The main goal of The Clause-typing System of Plains Cree is to support the claim that there are two types of clauses clearly identified in Plains Cree (a Canadian indigenous language of the Algonquian family): anaphoric and indexical. This proposal makes predictions about the distribution of clauses whose interpretations are dependent on structural embedding or previous discourse in contrast to those whose interpretations are only dependent on the speech situation. An interesting parallel is drawn by the author, Claire Cook, between expressions governed by Binding Theory and the clause types she proposes: anaphoric clauses are akin to pronominals that get their interpretation from the context or antecedent, and indexical clauses are akin to R-expressions that have a fully independent interpretation and cannot be governed by an antecedent-like element. Plains Cree proves to be a robust testing ground since the clause types under discussion are encoded in morphological paradigms and exhibit other features (such as evidential interpretation and differential subject marking) that allow for the determination of clause types.

This book is organized as follows. Chapter 1 introduces the central concept that there is a division between clause types that anchor to the speech situation (indexical) and those that do not, identified as the elsewhere case (anaphoric). Chapter 2 discusses the morphological distinction between indexical and anaphoric clauses, claimed to line up with the morphology of the independent and conjunct verbal orders from the Algonquian literature. Chapter 3 looks at indexical clauses in the independent order that cannot be c-commanded or contain externally licensed anaphoric elements, and that can only be evaluated deictically (i.e. with respect to the speech situation). Chapter 4 then discusses anaphoric clauses in the conjunct order that conversely must be c-commanded or preceded by an element acting as an antecedent, and that are evaluated with respect to the situation specified by the antecedent. Chapter 5 considers the structural relationships between anaphoric clauses and their antecedents, arguing that chained clauses are only governed by precedence (e.g. in the discourse), but adjoined and argument clauses can be c-commanded by their antecedent. Chapter 6 goes into the specifics of the distribution of different conjunct clauses in Plains Cree, for instance, determining which indicate a presupposition, or appear in the context of certain aspectual or modal properties. Chapter 7 briefly considers remaining questions, in particular what might be expected in other Algonquian languages, and concludes the book.

Chapter 2, ‘Mapping indexical and anaphoric CPs onto Plains Cree’s morphosyntax’, introduces the distinction between the independent and conjunct orders that identify verbal paradigms. Cook argues that the independent realizes indexical clauses, and the conjunct is the elsewhere case found with anaphoric clauses. The independent order appears in only (but not all) matrix clauses and uniquely exhibits proclitics agreeing with the phi-features of speech act participants, as well as a third person suffix $w$ and a particular set of ‘right-edge morphemes’. The conjunct order that appears in any subordinated clause, and certain matrix clauses, utilizes a distinct set of right-edge morphemes (e.g. agreement suffixes) and distinct left-edge clause-typing morphology. As sketched in (1), indexical clauses are marked by a ‘speech situation variable’ $s_{\text{0}}$, and anaphoric clauses have an ‘anaphoric variable’ $s$ on CP.

Chapter 3, ‘Indexical clauses: Plains Cree’s independent order’, looks specifically at the syntactic and semantic properties of indexical clauses instantiated by the independent order. Cook posits an anti-c-command condition for indexical clauses, and shows that indexical clauses cannot be embedded by verbs or subordinated by particles (50), but must remain ‘free’. Further, evidence is provided that indexical clauses cannot contain proforms (e.g. temporal or locative) creating clause-external dependencies, since this would violate the deictic nature of an indexical clause. The semantic content of indexical clauses is similarly independent: the reference time is fixed to the speech time (78), and a proposition must be evaluated from the perspective of the speaker, labelled the origo (92). Even when the speaker is not overtly an argument of an indexical clause, the interpretation of Plains Cree independent clauses indicates direct evidentiality that references the experiential knowledge of the speaker (97).

Chapter 4, ‘Anaphoric clauses: Plains Cree’s conjunct order’, considers the elsewhere clauses: those that are not indexical are anaphoric. Anaphoric clauses, as the name suggests, require an antecedent (under the Generalized Pattern of Anaphoric Dependence, Williams 1997), which allows for clause-external dependencies, unlike indexical clauses (127). Cook methodically considers contexts for conjunct, anaphoric clauses (e.g. embedded clauses, temporally dependent matrix ê-clauses, and clauses with differential subject marker -iyi), showing that they are possible only when an antecedent licenses the dependent elements by c-command, precedence, or both. Semantically the evaluation time and origo of an anaphoric clause are not by default the speech time and speaker (as for indexical clauses), but are instead determined by the relevant antecedent (155). For example, the left-edge element kî- is a ‘temporal shifter’ that can indicate that a state holds in the past in an indexical clause (i.e. the speech time does not coincide with the state) (158). In anaphoric clauses, however, kî- ‘sequences’ the verb with respect to the c-commanding or preceding antecedent clause, indicating that action precedes the previously identified time.

Chapter 5, ‘A syntactic classification of anaphoric clauses’, continues the discussion of anaphoric clauses diagnostically sub-classified into chained, adjoined and argument clauses. Chained clauses are dependent on their previous discourse (or ‘root span’) for their interpretation, but are never syntactically subordinated (185–186). Adjoined clauses are shown to be islands but can be c-commanded by their antecedent (unlike chained clauses), and argument clauses can also be c-commanded, but are not barriers to movement (194–196).

