1. The Typology.

Line Mikkelsen's monograph seeks the correct semantic, structural, derivational, and discourse-grammatical analyses of copular clauses. As the subtitle indicates, Mikkelsen presents a typology of three distinct types of copular clauses. The following examples are from page 1:

Employing data from English and Danish, Mikkelsen explores a number of semantic and syntactic differences between these three types of copular clauses. Predicational clauses can be viewed as canonical.¹

During her exploration of these clause types, Mikkelsen makes a number of significant claims concerning this typology. Five of these claims are presented in 2 for orientation.

Regarding the first four claims, Mikkelsen's reasoning is sound. In my view, however, the fifth claim is incorrect; names and personal pronouns can be property-denoting in certain contexts, as discussed below.

2. Structure.

Chapter 2 argues against Heggie's (1988a, 1988b) analysis of specificational clauses. Heggie proposes that specificational clauses are derived from predicational clauses via topicalization of the predicate DP and subsequent leftward movement of the copula. While Mikkelsen shows that Heggie's proposal is incorrect for English, she demonstrates that it is correct for certain clauses in Danish.

The sentence in 3 is ambiguous between a predicate topicalization reading and a specificational reading. The difference is non-truth conditional and hence nuanced. Mikkelsen uses a number of permutation diagnostics involving the negation ikke to tease the derivations apart.

The predicational reading has the negation following the subject Minna. In contrast, the specificational reading has the negation preceding the predicate expression Minna. Thus, in 4a den højeste spiller is the predicate DP, whereas in 4b it is the subject DP. The distinction becomes apparent when one considers the behavior of personal pronouns.

When the personal pronoun hun/hende ‘she/her’ functions as the subject as in 5a, the nominative form hun ‘she’ must appear. When the pronoun is in the predicate, as in 5b, the accusative hende ‘her’ must appear. Accusative is the default case for both Danish and English.

Chapter 3 briefly examines other proposals concerning specificational clauses (Moro 1997, Heycock and Kroch 1999, and Rothstein 2001). In view of the observations in later chapters, Mikkelsen opts for an analysis of specificational clauses that is quite similar to Moro's (1997) Government and Binding account. The central idea of Moro's analysis is that both expressions, that is, the subject DP and the predicate DP, are generated inside the VP. In standard predicational clauses, the referential subject DP raises to Spec-IP, and the predicate DP remains in the VP. In specificational clauses the situation is reversed; it is the predicate DP that raises to Spec-IP. The structure in 6 is from p. 42.

This tree illustrates the manner in which the predicate DP raises out of the small clause in the VP to the Spec-IP position, resulting in a specificational clause. This analysis represents Mikkelsen's major claim concerning the derivation of specificational clauses. Chapter 9 improves on this account, adapting the analysis to the Minimalist Program in terms of feature checking.

3. Meaning.

Chapter 4 sets up a typology of semantic types for DPs that is consistent in part with Partee's (1987) type-theoretic account of noun phrase interpretation. This typology acknowledges two major types of DPs: referential ones and property-denoting ones. According to Mikkelsen, the distribution of these DP types in the three types of clauses is as in 7 (where the term property-denoting is synonymous with predicative).

Chapter 5 explores the contribution of the subject DP to the meaning of a specificational clause. Numerous diagnostics are employed to tease apart the varying semantic types of the DPs, such as tag questions, left dislocation, relative clauses, question-answer pairs, grammatical gender in Danish, etc. The pronouns in the tag questions in 8 illustrate that the semantic traits of a subject DP vary dramatically according to whether the clause is predicational or specificational.

In 8, the pronouns in the subject position are anaphoric. The noteworthy aspect of specificational clauses is that the anaphoric connection to the subject is established by property-denoting pronouns (it, that), rather than by referential pronouns (he, she). Thus, the pronoun she in 8a refers to the individual denoted by Molly, whereas the pronoun it in 8b refers to the property denoted by the tallest girl.

Chapter 6 investigates the semantic contribution of the predicate DP to the meaning of specificational clauses. Mikkelsen's claim is that the predicate DP of a specificational clause is referential, and that certain expressions are obligatorily property-denoting. Therefore, if an expression occurs in the predicate which is incapable of being referential in that position (for example, adjective, bare NP), then the subject DP is necessarily referential and the specificational reading is no longer available. This situation is illustrated in 9a, which contains a bare NP and a tag question. If, however, the NP is introduced by a determiner, which means it is a DP, the specificational reading can obtain, as in 9b.

