Date: Tue, 29 Oct 2002 16:50:02 +0100 From: Pius ten Hacken Subject: Morphology: Review of Boucher (ed.) Many Morphologies
Boucher, Paul, ed. (2002) Many Morphologies, Cascadilla Press, xvi+267pp, paperback ISBN 1-57473-025-8, $28.95.
Publisher's announcement of the book: http://linguistlist.org/issues/13/13-2090.html
Reviewed by Pius ten Hacken, Universitaet Basel
This book grew out of a research project ''The Structure of the Lexicon'', funded by the French Ministry of Research, which included a summer school (1997) and a workshop on morphology (1998). It consists of nine papers, varying in length between 19 and 36 pages, preceded by a brief introduction. According to this introduction, by Paul Boucher and Marc Plenat, the book's title alludes to the variety of morphological problems and the diversity of formal solutions proposed. The review will be divided into nine micro-reviews summarizing and briefly commenting on the individual papers, followed by some general remarks. In a number of names, diacritics have been deleted to avoid character transfer problems.
Synopsis and Discussion of the Individual Papers
* In ''The Asymmetry of Morphology'', Anna Maria Di Sciullo presents a model of morphology claimed to be compatible both with Chomsky's (1995) Minimalist Program and with Kayne's (1994) theory of antisymmetry. It is based on morphological trees with two projection levels for affixes and roots. In this model, morphology and syntax are different components of grammar, working with similar operations applied in a different way. The morphological operations of SHIFT and LINK move elements in morphological trees. The central explanatory device is the Strict Asymmetry Hypothesis. The model is illustrated by application to a number of derivational suffixes, verbal compounds, and complex wh-elements such as ''everywhere''.
COMMENTS: Many readers without detailed knowledge of the author's earlier work might be put off by the introduction, which includes 33 references in slightly more than one page. Although she names her sections with general titles such as ''Derivation'' and ''Compounding'', these actually cover only a tiny selection of these fields and there is no indication of how (or whether) her treatment can be generalized.
Moreover, the author takes insufficient care in her formulation of arguments and in her treatment of the data. An example of the former is her confusion of ''does not imply that'' and ''implies that not'' in the explanation of asymmetry (p. 3). An example of the latter is her use of ''She went to the meeting you know about.'' as an example of a ditransitive preposition (p. 12). She also perpetuates the erratic claim that subjects cannot be included as non-heads in a verbal compound. This claim, formulated by Roeper & Siegel (1978) for ''-ed'', ''-er'', and ''-ing'', cannot be generalized. Allen (1978) gives counterexamples such as ''population growth''. A recent discussion of the ambiguity of examples such as ''government promotion'' and ''satellite observation'' can be found in Lapata (2002). Despite these obvious drawbacks, for specialists who know the background in the 33 references sufficiently well and are able to correct the errors in the presentation, the chapter offers some interesting ideas.
* In ''Middle Transitive Alternations in English: A Generative Lexicon Approach'', Christian Bassac and Pierrette Bouillon treat the alternation by which transitive verbs such as ''read'' can occur in middle constructions such as ''This book reads well''. They adopt the framework of the generative lexicon as developed by Pustejovsky (1995) and show how the internal structure of lexical entries can explain which verbs can occur in middle constructions of this type, how their meaning is affected, and why a modifier of a particular type is necessary.
COMMENTS: Adopting the classical structure of presenting a set of data, introducing a framework, and showing how the data can be explained in this framework, this well-written article presents a highly accessible argument for the theory of the generative lexicon. Although the section introducing the theory itself will be too brief for anyone not already familiar with the approach, the article will no doubt whet the appetite to learn more about it.
* In ''Unaccusativity Mismatches and Unaccusativity Diagnostics from Derivational Morphology'', Bozena Cetnarowska discusses the impact of morphological tests on the distinction between unaccusative (e.g. ''escape'') and unergative verbs (e.g. ''laugh''). The problem with this distinction is that diagnostic tests are language-specific and different tests do not coincide in borderline cases. By looking at problem cases with morphological tests in Polish and English, the author shows that these tests, when used with care, are no worse than syntactic tests. As opposed to the latter, they focus on ''deep'' rather than ''surface'' unaccusativity in Levin & Rappaport's (1995) terminology.
COMMENTS: This chapter is more of an overview than an argument for a particular point. As a consequence of the choice of a classification question as the main topic, no conclusion in terms of a theory explaining a set of data can be expected. The resulting tests, however, are language-specific and cannot be generalized. Although an attempt is made to justify them on the basis of a theory of grammar, they are not linked to a theory of classification.