Chapter 6, ‘A semantic classification of anaphoric clauses’, goes through the types of conjunct clauses to determine their semantic distribution. Cook considers a range of data that indicate that conjunct verbs prefixed by ê- are the elsewhere form for anaphoric clauses, while those with kâ- specify that a proposition is presuppositional, and simple conjunct verbs (Ø-) appear in subjunctive contexts (e.g. expressing desire, fear, directives, etc.). The second half of the chapter returns to the function of the ‘sequencing’ morpheme kî- and related issues of tense, aspect and modality.

The aim of this book – to prove that clauses can be indexical or anaphoric – is empirically reached through the myriad of examples from Plains Cree exhibiting the behaviour of independent and conjunct clauses. Although descriptive adequacy is obtained, I would argue that explanatory insight could be expanded upon for syntactic and semantic mechanics, specifically the details of how the system differentiates between clauses solely dependent on a speech situation versus those with an anaphoric element. Cook distinguishes the clause types by variables, $s_{0}$ and $s$, on CP (shown in (1) above) and states that ‘[i]n Plains Cree, this distinction is reflected in whether there is a pronominal element in Spec,CP (indexical); or a clause-typing element in $\text{C}^{0}$ (anaphoric)’ (15). She also repeatedly claims that anaphoric clauses are the elsewhere case and, in contrast, indexical clauses are restricted by the anti-c-command/precedence conditions. I now consider an alternative syntactic view: Plains Cree indexical clauses (rather than anaphoric) are default and the content of Spec,CP (rather than Spec,CP vs. C) is in complementary distribution between types.

Anaphoric clauses contain an element with a clause-external dependency, whether the clause is dependent on a matrix clause (e.g. for temporal interpretation), or there is a proform with a clause-external antecedent. A c-commanding/preceding antecedent is required to license the interpretation of the dependent element; hence there is no need for an anti-c-command/precedence restriction on clauses lacking these dependent elements. Cook claims that indexical clauses are restricted by the requirement that a deictic pronominal occupy Spec,CP, namely first or second person.Footnote ^[1] Third person arguments are possible in indexical clauses provided they do not create a clause-external dependency, but are reflected in a -w suffix rather than a proclitic in Spec,CP, shown in (2):Footnote ^[2]

Cook predicts that if the element in Spec,CP is necessarily deictic, then other Algonquian languages with third person (i.e. non-deictic) proclitics cannot have clause distinctions identical to Plains Cree. But suppose the absence of a third person proclitic is a morphological accident – the distribution of independent clauses remains the same when they are simply clauses lacking an external dependency. Despite the descriptively wider distribution of the conjunct, the deictic interpretation of the independent can be viewed as elsewhere since information about the speech situation is always available.

Now consider the makeup of the CP layer. Cook discusses operators in some conjunct clauses, for example the relative root -isi ‘manner’ in (3) indexed with clause-external êkosi ‘topic’ in (3b) (mood and negation operators are also discussed (221)).

Spec,CP in anaphoric clauses can be filled by an operator coindexed with a dependent element, and Spec,CP in indexical clauses is filled by a pronominal. One could suppose that conjunct C probes for anaphoric elements to raise to Spec,CP, shown in (4b), and independent C probes for phi-features, shown in (4a).Footnote ^[3]

Whether one proceeds with Cook’s syntactic sketches or some other permutation, it seems crucial to be able to identify when a clause is anaphoric. Above I sketch ‘Anaphor’ conceptually as a feature, but within compositional semantics (e.g. Heim & Kratzer Reference Heim and Kratzer1998) the distinction between indexical and anaphoric clauses is clear: a variable that is not dependent on a clause-external element will resolve its assignment within its clause, but one that is dependent on a clause-external element will still lack an assignment and can be targeted for raising to Spec,CP.

Concerning the presentation of the book, the introduction and Chapter 2 begin clearly and are accessible. Continuing through the book the argumentation becomes increasingly hard to follow, and occasionally I was not sure to what end some point was contributing, even on rereading. For instance, the lengthy introduction of ‘root spans’ at the end of Chapter 3 neither directly defines root spans nor makes it obvious why they are important – later they are discussed as the types of supra-clausal constituents an anaphoric clause can appear in. I feel that the discussion becomes confined to very specific questions of interpretation and form of the data that do not clearly connect to the larger discussion. This is not to say I expected less detail on the Plains Cree data, which is in fact the overwhelming strength of this work, but rather that it becomes somewhat inaccessible to the reader who is not immersed in the study of Plains Cree. In this vein, the final Chapter 7 could make further concluding remarks that point in a direction where this research may continue in addition to the specific data questions formulated there about other Algonquian languages in comparison with Plains Cree.

To summarize, this book offers a very valuable array of Plains Cree data, and includes the convincing claim that clauses may function as either anaphoric or indexical in this language. More remains to be said about how the Plains Cree clause system functions in the syntax and semantics, and how it is spelled-out in the morphology. These questions, however, could not even have been posed before Cook constructed this comprehensive review of clause interactions.