Since it is not possible in the tag in 9a, the subject DP is referential, which means that the predicate DP is non-referential and property-denoting. This situation obtains because Mayor of Santa Cruz is obligatorily non-referential. The sentence in 9b is ambiguous between the predicational and specificational readings. Under the predicational reading, the property being the mayor of Santa Cruz is predicated of the winner. Under the specificational reading, the property being the winner is predicated of the Mayor of Santa Cruz.

Chapter 7 examines the semantics of DP types in order to make sense of their distribution in copular clauses. Mikkelsen discusses the fact that certain DPs may not appear as subjects of specificational clauses, as the example in 10 involving a quantified expression illustrates.

According to Mikkelsens's account of such cases, quantified expressions such as both actresses may not be property-denoting, which means they cannot appear as the subject of a specificational clause. This explanation receives support from the example in 11, where the quantified expression cannot appear in the predicate of a predicational clause.

Mikkelsen extends the claim concerning quantified expressions to personal pronouns and names. In other words, she also posits that personal pronouns and names may not be property-denoting, which means they cannot appear as subjects in specificational clauses, as shown in 12.

I take issue with the second part of this claim in section 6 below, where I demonstrate that names and personal pronouns can be property-denoting in certain contexts, contrary to Mikkelsen's claim.

Mikkelsen also extends the analysis of specificational clauses to truncated clefts, such as That's/it's Susan. She demonstrates convincingly that these clauses do not involve ellipsis, but are rather a particular case of the specificational clause.

4. Use.

Chapter 8 explores the discourse notion of topic in the context of specificational clauses. The core phenomenon at issue is presented on page 133.

Mikkelsen cites Halliday (1967) as the first to have discussed these data. The difficulty they pose is seen in the awkwardness of the specificational answer to the question What is John?

Building on the work of Birner (1996), Mikkelsen points to the topic role in such cases. Topic is understood mainly in terms of the distinction between discourse-old and discourse-new material. The unique trait of specificational clauses in this regard is that they must be licensed by the preceding discourse (or situational context) that establishes the subject DP as discourse-old information. Under this approach, the specificational clause in 13b is fine because the question has established the subject DP the winner as discourse-old material. In contrast, the specificational clause in 14b is awkward because the question has not established the subject DP as discourse-old. In other words, discourse-new material cannot appear as the subject DP of a specificational copular clause.

This analysis of the distribution of specificational clauses in terms of the distinction between discourse-old and discourse-new material relies on the assumption that the specificational copular clause is “not normal,” whereas the predicational copular clause is “normal.” The standard distribution of semantic information, that is, the referential DP as subject and the property-denoting DP as the predicate, can take place regardless of the distribution of discourse-old and discourse-new material. This trait is what makes such clauses “normal.” The unusual distribution of the semantic information in specificational clauses, that is, the property-denoting DP as subject and the referential DP as the predicate DP, is what makes such clauses “not normal.”

Chapter 9 integrates the topic role into a Minimalist analysis by acknowledging a topic feature. Mikkelsen proposes that the subject DP of a specificational clause enters the derivation as the predicate DP with the interpretable feature [TOP], and that the functional category T appears with the unvalued feature [uTOP]. In order to provide this unvalued feature with a value, the predicate DP rises out of the VP to T to check off the [uTOP] feature on T. In so doing, it becomes the subject, as shown in 15 (adapted from page 172).

The categories vb and Pred are not important for the illustration. What is important is the manner in which DPpred enters the derivation in the complement of PredP and then rises to Spec-TP to check the unvalued feature [uTOP]. The derivation of predicational clauses is different in this regard. Instead of the DPpred moving, they have the DPref moving to Spec-TP.

5. Merits.

Mikkelsen's presentation of copular clauses is accessible to a wide audience. It is not overburdened with too many formalisms and irrelevant subject matter. If one overlooks the overabundance of organizational statements using first person references, the prose is smooth and easy to follow. Reading and studying the book has been enjoyable.

More importantly, Mikkelsen's analysis is convincing in many respects. Of the five claims mentioned in the introduction, four are sound. In this regard, the distinction between property-denoting and referential expressions is the key to understanding the semantics and syntax of specificational clauses. The subject of the specificational clause is property-denoting, whereas the predicate DP thereof is referential. The particular distribution of this semantic information explains our intuition about such clauses, namely that they are unusual.

