2. Description: Composition of Individual Chapters.

In “Coordination in morphology and syntax: The case of copulative compounds” (chapter 2), Susan Olsen concentrates on the morphologically complex copulative compounds (such as bartender-psychologist). Copulative compounds are semantically unique structures in the sense that they express a coordinative rather than a subordinate relation between two (or more) constituents. Olsen shows that morphological copulative compounds are distinct from their syntactic counterparts, so-called coordinative appositions.

In support of the argument that different principles govern the syntactic coordinative apposition structures and morphological copulative compounds, Olsen proposes the Principle of Ontological Coherence (POC) responsible for determining the grammaticality of the latter (morphological) component of grammar. Olsen defines the POC: “a complex concept as the denotation of a morphological object that picks out a coherent individual from one of the domains of individuals” (p. 19). To clarify the true nature of the cognitive mechanisms involved in generating both determinative compounds—those characterized by a subordinate relation, such as computer monitor—and copulative com-pounds, Olsen proposes that copulative compounds are simply a type of determinative compound. The difference between the two types of compounds is derived by means of a compound template consisting of a grammatical level of lexical-semantic representation (termed “semantic form”, SF) and the level of conceptual structure (CS). The copulative interpretation is the most neutral and is obtained by default when the context does not provide an alternative meaning of the compound structure.

In “Aspect, infinitival complements, and evidentials” (chapter 3), Elly van Gelderen takes issue with Giorgi and Pianesi's (1997) analysis of English eventive infinitives lacking the -en ending as inherently perfective, unlike their counterparts in Dutch and German. The crux of Giorgi and Pianesi's argument is that these perceptive verb complements (PVC) carry a perfective feature in English (but not in modern Dutch or German). Van Gelderen argues that the loss of the verbal suffix -en does not coincide with the perfective interpretation of bare eventive infinitives. Van Gelderen proposes that the shift enabling bare eventive infinitives to be interpreted as perfective elements occurred through two diachronic changes. First, following the loss of the ge- prefix that previously marked perfectivity, the suffix -ing underwent reanalysis as the result of the checking relation between the verb and an aspect phrase (AspP) in the middle field. Second, a parametric shift in unmarked aspect (from unmarked imperfectivity to unmarked perfectivity) further solidified the former diachronic change.

According to van Gelderen, there are three kinds of perception verbs in English, hence three distinct variants of PVCs. To illustrate this point, van Gelderen (pp. 54–56) shows that three specific instances of the verb see are evidenced in English:

In 2a, we see an instance of an activity see, which is usually replaced by clearer activity verbs, such as watch. Example 2b contains a stative see, and 2c contains a modal see, which van Gelderen argues to have been grammaticalized to the status of an auxiliary/evidential. The key difference between the three types of see represented above reduces to their complement selections—the activity see in 2a selects a CP complement, the stative see in 2b selects an aspect (AspP^o) complement, and the modal see in 2c selects a vP complement. The impossibility of an imperfective reading in 2c is predicted under this analysis—the PVC selected by the evidential see does not select an AspP^o as its complement (that is, this is the only viable option for stative see in 2c).

Helen de Hoop's “The problem of unintelligibility” (chapter 4) is a unique contribution to this volume, since it is the only paper couched within a non-Chomskyan framework (i.e., Optimality Theory). A bidirectional version of Optimality Theory is employed by de Hoop to explore instances of unintelligibility, that is, when a well-formed syntactic expression fails to give rise to a felicitous interpretation, as in 3.

This version of Optimality Theory assumes that both the speaker and hearer are able to take into account each other's perspective. Occurrences of unintelligibility arise when an optimal interpretation of a syntactic structure is not present in the candidate set of potential interpretation outputs. De Hoop's solution to this theoretical issue is straightforward: extend the candidate set of possible rival interpretations to include contradictory statements. Furthermore, de Hoop posits five pragmatic-oriented ranked constraints that have the ability to map syntactic structure to semantic interpretation. For example, by introducing pragmatic-oriented constraints such as “Be Informative” and “Avoid Contradiction,” and ranking them in this order, sentences such as 3 above will receive a correct interpretation (that is, the number of female professors who are men with beards and glasses is greater than the number of female professors who are not).

