreement relation that licenses pro-drop. Two other differences are pointed out: only pronouns are in A-positions, and the two languages under study, Navajo and Lummi, lack determiner quantification. These contrasts between PA and LA languages are directly related to the way the mapping between information structure and morphosyntactic structure is implemented.
In “On Zero Agreement and Polysynthesis”, Marc C. Baker presents a case study of Mapudungun, that he argues has implications for ideas like the Polysynthesis parameter and the Pronominal Argument hypothesis (Jelinek in this volume), if the phenomena is to be better understood. The languages studied vary as regards the Polysynthesis parameter: Mohawk, for instance, has the “yes” value, whereas Chichewa and Kinande have the “no” value. The analysis of non-configurational languages depends on the ability to recognise null agreement affixes. Mohawk verbs seem not to agree with third-person neuter inanimate objects. Another language, Mapudungun (Chilean and Argentinean Andes) has an overt third-person object agreement marker fi, which seems to be optional. Baker mentions several characteristics, such as the Ø agreement, the treatment of morphological causatives, embedded clauses, noun incorporation. He also tries to understand how children acquire the language, and he carried out an experiment with a young child for six months. His study of Mapudungen, even if it displays specific data, may help study other polysynthetic languages, and the question of agreement.
As a conclusion, this volume, which is well organised, will interest linguists concerned with the means to retrieve absent arguments in a sentence. Languages vary a lot as regards which arguments can be absent and why the possibility of this absence exists. The authors take up a question that has been much debated: can agreement morphology have direct import for the selection of the pro-drop parameter? They present various approaches and their analyses bridge the gap between a Government and Binding-based approach and a Minimalist-based one, as the latter is less interested in the pro-drop phenomenon. The solutions seem to be equivalent from the empirical point of view, at least with the data available and discussed. The very positive point is that the authors often quote each other, and this makes the discussion of the data debated easy to follow, even when they discuss languages the reader might not be familiar with. Questions about how children acquire a language are also addressed: Do all children begin by assuming that subjects can be absent, and then stop omitting them, or is it the contrary? The volume is also a good read for those interested in the interactions of syntax and morphology in various languages..
Cercles © 2012
All rights are reserved and no reproduction from this site for whatever purpose is permitted without the permission of the copyright owner. Please contact us before using any material on this website.