LINGUIST List 20.3997

Sat Nov 21 2009

Review: Morphology; Semantics: Lieber (2009)

Editor for this issue: Monica Macaulay <monicalinguistlist.org>


        1.    Barbara Schlücker, Morphology and Lexical Semantics

Message 1: Morphology and Lexical Semantics
Date: 21-Nov-2009
From: Barbara Schlücker <schlueckzedat.fu-berlin.de>
Subject: Morphology and Lexical Semantics
E-mail this message to a friend

Discuss this message

Announced at http://linguistlist.org/issues/20/20-344.html

AUTHOR: Lieber, Rochelle TITLE: Morphology and Lexical Semantics SERIES: Cambridge Studies in Linguistics 104 PUBLISHER: Cambridge University Press YEAR: 2009

Barbara Schlücker, Institut für deutsche und niederländische Philologie, Freie Universität Berlin

INTRODUCTION

This book is the paperback edition of the 2004 monograph. It deals with the semantics of morphology, a topic to which little attention has been given so far. In fact, the book is on the semantics of word formation rather than on inflection and the main focus is on derivation although compounding and conversion are also dealt with along the way. In the introduction, Lieber identifies four problems concerning the meaning of word-formation processes: (1) The polysemy question, i.e. why are derivational affixes often polysemous, e.g., '-er' in 'writer' (agent noun), 'opener' (instrument noun), 'loaner' (patient noun)? (2) The multiple-affix question, i.e. why does English often have several affixes that perform the same kind of function or create the same kind of derived word, e.g. '-er' and '-ant' which both create agent nouns? (3) The zero-derivation question, i.e. how do we account for word formation in which there is semantic change without any concomitant formal change, viz. conversion? (4) The semantic mismatch question, i.e. why is the correspondence between form and meaning in word formation sometimes not one-to-one?

The lexical semantic representations in the framework Lieber develops in the course of dealing with these problems consist of two major parts, the semantic skeleton and the semantic body. The skeleton of the representation is hierarchically arranged and fully decompositional, based on six semantic features which distinguish the major ontological categories of lexemes and the basic concepts of time, space, and quantity. The body of the representation is noncompositional, holistic, and presumably only partially formalizable. It comprises the perceptual and cultural knowledge that forms the bulk of the lexical representation.

The focus of the book is on the skeleton, dealing with the body aspects only occasionally. Contrary to the body, the semantic skeleton is stable and formalizable and it contains those aspects of meaning that have consequences for syntax. The Principle of Co-indexation is introduced in order to integrate the (skeletons of the) parts of complex words, both compounds and derivations, into one single referential unit.

Along with numerous examples, this book also contains a series of detailed case studies on several English affixes such as '-ee', '-ize', '-ify' and 'over-'.

As noted above, this book is the paperback edition of a 2004 monograph. This raises the question of why we should review this book again. There are several reasons to do so. First, publishing the paperback edition is evidence of the great interest in this book. Second, to my knowledge, no other monograph on the semantics of derivation has been published since then. Third, as compounding is dealt with only briefly in this book, I will also discuss a recent paper on compounding by the same author which extends the framework as developed in this book further (Lieber 2009). This paper also refines the notion of semantic body, which receives only passing attention in the present book.

In the following, I will summarize the main parts of the framework and comment on the questions and solutions presented.

SUMMARY

(I) The semantic features Building on the works of, among others, Szymanek (1988), Jackendoff (1990), Pustejovsky (1995), and Wierzbicka (1996), Lieber introduces six semantic features to be used as a basis for the framework. Some of them seem to be similar to semantic features or primitives introduced in previous works by others, but they often differ from those, among other things, with regard to their ''grain size''. These six features are regarded as a beginning system, not yet capable of capturing all lexical meanings and to be extended in future work (note that Lieber 2007 adds the feature [scalar]). All features are used in a cross-categorial way. When combined, they can be used to define the major syntactic categories noun, verb, adjectives, and adpositions. The features are used both in an equipollent and privative way, i.e. they are binary in value but they may also be entirely absent in the semantic skeleton of a lexical item if that feature is irrelevant for the item in question.

The first three features to be introduced are [material], [dynamic] and [IEPS], the last of which stands for ''Inferable Eventual Position or State''. Presence of the feature [material] indicates (the conceptual correspondence of) nouns; the positive/negative value distinguishes concrete from abstract nouns respectively. Presence of the feature [dynamic] signals the concept of situations; the positive/negative value indicates an event/process or a state respectively. [IEPS] adds a path component to the meaning. If it is absent, a path meaning is irrelevant to the lexical item (e.g., 'eat'), if it is present, either positive or negative, it adds to the item the meaning of a direct or a random path respectively, e.g., 'descend' or 'walk'.

