In this monograph, Yehuda N. Falk presents a view of subjecthood which seeks to reconcile the major insights stemming from both the functional-typological and formal approaches to subjects developed within the last thirty years or so. Working within the framework of Lexical Functional Grammar (LFG), he takes grammatical functions (GFs), of which the subject is a major manifestation, to be a universally relevant level of syntactic representation, parallel to the levels of constituent structure and argument structure. The subject GF thus emerges as relevant for all languages (165). The subject GF, however, does not correspond to the traditional notion of subject. The latter, according to Falk, reflects the intersection of the subject GF with another type of grammatical function, an ‘overlay’ or secondary function, called the syntactic pivot (PIV). The PIV, unlike the subject GF, is not universal.

The subject GF itself is characterised as the element with the function of expressing as a core argument the hierarchically most prominent argument on the argument structure hierarchy (39), which in LFG is called θ. In the unmarked case, θ corresponds to the highest-ranking argument on the thematic hierarchy, which will generally be the agent, provided there is one, otherwise the patient/theme, etc. However, since the level of argument structure in LFG is a syntactic rather than a semantic level and thus contains both thematic elements and non-thematic ones (e.g. expletives and idiom chunks), θ may also be non-thematic, as there is in There is a book on the table. Further, if θ is displaced from core argumenthood, as, for example, in passive constructions, the subject GF will correspond to the next highest-ranking argument (i.e. the patient/theme and so on). This mapping between arguments and GFs is achieved not via a one-to-one association between thematic roles and GFs but via a systematic hierarchy-to-hierarchy mapping (37). Significantly, in contrast to earlier LFG analyses, most notably that of Manning (Reference Manning1996), the mapping of thematic and argument structure onto GFs is taken to be cross-linguistically uniform. In terms of the nowadays widely used labels S (sole argument of intransitive verb), A (agentive argument of transitive verb) and P (patient-like argument of transitive verb), Falk takes the subject GF to correspond invariably to S and A, rather than to S and A in nominative/accusative languages but S and P in ergative/absolutive languages. The argument-to-GF mapping in so-called active languages, where S evinces A-properties with one set of verbs or constructions and P-properties with others, is dealt with by assuming that the GF whose corresponding argument role is missing is skipped (47). Accordingly, an S manifesting P-properties will be assigned the object GF and not the subject GF.

In contrast to the subject GF, which expresses the most prominent syntactic argument, the PIV is the element that has the function of connecting its clause to other clauses in the sentence (74). Falk conceives of it not as an argument function but as an overlay function which is assigned to arguments to enable cross-clausal continuity (77). In fact, Falk suggests that within LFG it is only via the PIV that reference to any function in a lower or coordinate clause can be made. This is referred to as the Pivot Condition (78). Thus, by definition, the PIV is any element which can be shared to make inter-clausal linkage possible. Since inter-clausal linkage is cross-linguistically not restricted to the sharing of arguments that belong to a specific set of core arguments, the PIV may involve S and A, S and P or indeed any core argument as long as it is a discourse topic (175). Moreover, the PIV, unlike the subject GF, is not a necessary feature of all languages. Therefore, under Falk's approach to subjecthood, typological variation resides not in the presence vs. absence of a subject GF or in the nature of this GF, but rather in the presence/absence and nature of the PIV.

The major typological distinction that Falk draws with reference to the above is between languages in which the subject GF and PIV coincide, (i.e. both correspond to S and A) and those in which they do not. Somewhat confusingly, given his distinction between the subject GF and the PIV, though in line with the more traditional conception of subjecthood, he calls the former ‘uniform subject languages’ and the latter ‘mixed subject languages’ (12). Needless to say, English and the other well-known syntactically nominative/accusative languages are uniform subject languages. Mixed subject languages are essentially those which have been described as featuring syntactic ergativity, such as the Australian languages Dyirbal and Yidiny, the Inuit languages (e.g. West Greenlandic), the Mayan languages (e.g. Jacaltec and Tzutujil) and the languages of the Philippine type (e.g. Balinese, Karao and Tagalog). In addition to uniform subject and mixed subject languages, Falk recognises a group of languages for which this distinction is not relevant, by virtue of the fact that their PIV is based not on argument but on discourse status (178), more specifically topic status. Such languages, which include Mandarin, are called topic-pivot languages, in contrast to argument-pivot languages, which encompass both the uniform subject and mixed subject type. Finally, there are pivotless languages (178–193), which have no PIV, be it an argument-based or a discourse-based one. These are languages which lack long-distance dependencies and functional (as opposed to anaphoric) control constructions. (Whether they also lack syntactic passives and antipassives, as the discussion of pivots by other scholars would lead one to expect, is not clear as the issue is not addressed here.) The major examples of pivotless languages provided by Falk are the Muskogean languages Choctaw and Chickasaw and, somewhat surprisingly, Warlpiri (on which see below).