Werner Abraham's “VP-internal subjects as ‘unaccusatives’: Burzio's ‘Object Account’ versus the ‘Perfectivity Account’” (chapter 5) is the second paper in Part I to address the relation between perfectivity and syntax. Abraham takes issue with unaccusativity and Burzio's Generalization—which links a verb's ability to assign accusative case with its ability to license an external argument—and hypothesizes that both can be understood within the concept of perfectivity. Abraham tests the definiteness effect—that is, the inability of a “strong” NP to show up in existential sentences—which is traditionally considered to be a hallmark empirical test in support of Burzio's Generalization (data from Abraham, p. 88):

The data in 4 support the claim that only ergative verbs, as in 4a, can appear in weak existential or presentational clauses. Accordingly, ergative verbs are unable to assign (structural) accusative case because these verbs do not license active functional projection specified for agreement (AgrOP). Abraham also adopts Diesing's (1992) Mapping Hypothesis, which states that definite arguments must move to the specifier position of an agreement projection prior to LF-interpretation. By contrast, indefinite objects are required to remain within the VP unless they receive a generic interpretation. Abraham argues that existential clauses lack an agreement phrase, which would explain the definiteness effect in the context of the Mapping Hypothesis framework. Abraham further argues that existential clauses do not function as optimal diagnostic tools in testing unaccusativity. As an alternative, he proposes that properties such as telicity, change of state, and the possibility of an agentive reading should be used as diagnostics for unaccusativity.

In “Either, both, and neither in coordinate structures” (chapter 6), Petra Hendricks suggests an alternative account of these three lexical items that contradicts standard assumptions. She proposes that either, both, and neither are focus particles quantifying over a set of relevant alternatives, rather than traditional conjuncts. Hendricks sets up two criteria by which she evaluates the lexical particles individually. First, she compares each lexical item with focus particles such as only with respect to distribution, interaction with sentential intonation, and contribution to the interpretation of the sentence. Second, she compares each, neither, and both with respect to their ability to c-command the element present in the first conjunct that receives contrastive stress. In this regard, these three traditional conjuncts strongly resemble focus particles. Based on these tests, Hendricks concludes that the same rules governing the usage of focus particles are also applicable to these three traditional conjuncts.

The paper “Information structure meets Minimalist syntax: On argument order and case morphology in Bavarian” (chapter 7), Helmut Weiß proposes an analysis of two types of short scrambling in double-object constructions in Middle Bavarian. The two distinct types of scrambling he examines are (i) the obligatory scrambling of a definite direct object over an indefinite object, and (ii) the optional object inversion involving focus (whereby the unfocused element precedes the focused element). Weiß contends that the first type of scrambling mentioned above—that is, the obligatory scrambling of a definite object over an indefinite object—is feature-driven. Contrary to other recent analyses of short scrambling in West Germanic (see, for example, Abraham and Molnárfi 2002, 2003 among others), Weiß asserts that morphological properties (features) of the scrambled argument are at work here. More specifically, he (p. 153) proposes the Principle of Strong Morphology as a potential explanation of the cross-linguistic variation between languages that allow scrambling (such as Bavarian) and those that do not (such as English).

According to this proposal, strong morphology makes overt movement for checking requirements superfluous. If morphological strength triggers syntactic movement, these two models are not completely identical. What this means for Weiß's proposal is that for a language exhibiting morphological Case (such as Bavarian), the Case feature can be checked after Spell-Out, and in a language lacking morphological Case (such as English) it must be checked prior to Spell-Out, that is, in the syntax.

Klaus von Heusinger's contribution to this volume “Focus particles, sentence meaning, and discourse structure” (chapter 8) attacks the traditional notion of parsing the sentence information structure into a propositional unit (old information) and focus (new information). Von Heusinger argues that such a view of information structure leads to several problems with the compositional process in complex cases. Von Heusinger maintains that information should be represented in two overlapping units, rather than in two disjunctive ones: the foreground unit denotes the entire sentence, while the background unit corresponds to the whole sentence minus the focused expression. To support his claim, von Heusinger shows how overlapping units of information structure provide a more parsimonious explanation of the semantic interpretation of adverbs of quantification. Consider 6 and 7 from von Heusinger (p. 170).