The feature [Loc] (location) signals location in space or time. More precisely, presence of [Loc] indicates relevance of position or place in time or space for the given lexical item; [±Loc] means either position/place or explicit lack of position/place, as in 'remain' or 'lack' respectively. The features [B] (bounded) and [CI] (composed of individuals) finally are needed in order to capture quantificational meaning. [B] signals the relevance of intrinsic spatial or temporal boundaries, either in a situation or (the conceptual correspondence of) nouns. [CI] finally indicates that the meaning of a lexical item is composed of separable similar units. When combined, [B] and [CI] can be used to classify singular and plural count and mass nouns, e.g., [+B, -CI] for 'person', [-B, -CI] for 'furniture' or 'water', and [-B, +CI] for 'cattle'.

(II) The Principle of Co-indexation Chapter 2 introduces the Principle of Co-indexation. This principle governs the integration of the skeleton parts of a complex word into one single referential unit. The Principle of Co-indexation says that in a configuration in which semantic skeletons are composed, the highest nonhead argument is to be indexed with the highest (preferably unindexed) head argument. The complex word determines how many arguments are eventually projected into the syntax; for this reason, co-indexation is responsible for the syntactic properties of the resulting complex word. The main difference between compounding and derivation is that in a derived word the skeleton is created by subordination and in a compound by concatenation.

(III) Polysemy of affixes The answer to the polysemy question, i.e. the question why derivational affixes are often polysemous, has to do with the fact that the semantic content of these affixes (as codified in the skeleton) is minimal, abstract and underdetermined and that these affixes normally do not have a semantic body. Combined with a semantically richer base, the semantic contribution of the affixes can be spelled out (and lexicalized) in many different ways. This kind of polysemy is called ''constructional polysemy'', as in the case of '-er'. However, polysemy of affixes may also arise under pragmatic pressure. This happens if lack of a specific derivational affix with the required sense in a given situation/context forces a (semantically close) productive derivational process to fill the semantic gap, as in the case of '-ery', where the basic collective reading (as in 'jewelry') is stretched to denote places (such as 'piggery'). This second kind of polysemy is called ''sense extension'' (see Copestake and Briscoe 1996).

(IV) Multiple affixes The answer to the multiple-affix question, i.e. why English often has several affixes that perform the same kind of function or create the same kind of derived word (such as the negative prefixes 'in-', 'un-', 'dis-', 'non-') is again related to the assumption that the semantic content of affixes is underdetermined, i.e. their skeleton is characterized by a small number of semantic features, and that they normally do not have a semantic body. The overlap of meaning results from the minimal semantic representations of these affixes; slightly different nuances of meaning emerge from the combination with the meaning of different bases in the process of derivation. So, for example, the meaning of all four negative prefixes 'in-', 'un-', 'dis-', 'non-' is only characterized by [-Loc] and it is only in combination with the different bases that the meaning nuances privation, contrary negation, contradictory negation, and reversativity arise.

(V) Conversion As to the question of semantic change without a concomitant formal change, AKA conversion, Lieber argues that this is not a process of zero-derivation because the diversity of semantic patterns exhibited by such changes is much wider than that of a single affix, even if the relative parsimony of the affix meaning and the resulting polysemy of affixes is taken into consideration. For this reason, it is impossible to identify a single (zero-) affix that would be able to cover all possible meanings. Rather, conversion is considered as one form of coining new lexical items and these new forms enter the lexicon just like any other novel lexical item, namely by listing. Thus, conversion is analyzed as relisting (see Lieber 1992).

(VI) Semantic mismatches The final question, concerning semantic mismatches, is on phenomena like derivational redundancy, empty morphs and semantic subtraction, i.e. on cases where apparently there is no one-to-one correspondence between form and meaning. Semantic subtraction means that a complex word formally has the structure [[base+X]+Y], where both X and Y are affixes, such that Y attaches to an already affixed base, but semantically, Y attaches to the base directly rather than to the affixed base, thus yielding the structure [base+Y]. An example for semantic subtraction would be 'realistic', where the suffix '-ic' semantically attaches to the base 'real-' rather than to personal noun 'realist' - contrary to a form like 'novelistic'. However, according to Lieber, semantic subtraction as well as empty morphs should not be characterized as semantic mismatches but rather as cases of allomorphy. As for derivational redundancy, Lieber assumes that there is no strict semantic rule preventing redundancy. However, the resulting, seemingly redundant forms must obey the pragmatic restriction of ''usefulness''.

