This book investigates the polysemy of deverbal nominalizations formed with the suffix -ment in present-day English, by closely addressing two semantic verb classes from which such nominalizations are productively built: change-of-state and psychological verbs. Adopting the theoretical framework of Frame Semantics, the author argues that the interpretation of such derivations involves a constant interplay of the semantic properties of the affix with those provided by the verbal base.

The book consists of eight chapters, including the introduction and conclusion. The introduction lays out the challenge that affix polysemy poses for linguistic theory and the desideratum that such a theory should account for both the contribution of the base and that of the affix in building the meaning of a derived word. Unlike in some recent approaches (e.g. Lieber Reference Lieber2016), which focus on affixes leaving aside the base, Kawaletz aims to model affix polysemy as a function of the array of semantic elements that the base makes available and the predisposition of the affix to select only some of these within a broad notion of compositionality. The interaction between the base and the affix, relying on a decompositional analysis of derived words, becomes crucial for this modeling. In this respect, Kawaletz offers a finer-grained analysis for affix polysemy than usual in purely descriptive or underspecified approaches, such as in Bauer et al. (Reference Bauer, Lieber and Plag2013) or in Pustejovsky (Reference Pustejovsky1998) and Lieber (Reference Lieber2004, Reference Lieber2016). The framework chosen for this modeling is that of Frame Semantics (Barsalou Reference Barsalou1992a, Reference Barsalou, Lehrer and Kittay1992b; Petersen Reference Petersen, Skilters, Toccafondi and Stemberger2007; Löbner Reference Löbner2013), which makes crucial use of frames, i.e. attribute-value structures used as fundamental knowledge representations, comparable to those from Head-driven Phrase Structure Grammar (Pollard & Sag Reference Pollard and Sag1994) or Lexical Functional Grammar (Bresnan Reference Bresnan1982).

For her close analysis of -ment nominalizations, the author aims to provide an answer to the following three research questions: (1) Which readings are possible in newly formed -ment derivatives? (2) What are the semantic contributions of the base and of the affix? (3) How can this be modeled in a frame-semantic approach? To this end, she uses a dataset of 40 -ment neologisms extracted from the Oxford English Dictionary (OED) and the Corpus of Contemporary American English (COCA), which are built on change-of-state and psychological verbs, and evaluates their possible readings on the basis of 369 corpus attested examples, collected in a purposeful sampling approach.

Chapter 2 ‘Frames’ (pp. 7–22) introduces Frame Theory, the theoretical background against which the author is building her analysis of -ment nominalizations and their base verbs. After a brief presentation of the evolution of the theory, Kawaletz describes the model assumed in the context of her project building on Barsalou (Reference Barsalou1992a, Reference Barsalou, Lehrer and Kittay1992b), with clarifications on the representation of frames, attributes, uniqueness conditions, constraints and type signatures. Furthermore, she introduces generalized lexical event frames, which generalize over sets of lexemes that share a semantic core of attributes (e.g. verbs of directed motion as involving an actor and a goal), by contrast to instantiated frames, which represent particular tokens in context. The last section of the chapter deals with how derivation may be modeled with frames, namely as frame unification or as referential shift. In unification, the frames of the parts are conjoined to form the whole. Referential shift builds on metonymy, such that the frame's reference is shifted to the value of one of its attributes, and this is what Kawaletz uses in her approach. For instance, from the base verb walk, one can derive the noun walk, whose frame has the same reference as the verb on the meaning ‘act of walking’, but not on the meaning ‘route for walking’; here, the frame reference gets identified with the value of the verb's Path attribute. The modeling of these referential shifts is done via lexical rules, which constrain for instance a suffix like -er to get the reference of the Agent (‘the one who walks’) or of the Instrument of the base verb (‘walking aid’).