While both 6 and 7 compare two sets of time points or intervals for which a certain restriction holds, attaining all possible interpretations is only possible when the whole sentence is considered. The operator most for the expression usually compares two tests of time t. The first set is characterized by the background information, while the second is characterized by the whole sentence or the conjunction of focus and background. Von Heusinger (p. 170) provides the interpretation in 8.

The concept of Foreground-Background Semantics as outlined in this article presents a challenge for semantic theories of information structure, because it tries to integrate a wide range of phenomena. One prime goal of this proposal is to return information structure interpretation to its true status as a part of a larger discourse representation.

The final chapter of this volume, “On the interpretation of multiple negation in spoken and written Afrikaans” by László Molnárfi, deals with the phenomenon of negative concord in spoken and written Afrikaans. The possible readings of 9 are logical negation canceling (LNC) and negative concord (NC) (data from Molnárfi, p. 196).

In the spirit of Abraham 2000, this paper pursues the use of certain grammatical forms as specific means of oral communication employed to facilitate sentence parsing. Molnárfi takes issue with analyses of negative concord in West Germanic, which propose to account for these constructions in terms of a formal operator-licensing mechanism. According to Molnárfi, negative concord should be analyzed as an instance of top-down feature percolation, where silent copies of the first negator infiltrate all terminal nodes within the scope of negation. Accordingly, written and spoken Afrikaans differ with respect to the constraints that govern the morphological Spell-Out of the negation copies in both spoken and written language.

3. Critical Analysis.

As Fortin (2004) observes, the papers in this volume are more coherent in their approaches to the syntax-semantics and syntax-pragmatics interfaces than one might expect from such a wide-ranging volume. There are two papers (Abraham and van Gelderen) that discuss perfectivity; two cover focus particles (Hendriks and von Heusinger); and two that deal with definiteness (Abraham and Weiß). Rather than provide a critical analysis of all of the papers and general arguments presented in this volume, I address three topics relevant to my own research expertise. First, I discuss how a pragmatic filter may indeed exist prior to LF. Then I turn to Weiß's Principle of Strong Morphology, and show that although it is an attractive attempt to view scrambling in Middle Bavarian as determined by a morphosyntactic feature, problems still remain. Finally, I focus on Molnárfi's account of negative concord in Afrikaans, and the challenges that his theory of downward feature percolation present for standard assumptions about the model of the grammar in the minimalist program.

An important commonality expressed somewhat tacitly in de Hoop's and von Heusinger's articles is the potential introduction of pragmatic considerations filtering well-formed representations before they can be interpreted for semantic meaning (that is, at LF). De Hoop's introduction of pragmatic-oriented constraints such as “Be Informative” and “Avoid Contradiction” to ensure optimal mapping strategies of syntactic structure to LF, and von Heusinger's proposal of information structure (background and foreground units) suggest the possible presence of a pragmatic filtering component. This hypothesis, namely, that there appears to be some sort of pragmatic constraint shaping representations prior to their semantic interpretation, can perhaps be better understood through Game Theory. In terms of Game Theory, these constraints can be seen to exist in a symmetric game; a game where the payoffs for playing or using a particular strategy depend only on the other strategies employed, not on who is playing the game. This stands in contrast to games played in normal form, where it is generally assumed that the players involved act simultaneously without knowing the actions of others (an assumption that is clearly unwanted and hardly practical in any theory of discourse analysis). This is similar in spirit to de Hoop's idea of introducing a bi-directional version (which assumes that both the speaker and hearer are able to take into account each other's perspective) of Optimality Theory. From the standpoint of linguistic economy, the introduction of Game Theory to function as a finite set of pragmatic constraints serves as a way to restrict the generation of an infinite number of candidates by GEN to be evaluated at LF.