* In ''Many Plurals: Inflection, Informational Additivity, and Morphological Processes'', Susan Steele argues for a processual view of morphology on the basis of data from Luiseno. In a morpheme-based analysis, Luiseno ''um'' is treated as a plural suffix. On closer inspection, such an analysis cannot be upheld. Instead, a number of different processes are proposed which change the features and affect the form of a word without resulting in one-to-one matches between forms and meanings.
COMMENTS: This article addresses one of the foundational questions in morphology, whether rules should be formulated as operations on morphemes or as processes, Hockett's (1954) Item & Arrangement vs. Item & Process. In the presentation and discussion of Luiseno data, a large degree of tolerance or knowledge of the literature on this language seems to be presupposed. Morphological categories are rather different from the ones familiar from European languages and a reader looking for any specification of whether we are dealing with nouns or verbs will be disappointed.
* In ''Gender 'Polarity': Theoretical Aspects of Somali Nominal Morphology'', Jacqueline Lecarme discusses a case of polarity, where pluralization of nouns is often said to change gender to the opposite value. A more precise inspection of the Somali data shows that it is possible to predict the gender of plural nouns on the basis of the plural marker and independently of the gender of the base. A similar analysis applies to Breton. These data are explained in the framework of Halle & Marantz's (1993) Distributed Morphology. It is proposed that Somali number is inserted in a different position in the tree from normal inflectional number.
COMMENTS: The careful presentation of an interesting set of data in this chapter effectively destroys the theoretically problematic concept of gender polarity. The second part of the chapter cannot be understood without familiarity with Distributed Morphology. An alternative analysis which is not considered at all is that in some languages nominal number is derivational rather than inflectional, cf. the argument for Japanese in ten Hacken (1994). It seems that the inflectional nature of nominal number is simply universally stipulated here.
* In ''Surface-to-Surface Morphology: When Your Representations Turn into Constraints'', Luigi Burzio develops an approach to word formation in the framework of Optimality Theory in which the task commonly attributed to word formation rules is taken over by constraints. The central constraint is Output-to-Output Faithfulness (OO-F), which requires base word and derived word to be as similar as possible. OO-F is subject to Gradient Attraction (GA). GA is an organizational force in the lexicon which tends to make similar items more similar. As a consequence, once OO-F is violated in one respect, it is easier to violate it in other respects as well. GA is itself explained by the Representational Entailments Hypothesis, a general cognitive device governing the strength of expectations in view of experiences.
COMMENTS: This is an absolutely fascinating article for anyone interested in the question of whether analogy can replace rules in morphology. It is extremely well-written and accessible even with only basic knowledge of Optimality Theory, because crucial points are explained more than once from different perspectives. Even for those who refuse to accept its sweeping conclusion, the article contains many points of inspiration.
* In ''An Experimental Constructional Database: The MorTAL Project'', Nabil Hathout, Fiammetta Namer, and Georgette Dal describe the functioning of two tools used in the construction of a lexicon database with derivational information for French, MorTAL. One of the tools, DeCor, is entirely based on statistical corpus analysis, the other, DeriF, also uses linguistic information. As a case study of the French suffix ''-able'' shows, DeCor achieves 85% accuracy, measured against the analyses found by DeriF. DeCor is more robust and requires less specification work for each affix, but DeriF yields more reliable results.
COMMENTS: This chapter was written for computational linguists. The section outlining the position of the work described here in the context of the general field reads more like an obligatory section of a research grant application than a real explanation of the background. The description of the tools is too superficial to give more than a rather general idea of their functioning. This is not unusual in computational linguistics, because one does not want to give away the advantage in the competition for future research grants. In the evaluation of DeCor and DeriF, DeriF seems to be taken as a standard of comparison. From a methodological point of view, this appears a rather unfortunate decision.
* In ''Applications of Computational Morphology'', Beatrice Daille, Cecile Fabre, and Pascale Sebillot describe a number of computational systems for the acquisition and use of lexical databases with morphological information, as well as a number of applications in which the morphological information in these databases is useful. The applications include lemmatizing unknown text, recognition of terms, and document retrieval.
COMMENTS: While the authors claim to review ''[t]he best known computational systems'', the selection of lexical databases, systems for lexical knowledge acquisition, and examples of applications is small and biased towards a French perspective. This bias makes it a useful supplement for people working in this area with a more international orientation.