Chapter 3 ‘Methodology’ (pp. 23–46) explains in detail how the author collected, filtered and semantically classified the dataset. To capture the meaning on which the suffix -ment is productive in present-day English, the author collected a total of 40 different -ment neologisms from the OED and from COCA. Some of these were extracted from the OED as deverbal formations first attested starting with 1900. This list was complemented by hapax legomena from COCA, which were then grouped according to the semantic classes of their base verbs. For this, the verb classification from VerbNet (Kipper-Schuler Reference Kipper-Schuler2005) was employed, and the two richest verb classes were selected: change-of-state and psychological verbs. After further revisions and additions of data, the author put together a list of 40 -ment nominalizations (18 for change of state and 22 for psych verbs), most of which are considered neologisms. To collect as much possible information on their polysemy, Kawaletz searched these items in further corpora and managed to collect a sample of 369 tokens (192 for change-of- state nouns and 177 for psych nouns) for the 40 types of -ment nominalizations. To classify the meaning of these nouns, two large semantic categories were employed: participants and events. For the former, the core roles are Agent, Stimulus, Patient, Experiencer, Instrument and Result, which rely on the VerbNet classification. For the latter, Kawaletz offers an elaborate taxonomy of event types as covered by the two verb classes, which is explained and revised in the subsequent two chapters. For now, I mention the first level of hyponyms to event (as in eventuality), which are action, state, change-of-state, causation and psych-reaction, most of which have further hyponyms. The 369 tokens of nominalizations were annotated by three trained linguists, such that at least two agreed on the identified reading.

Chapter 4 ‘Change-of-state verb bases’ (pp. 47–107) handles -ment nominalizations built on change-of-state (COS) verbs. This chapter is split in two parts: ‘The semantics of COS verbs’ and ‘The semantics of COS nouns’. The first offers an overview of the previous literature on COS verbs as relevant for the Frame Semantics formalization, the main subclasses of such verbs and a formalization in terms of frames. The author classifies the COS bases of her -ment nominals in three subclasses: causative-only COS (abridge, bedraggle, befoul, besmirch, debauch, embitter and uplift), causative/inchoative COS (congeal, decenter, diminish, disband, discolor, disperse, increase, progress and worsen) and causative/inchoative reversible COS (embrittle and unfold). The first class receives a change-of-state causation frame type, which involves Agent, Patient and Instrument roles and two subevents: a Cause taking over the three roles and an Effect (with a Patient role and a Result state). Causative/inchoative COS verbs receive a frame in which both the full change-of-state causation and the simple Effect (or change-of-state) subevent embedded in it are available. Causative/inchoative reversible COS verbs receive an additional specification for the physical state of the Patient and the Instrument as solid, and for the Result state as involving the special type of state having-form.

The second part of chapter 4 is dedicated to -ment nouns built on COS verbs with an overview of their possible readings and the frame semantic formalization of these. The third section is the most interesting one for lexicalist semanticists of different theoretical orientations, as it addresses the possible readings that such nouns may realize, at least in the dataset considered here. What caught my attention in this section is the discussion on Patients, Results and Implicit products and especially the observation that if the event involves some Result or Implicit product, it will rather be this that the -ment noun will express, and not the Patient of the event. Moreover, -ment will express either the Result or the Patient, never both. The fourth and last section offers a formalization of these readings with frames, in which the author revises the frames and the type hierarchy of COS verbs presented in the first part of the chapter. This section would be most appealing to researchers who work with frames and are interested in the technical details of such an implementation.