Weiß's approach to obligatory scrambling in Middle Bavarian, understood through his Principle of Morphological Strength, supports the general minimalist thesis that pragmatic considerations neither govern nor affect syntactic operations. Rather, pragmatic well-formedness is viewed as a “side effect” of syntactic operations. I find Weiß's proposal in its current state problematic both on conceptual and empirical grounds.

First, addressing the empirical grounds, evidence showing that ubiquitous A-movement properties are present in obligatory scrambling is required to argue that some sort of morphological feature is involved in scrambling. Work by Webelhuth (1992) and others clearly shows that an isomorphic reflex displaying traditional A- or A-bar reflexes of middle field scrambling in West Germanic appears improbable. If A-movement characteristics are manifest in Middle Bavarian scrambling, then such data would be paramount in supporting Weiß' theoretical underpinnings.

From a conceptual standpoint, it remains unclear how the Principle of Morphological Strength is superior to earlier instantiations of minimalist theory in accounting for the overt versus covert movement dichotomy. If strong features (in this instance Case) can be checked after Spell-Out, does this require the covert movement of the entire constituent or just the strong feature in question? Furthermore, if we adopt a multiple Spell-Out model of the grammar, these Case features would have to be expelled to the edge of the vP to ensure their survival for operations that will take place in the higher strong phase CP. Therefore, although Weiß's proposal offers an interesting perspective on the scrambling puzzle in West Germanic, in many ways its application is ineffective as it fails to eliminate problematic issues with this particular movement type that have plagued syntactic theory since its conception.

What is unique about Molnárfi's contribution to this volume is his assertion that the realization of NC in Afrikaans is due to a top-down percolation of a negation feature [Neg] to the rightmost terminal node in the syntactic hierarchy. The presence of the first negation particle (Neg₁) and the rightmost particle (Neg₂) serve as the boundaries of the scope domain of Neg₁. As somewhat of a radical departure from standard assumptions that maintain that Neg₁ is the syntactic head of the maximal projection NegP, Molnárfi proposes that this element is the morphological spell-out of the feature [Neg]. More important for Molnárfi's argument is his assumption that the second negative element (Neg₂) does not assume the independent status of a terminal node in syntactic structure (Molnárfi, pp. 205–206).

As illustrated in sentence 10a, the second negator can freely be associated with different terminal nodes, depending on which node is most deeply embedded. Phonologically weak copies of the second negator easily cliticize on such verbal, adjectival, and nominal heads. In 10b, we see that the two negation copies cannot be directly adjacent to one another. Biberauer (2006) interprets the ungrammaticality of such constructions as potential “syntactic OCP” (Obligatory Contour Principle) effects akin to autosegmental phonology. Lastly, as demonstrated in 10c, the lack of syntactic autonomy of the second negator also explains why negation in Afrikaans can be doubled (copied) beyond the sentence boundary. These examples challenge long-held standard assumptions about the model of grammar as interpreted under current minimalist desiderata.

As stated by Molnárfi himself, the percolation of features in a top-down fashion is difficult to account for in a phase-based model of the grammar employing multiple Spell-Out. Example 10c is difficult, if not impossible, to account for in current minimalist theory. Two recent papers addressing the distinction between syntactic ontological commitments (that is, phases and prolific domains) and phonological units have been written by Fuß (forthcoming) and Grohmann and Putnam (forthcoming). In the latter paper, the authors propose that PF and LF are assembled simultaneously in a bi-directional fashion, with PF being dynamically composed top-down and LF being assembled in the traditional bottom-up approach. In this regard, Molnárfi's paper contains data that may benefit further studies on the model of the grammar as well as those centered on negative concord structures.

Although certain minor errors exist in this volume (for example, “suboordonated” for “subordinated” on page 221, “prefigation” for “prefixation” on page 101 and some incorrectly numbered footnotes and unclear citations), this collection of articles is a valuable contribution to the field of pragmatics-syntax and semantics-syntax interface. Both the empirical data and theoretical assumptions put forward in this work supply unique perspectives on current problematic issues involving syntactic interface research.