* In ''A Common Basis for Syntax and Morphology: Tri-Level Lexical Insertion'', Joseph E. Emonds proposes a new model as a basis for a unified account of syntax and morphology. Following Lieber (1992) he assumes that morphemes (free stems, bound stems, affixes, and zero morphs) have lexical entries. Syntax and morphology build a single tree and use the same notion of headedness. Right-headedness is the universal default, but it can be overridden by language-specific rules. These rules, elaborated for English and French, do not exploit the distinction between syntax and morphology but those between open and closed projections (only IP, DP, AP, and PP are closed) and between free and bound elements. Affixes and stems are distinguished on the basis of their semantic specificity. Derivation and inflection are distinguished by the level of lexical insertion of the affix. Inflectional affixes are inserted at Phonetic Form whereas derivational affixes are inserted either at the start of syntax (if they are non-productive) or at the end of the transformational cycle, before spell-out (if they are productive). The effects of these different levels of insertion are shown for the English suffixes ''-ing'' and ''-ment''.
COMMENTS: This chapter offers a well argued and highly readable sketch of an intriguing approach to morphology. There may be some quibbles about details in the presentation, e.g. the structure [[[development] [of new roads]] into the hills] on p. 256, but on the whole both the presentation and the theory give an impression of careful elaboration. As it is presented here, the question remains, of course, to what extent the approach can be generalized to phenomena which motivated Anderson (1992) to adopt an Item and Process model, such as reduplication, ablaut, and suprasegmental modification. Since this chapter is presented as a brief outline of the main ideas in Emonds (2000), the reader's interest is attracted to the discussion of these issues there.
The first general observation about this book which comes to mind is that the title is entirely appropriate. Each chapter takes a quite different approach or theory as a basis, not in order to introduce it, but rather to illustrate how it is used in state-of-the-art research. The risk taken by the editors is then that few people will be able to appreciate more than one or two chapters. Avoiding this imposes an extremely difficult task on the authors. In this reviewer's opinion, Burzio and Emonds mastered this task to a remarkable degree. Some of the other contributions constitute an invitation to the reader to find out more about the theory adopted, because of the intriguing results. Yet others are less successful, as argued in more detail above.
The decision to include two chapters on computational morphology in this volume is difficult to understand. The way questions are asked and addressed in computational morphology differs fundamentally from the way they are dealt with in theoretical morphology. It hardly seems plausible that these chapters will appeal to the readership of the others. It is also strange that the more general chapter follows the more specific one.
The readership of this book consists of specialists in generative morphology who know the major current approaches in quite some detail and are not afraid to deal with specialized data sets from languages they do not know. Each chapter has its own bibliography, which is not surprising in view of the minimal overlap and facilitates the use of individual chapters in photocopies.
Allen, Margaret Reece (1978), ''Morphological Investigations'', Unpublished Doctoral Dissertation, University of Connecticut.
Anderson, Stephen R. (1992), ''A-Morphous Morphology'', Cambridge: Cambridge University Press.
Chomsky, Noam (1995), ''The Minimalist Program'', Cambridge (Mass.): MIT Press.
Emonds, Joseph E. (2000), ''Lexicon and Grammar: The English Syntacticon'', Berlin: Mouton de Gruyter.
ten Hacken, Pius (1994), ''Defining Morphology: A Principled Approach to Determining the Boundaries of Compounding, Derivation, and Inflection'', Hildesheim: Olms.
Halle, Morris & Marantz, Alec (1993), 'Distributed Morphology and the Pieces of Inflection', in Hale, Kenneth & Keyser, Samuel J. (eds.), ''The View from Building 20: Essays in Linguistics in Honor of Sylvain Bromberger'', Cambridge (Mass.): MIT Press, p. 111-177.
Hockett, Charles F. (1954), 'Two Models of Grammatical Description', ''Word'' 10:210-231.
Kayne, Richard S. (1994), ''The Antisymmetry of Syntax'', Cambridge (Mass.): MIT Press.
Lapata, Maria (2002), 'The Disambiguation of Nominalizations', ''Computational Linguistics'' 28:357-388.
Levin, Beth & Rappaport, Malka (1995), ''Unaccusativity: At the Syntax - Lexical Semantics Interface'', Cambridge (Mass.): MIT Press.
Lieber, Rochelle (1992), ''Deconstructing Morphology: Word Formation in Syntactic Theory'', Chicago: University of Chicago Press.
Pustejovsky, James (1995), ''The Generative Lexicon'', Cambridge (Mass.): MIT Press.
Roeper, Thomas & Siegel, Muffy (1978), 'A Lexical Transformation for Verbal Compounds', ''Linguistic Inquiry'' 9:199-260. ‡
ABOUT THE REVIEWER:
Pius ten Hacken is Privatdozent for general linguistics at the Universitaet Basel. His research specializations include morphology, computational linguistics, and the philosophy and history of linguistics. Ø