Chapter 5 ‘Psych verb bases’ (pp. 109–56) is dedicated to -ment nominalizations derived from psych verbs. The author identifies her psych verb bases as belonging to two of Levin's (Reference Levin1993) four psych verb classes, namely, amuse verbs and marvel verbs. The former are transitive object experiencer (OE) verbs, and the latter are intransitive subject experiencer (SE) verbs with a PP. After a brief overview on the various theoretical perspectives on OE and SE verbs, the author follows aspectual tests from Van Valin (Reference Van Valin2005) and determines that the relevant OE verb bases in her dataset are either causative accomplishments, which involve a COS subevent (see endull, enrage, soothe and uplift) or causative states, which involve a state subevent instead (see abash, affright, annoy, bemuse, bumfuzzle, confound, convince, dishearten, dumbfound, enrapture, nonplus, perturb, reassure, stagger, upset and worry). For the SE verbs she distinguishes between the simple activity muse over and the simple states approve of and worry about. Building on previous literature, the author then proposes a formalization with five different frames for the OE verbs and three for the SE verbs. These types differ in terms of event structure (mono- vs. bi-eventive), their attributes (e.g. Cause vs. Explanation; with or without Agent, Instrument or Actor) or the values of these attributes. In view of the readings that the corresponding -ment nominalizations may instantiate, the author singles out the frame types psych-state causation and change-of-psych-state causation as relevant for OE verbs/nouns, and experiencer psych-action and psych-state for SE verbs/nouns. All of these may be realized as transpositional readings in nominalizations, while nominalizations are also argued to be able to realize only a subevent of Cause (as event) or Effect (as psych-state or change-of-psych-state) for OE verbs, as well as the Stimulus participant for both SE and OE verbs. At the end of chapter 5, the author offers a frame formalization of the nominalization types identified in her dataset. This leads to a revision of the type hierarchy for participants and for event types (pp. 151–3). The readings of the nominalizations are obtained by means of lexical rules, which take as input the frame of the verb and as output the targeted readings in relation to the semantics of the base.

Chapter 6 ‘Gaps and ambiguity’ (pp. 157–71) discusses gaps and ambiguity with respect to the readings of -ment nominalizations. Here, the readings of Instrument, Causer and Causing-event stand out as showing most gaps for COS nominalizations, and Stimulus and Causing-event for psych nominalizations. The author groups these under the larger category of Originator and argues that the fewer attestations may be due to the competition with other subject-oriented suffixes such as -er and -ant. These readings also turn out to be the most ambiguous, as the context often cannot disambiguate between them.

Chapter 7 ‘Discussion’ (pp. 173–82) summarizes the results of the previous three chapters in relation to the three research questions proposed at the beginning and some methodological issues. In answer to the first question, regarding what readings are available for -ment nominalizations, Kawaletz sums up ten types of eventuality readings (transpositions of full verbal event structures or shifts to subevents) and six for participants, most of which represent finer distinctions compared to the ones proposed in previous literature. An important observation is that [+animate] participants are not possible, while Location and Explicit product are not attested for these two classes of verbs, because they do not encode such participants in their event structure. These readings also provide an insight into the second question, concerning the contribution of the base and that of the suffix. On the one hand, the event readings can be traced back to the base verbs, while the lack of Location and Explicit product readings can be directly related to the non-existence of such participants in the lexical semantics of these verbs. On the other hand, the non-existence of Agent, [+animate] Patient and Experiencer readings is related to the restriction of -ment as not realizing [+animate] readings. For the third question, on the modeling in Frame Semantics, Kawaletz argues that the most feasible approach is to model the derivation of each individual reading from verbs to nouns in one lexical formation rule, which expresses the relation between the frame of the verb and that of the nominalization. The possible readings for a -ment noun are determined by the interplay between the type signature and the inheritance hierarchy for -ment, yielding a reading when no incompatibilities arise: e.g. we may find a shift from the psych verb annoy to a Result-state reading of annoyment, because the attribute Result-state is compatible both with psych verbs and -ment. However, an Experiencer reading is not possible, as the corresponding attribute requires [+animate] entities, which conflicts with the [-animate] requirement of -ment.

Chapter 8 ‘Conclusions and outlook’ (pp. 183–6) offers a summary of the results and a list of desiderata for future research. Among the latter, an extension to bases of other semantic classes and/or categories and to other derivational processes would help to identify possible universal principles in derivation, while supplementing the corpus-based research offered here with various computational tools would lead to more reliable results. A final thorny question is how to distinguish polysemy from context-induced coercion, and here experimental methods are suggested, which would again help to refine the systematic patterns of nominalizations from idiosyncrasies.