6. Problems.

There are also problems with Mikkelsen's account, and the notes on pages 72 and 95 bear witness to these. In each case, Mikkelsen adds the note in order to hedge a significant claim made in the body of the text, claims that do not hold up to scrutiny.

The first shortcoming is in Mikkelsen's analysis of VP ellipsis (VPE) (pp. 99–107). Mikkelsen observes that the following sentences are quite marginal, if not entirely disallowed (p. 100):

Based on such data, Mikkelsen develops the argument that VPE is not possible in such cases because the predicate DPs, that is, Betty in 16a and Harry in 16b, are non-predicative, that is, they are referential. This motivates Mikkelsen to maintain tentatively that VPE cannot target referential DPs.

This claim is odd in view of Mikkelsen's heavy reliance on tag questions with specificational clauses. Tag questions are a specialized manifestation of VPE, as Mikkelsen herself acknowledges in note 10 on page 72. For instance, in a tag question with a specificational clause, as in 17a, VPE targets the referential predicate DP Susan. Furthermore, it is not difficult to produce similar instances of VPE that do not involve a tag question at all (see 17b).

In light of such data, Mikkelsen's claim that VPE may not target a referential DP is unconvincing.

The second area where Mikkelsen's account is lacking—the treatment of equative clauses—is more serious for the overall analysis. The discussion throughout the book mainly focuses on the differences between predicational and specificational clauses, whereas equative clauses receive considerably less attention. Then in the conclusion (p. 193), Mikkelsen suggests that it might be possible to collapse predicational and equative clauses into a single category. It is unfortunate that Mikkelsen did not pursue this possibility from the start.

The archetypical equative clause—4.20 on page 58—behaves like a predicational clause with respect to the diagnostics used by Mikkelsen.

However, another type of equative clause behaves either like a predicational clause or like a specificational clause.

In 20, both the property-denoting it and the referential she are possible in the tag.

Mikkelsen's claim about the nature of names (Molly, McGovern) bears on the correct analysis of these data (see claim 5 in section 1). She maintains that names (and personal pronouns) are obligatorily referential (p. 95), yet she also acknowledges that epistemological factors can allow a name to be property-denoting (pp. 51–53, p. 73, p. 95 note 1). Mikkelsen thus makes a significant claim that is contradicted both by certain data, such as 5.36 on page 73, and by her own acknowledgment in a previous section, that is, the discussion on pages 51–53.

If one abandons the assumption that names and pronouns must be referential, allowing them to be property-denoting in certain specialized contexts, the justification for the equative category disappears. Sentences 18b and 19b are purely predicational, the names being property-denoting. However, sentences such as 18 are ambiguous between the predicational and specificational readings. The ambiguity arises due to the fact that both the name and the pronoun have the option to be property-denoting or referential.

Sentence 21a can be uttered in a context where the listener does not know who Molly Jacobson is and the speaker points her out in a crowd. By comparison, sentence 21b could be uttered in a context where the listener knows who Molly Jacobson is, but can't recognize her in a crowd (at a costume party for instance), so the speaker points her out.

Given this analysis, Mikkelsen's claim that names and pronouns cannot be property-denoting does not hold up to scrutiny. Since such expressions can be property-denoting in certain specialized contexts, the justification for the equative category—namely that they are unique insofar as both DPs are referential—disappears. All copular clauses obligatorily involve one property-denoting expression and one referential expression, and a constellation where both DPs are either referential or property-denoting is impossible. This analysis collapses the three-way distinction into a two-way distinction, significantly simplifying the typology.

7. Further research.

Our knowledge of copular clauses would be augmented considerably via cross-language comparisons. Mikkelsen's account draws on data from English and Danish. These two languages are similar insofar as both clearly have specificational clauses. With other languages, however, it is not immediately clear whether specificational clauses exist at all. Indeed, the data in 22 support the view that specificational clauses do not exist in German, Spanish, and Czech.

In each case, the copula must agree with the pronoun, and not with the pre-verbal DP. In English, the agreement relation is reversed, as in The best player is/*are you. Thus, if the examples in 22 contained specificational clauses, agreement would obtain with the pre-verbal DP. These data are consistent with a predicate topicalization analysis of the preverbal DPs.

If, as I suggest, specificational copular clauses do not exist in languages such as German, Spanish, and Czech, then copular clauses in general qualify as a significant area of inquiry for language typology. Languages can be classified according to whether or not they allow specificational clauses. It is likely that inflectionally rich languages, such as German and Czech, lack specificational clauses, while inflectionally poor languages, such as English and Danish, allow them.