As indicated above, Lieber (2009) develops the framework of the monograph at hand further with regard to two aspects which I will present briefly below, namely the nature of the semantic body and compounding.

(VII) The semantic body Lieber (2009) explores the nature of the semantic body as well as the relationship between the skeleton and the body. Just as in Lieber (2004), this paper proposes that the body contains various perceptual, cultural and other encyclopedic aspects of meaning that (often) vary from speaker to speaker, but this paper assumes more systematicity and structure of these meaning aspects. Elaborating on the metaphor of skeleton and body, the semantic body consists of two layers: the ''muscular structure'', which is the more systematic part, and the ''fat'', which is unsystematic and which contains speaker-individual knowledge. There is a universal inventory of 'muscular-body' features, such as {animate}, {human}, {n dimension}, or {function}. Examples of individual, unsystematic, encyclopedic meaning elements of the ''fat'' part are {for sleeping, contains comfortable surface, ...} for the lexical entry for 'bed'. Although the Principle of Co-Indexation governs the integration of only the skeleton parts of a complex word, the body parts also have an influence on this process, as they can affect realizability. For example, the first constituent of the compound 'cookbook author' bears the body feature {-animate} but the second {+animate}. Although according to the Principle of Co-indexation the (only) argument of 'cookbook' should be co-indexed with the first argument of 'author', the incompatibility of these body features enforces co-indexation with the second (internal) argument of 'author', leading to the interpretation of an author of cookbooks instead of an author who is a cookbook.

(VIII) Compounding Lieber (2004) as well as Lieber (2009) focus on nominal (N+N) compounds, leaving aside adjectival and verbal compounding (except for a side note on V+V in the 2009 paper). Lieber (2009) proposes three different analyses for attributive (root), synthetic, and copulative compounds, in each case for endocentric as well as exocentric compounds. Attributive endocentric compounds, the most prominent group of compounds in English, are seen as a semantic default, whereas synthetic and copulative compounds, whose internal semantic relations are much more constrained (both in terms of their skeleton and body), are ruled out. Copulative compounds are simply realized by co-indexation of the sole referential argument of both stems; this is possible only if the skeleton and the (muscular) body part of both stems are identical, e.g. 'scholar-athlete'. Synthetic compounds, such as 'truck driver' or 'table leg', exhibit an argumental relation between both stems, i.e. the non-head stem is interpreted as an argument of the head. According to the Principle of Co-indexation, it is co-indexed with the highest head argument, which in the case of 'truck driver' is the internal argument. However, special semantic requirements as with '-ee' in the case of 'employee' (requiring a sentient, nonvolitional argument) may also lead to co-indexation with the external argument as in 'city employee'. Similarly, body restrictions may enforce irregular co-indexation, see 'cookbook author' above. If the structural requirements of copulative and synthetic compounds are not met, the default interpretation is the attributive interpretation, such as 'dog bed' or 'phone book'. As the resulting complex word must have only one (referential) argument, both stems must be co-indexed and as it is clear that this may neither be a relation of coordination nor subordination, the relation between both stems is left open, provided that the second one determines the syntax and semantics of the whole.

Attributive and synthetic compounds differ in an interesting way with regard to exocentric cases: whereas so-called exocentric attributive compounds (e.g. 'birdbrain') are in fact endocentric compounds which are interpreted metonymically, see Booij (1992, 2002), exocentric synthetic compounds (e.g. 'pickpocket') seem to be structurally exocentric, because this exocentricity arises from an unindexed verbal external argument which is free to receive an implicit interpretation. Therefore, these exocentric compounds are restricted to an interpretation consistent with external arguments, i.e. agentive (as in the case of 'pickpocket') or instrumental.

EVALUATION

The ultimate objective of Lieber's work is to develop a framework for a semantic approach to morphology which accounts for all word formation processes, which is based on case studies and which can be applied not only to English but also cross-linguistically. This is a very ambitious goal and Lieber clearly does not claim to accomplish it in her monograph. Rather, she characterizes her framework as a beginning system which has to be refined and extended; in the conclusion, she identifies several questions and issues for further research, among other things, the issue of inflection.