The laborious investigation of neologism data that Kawaletz carried out strikes me as very useful for the study of affix polysemy, as one can also evaluate now against the background she established the readings of more established -ment nominalizations, to possibly find out what is systematic and what is idiosyncratic in the meaning of this suffix. The question of possible frequency effects that the author raises in chapter 6 when discussing the gaps related to Originator readings is a valid one and could be further tested with more established nominalizations. However, it is also interesting that the gaps in the data are so few and precisely related to this thematic role, and I wonder whether this observation could be used to answer more general questions about the nominalization process and how metonymic shift happens. In this book all participant readings with action nominalization suffixes are considered metonymic shifts from the event to its participant. However, not all of these readings behave similarly: while some are recoverable from the verb itself, others are not. Thus, result readings are found in verbal forms (e.g. in past participles), and one may argue that such nominalizations are transpositions from the meaning that some form of the verb may realize. Possible product readings would then be metonymic shifts from the transpositional result nominalization and not directly from the event's participant. As the author correctly points out, Originator readings systematically come with the -er suffix, they are never expressed by verb forms, and the fact that they present gaps for -ment could be an indicator that they do not originate in a direct transposition/derivation from the verb's meaning but come about by means of some more idiosyncratic metonymic shift after the derived noun has been formed.

A related question arises as to the inchoative readings (what the author calls change-of-state readings) of -ment nominalizations based on COS verbs. The book predicts (and even claims) that the inchoative subevent could be nominalized even when there is no corresponding inchoative reading of the verb. I found the data and the discussion on pages 76–7 insufficiently clear or convincing. Most importantly, I consider this a very important question for the study of polysemy, which deserves more attention, as highlighted above. Can -ment indeed nominalize any subpart of the event structure, independently of whether the verb expresses it? I think that a positive answer to this question, as pursued in this book, would lead to much more ambiguity than desirable in our theory. Only a close scrutiny of the empirical facts could offer an answer. I would expect metonymic shifts to be less systematic/productive than transpositions and to happen at the level of the nominalizations (as also in Bierwisch Reference Bierwisch and Motsch1989, Reference Bierwisch, Giannakidou and Rathert2009), and not in the derivational process from verb to noun, as assumed in the Frame Semantics theory developed in this book. But this certainly needs further investigation.

As mentioned in relation to chapter 4, I found the generalization that -ment expresses either the Result or the Patient of a COS verb but never both very interesting and intriguing. A question that arises here is whether the realization of Patient vs. Result nominalizations may lead to a further distinction among these verbs, and whether the contrast between manner and result verbs introduced by Rappaport Hovav & Levin (Reference Rappaport Hovav, Levin, Butt and Geuder1998, and subsequent work) and further refined by Beavers & Koontz-Garboden (Reference Beavers and Koontz-Garboden2020) might have a bearing on this discussion. In general, at a theoretical level, I highly appreciated the decompositional approach to event structure pursued in the book, but I was surprised by the fact that current semantic insights from Beavers & Koontz-Garboden's (and even from Rappaport Hovav and Levin's more recent works on COS verbs) have not been considered for the discussion. Consulting this literature would have prevented the author, for instance, from categorizing inchoative COS readings as simple events, since such events have also been shown to be complex (cf. both repetitive and restitutive readings of the adverb again in The door opened again).

Let me conclude by emphasizing once again how welcome this systematic and empirically oriented research on the polysemy of the nominalizing suffix -ment is for future work in the study of polysemy in nominalizations and for a theory of derivational morphology. One important aspect is the analysis of neologisms, which are often challenging, given their unsettled meanings. In this book, the author made a great contribution for the suffix -ment. The decompositional analysis of the meanings of nominalizations has the merit of offering the possibility of direct comparison between the frame-semantic implementation provided in the book and other formal accounts on the event structure of both verbs and nominalizations. In this respect, the author offered a felicitously simplified and extensive picture of the theoretical debate. The implementation in Frame Semantics is well organized and easy to follow even for non-specialists and, although mostly descriptive, it makes clear predictions that could be further tested in future research. Finally, I believe that the comprehensive picture of the different types of readings that -ment may realize, the gaps and the systematicity it exhibits, has the potential to incite more targeted and groundbreaking research on what may be the source of semantic (ir)regularities, towards a finer-grained theory of nominalizations.