Some points of criticism are accounted for to some extent in the 2009 paper. Among these points are a more detailed description of the semantic body, including the question about its internal structure, about the way in which it is embedded in the meaning of a lexical item and about the relation between the skeleton and the body. The issue of compounding is also dealt with in a more detailed way in the 2009 paper. However, several questions remain unanswered: regarding compounding, the interesting and important case of phrasal compounding is not discussed and at the moment, I cannot see how co-indexation of a phrase and a head stem in phrasal compounds like 'underwater camera' or 'God-is-dead movement' would work. A more fundamental objection concerns the formation of lexical semantic classes, based on the six semantic features. The stated purpose of the semantic skeleton is to represent those (and only those) aspects of meaning that have consequences for syntax. However, I am not convinced that this is really accomplished in the current system, at least in the case of the classification of verbs. For example, it is unclear to me why we would need [IEPS], in contrast to features like [dynamic] or [B]. [+IEPS] is said not to imply either telicity or boundedness (p. 31, note 11). Instead, the feature [B] is used to distinguish between (verbs denoting) temporally punctual situations and temporally durative ones (both [+IEPS] and [-IEPS]). This distinction is reflected in the ability to combine with durative adverbials ('We walked / fell / *arrived for an hour'). Thus, [B] is a semantic feature which clearly has an impact on syntactic behaviour. However, as far as I can see, the same does not hold for [IEPS]. (1) and (2) are the examples Lieber gives as illustration of the different behaviour of [+IEPS] and [-IEPS] verbs. According to Lieber, (1) ''is strange because the [+IEPS] change of place verb 'descend' implies a PATH with non-equivalent initial and final points and a steady progression from one to the other'' (p. 31).

1. ?After having descended the ladder, Morgan found himself to be in exactly the same place he had started from. 2. After having walked for five hours, Daisy found herself to be in exactly the same place she had started from.

However, this is clearly a semantic failure. Syntactically, (1) is fine. So I cannot see why the PATH meaning is needed in the semantic skeleton of these verbs and why it does not suffice to represent it as part of the semantic body. [+IEPS] is also used for prepositions like 'into', 'onto', 'toward' etc. (pp. 101-108). I am not sure whether the PATH meaning is syntactically relevant or if instead the feature [B] could be used in order to capture their different syntactic behaviour, with the PATH meaning again encoded in the body.

In addition, there are a few minor points on the verb classification, for example why telicity is completely excluded from verb meaning in the simple lexicon (p.141). Obviously, telicity often only appears at higher levels of semantic composition, but I would argue that there are also verbs which are inherently telic, i.e. at the lexical level, like 'find' or 'arrive'.

To sum up, although I do not agree on all technical details, this monograph provides a very interesting and useful framework. Perhaps the greatest value of this book is that it shows that semantics is essential for a proper theory of word formation. To give just one example, the relisting-analysis for conversion is highly convincing because of its very simple semantic motivation. ''Morphology and Lexical Semantics'' is a highly valuable book, which is very stimulating for further discussion and research.

REFERENCES

Booij, Geert (1992) Compounding in Dutch. Rivista di Linguistica 4, 37-59.

Booij, Geert (2002) ''The Morphology of Dutch.'' Oxford: Oxford University Press.

Copestake, Ann & Ted Briscoe (1996) Semi-productive polysemy and sense extension. In: James Pustejovsky & Branimir Boguraev (eds.), ''Lexical Semantics: The Problem of Polysemy.'' Oxford: Clarendon Press, 15-68.

Jackendoff, Ray S. (1990) ''Semantic Structures.'' Cambridge, MA: MIT Press.

Lieber, Rochelle (1992) ''Deconstructing morphology: Word Formation in Syntactic Theory.'' Chicago: University of Chicago Press.

Lieber, Rochelle (2004) ''Morphology and Lexical Semantics.'' Cambridge: Cambridge University Press.

Lieber, Rochelle (2007) The category of roots and the roots of categories: what we learn from selection in derivation. Morphology (2006), 16/2: 247-272.

Lieber, Rochelle (2009) A lexical semantic approach to compounding. In: Rochelle Lieber & Pavol Stekauer (eds.), ''The Oxford Handbook of Compounding.'' Oxford: Oxford University Press, 78-104.

Pustejovsky, James (1995) ''The Generative Lexicon.'' Cambridge, MA: MIT Press.

Szymanek, Bogdan (1988) ''Categories and Categorization in Morphology.'' Lublin: Catholic University Press.

Wierzbicka, Anna (1996) ''Semantics: Primes and Universals.'' Oxford: Oxford University Press.

Barbara Schlücker is a lecturer in linguistics at the department of German and Dutch at the Freie Universität Berlin, Germany. She holds a doctoral degree in German linguistics (2006). Her research interests are semantics and pragmatics as well as morphology and syntax. Her current research project is on the semantics of nominal compounds and phrases in German.