As Falk himself acknowledges, much of the discussion presented in this monograph, as well as the language data cited, will be familiar to anyone who has been following the literature on subjects from the mid-1970s onwards. Falk's major contribution to the discussion of subjecthood relates to the notion of pivot. His bifurcation of the traditional notion of subject into subject GF and PIV builds on the distinction between role- and reference-related subject properties, first discussed by Schachter (Reference Schachter, Peter and Sadock1977) and subsequently captured in terms of the contrast between semantic vs. pragmatic pivots by Foley & Van Valin (Reference Foley and Van Valin1984). The major difference between his PIV and that of the syntactic pivot of other scholars, such as Dixon (Reference Dixon1994) ; Givón (Reference Givón and Talmy1997), Kibrik (Reference Kibrik1997) and Van Valin & LaPolla (Reference Van Valin and Randy1997), is that it is not construction-specific but language-specific (83). In other words, languages may have only one type of pivot. If they have an argument pivot, it must be either an S/A one or an S/P one, but not both; and if they have a topic pivot, then they cannot have an argument pivot. This entails that various phenomena that are generally analyzed as pivot-sensitive emerge as pivotless under Falk's approach.

Falk bases his recognition of syntactic pivot not only on the identity of the arguments involved in the co-reference between clauses but also on the formal properties of the linkage (89–95). In the case of cross-clausal coordination, for example, only constructions that must be analysed as having a single syntactic element bearing a grammatical function in more than one clause qualify as pivot-sensitive. Constructions which may be viewed as based on overt or covert anaphoric relations or subconstituent coordination do not qualify. Consequently, whereas cross-clausal coordination in English is typically considered to reflect an S/A pivot, that in Dyirbal an S/P pivot and that in Yidin both an S/A and an S/P pivot (depending on the coordination), under Falk's analysis only Dyirbal coordination is pivot-based. According to Falk, cross-clausal coordination in English is better seen as involving not ellipsis of S or A (e.g. [[Betty came in], [Ø sat down] [and Ø picked up the phone]]) but rather subclausal constituent coordination, i.e. coordination of VPs (e.g. [Betty [came in], [sat down] [and picked up the phone]]) and thus no ellipsis at all. Cross-clausal coordination in Yidin, on the other hand, is interpreted as involving the semantic sharing of anaphoric elements, i.e. the presence of null pronouns, rather than the syntactic sharing of an overt and an elided constituent. Falk hypothesises that given his more restricted view of what constitutes a pivot-related construction, all languages previously analysed as exhibiting multiple pivots (in his sense of the term and not merely subject GF vs. pivot phenomena) will on closer inspection emerge as having unique pivots. Needless to say, whether only a single pivot needs to be recognised per language is only in part an empirical matter, since much depends on the theoretical assumptions that one makes in relation to clausal architecture and the distinction between null pronouns (covert anaphora) as opposed to argument sharing.

Another important feature of Falk's notion of pivothood is the distinction between topic-pivot and argument-pivot languages. This distinction is not as well explored as that relating to (syntactic) pivot-sensitive as opposed to pivotless constructions, presumably due to the underdeveloped state of knowledge in regard to discourse functions within formal theories. Significantly, Falk's topic-pivot languages include not only the expected topic-prominent languages of Li & Thompson (Reference Li, Thompson and Charles1976), such as Mandarin or Lahu, but also what others have viewed as being purely role-dominated languages such as Acehnese. This is to a large extent a consequence of Falk's definition of PIV. Since any element which functions in cross-clausal continuity must be a PIV, languages in which cross-linkage phenomena are sensitive to core argument status per se, i.e where cross-linkage applies irrespective of which core argument is involved (rather than to some subset of the core arguments), cannot be interpreted as lacking a PIV (178). Since the PIV is not determined by argument type, Falk suggests that it must be determined by discourse properties. Although Mandarin and Acehnese are the only languages that he cites in this connection, one can assume, given his argumentation, that he would also have to treat as topic-prominent other languages which have been argued not to differentiate among the core arguments in relation to clause linkage phenomena, such as Archi, Meithei, Mongsen, Nunggubuyu and Riau Indonesian. While Falk takes the (re)classification of Acehnese and Mandarin as pivot-sensitive languages to indicate the superiority of his approach to pivothood in comparison to that of others, in particular to that of Van Valin & LaPolla (Reference Van Valin and Randy1997), it is difficult to agree with him on this point until clear evidence is provided that discourse topicality is indeed what underlies inter-clausal continuity in all such languages.

Finally, Falk's hypothesis regarding the lack of pivots in morphologically as opposed to syntactically ergative or split ergative languages deserves mention (194). Since Falk views unmarked nominative case to be a function of an S/A pivot and unmarked absolutive case to be a function of an S/P pivot, he considers the overt marking of A in a language exhibiting ergative case marking as incompatible with a potential S/A pivot. Accordingly, he speculates that pivotlessness should be rife among morphologically ergative languages. Unfortunately, he does not make much use of the literature to support his hypothesis and cites only Warlpiri and, potentially, Hindi as cases in point. A glaring omission in this context is the failure to mention the morphologically ergative Dagestanian languages, such as Archi, Bezhta, Godoberi, Lak and Tsakhur, which have been argued by Kibrik (Reference Kibrik1997) to lack a syntactic as opposed to a semantic pivot, i.e. to be role-driven as opposed to reference-driven.

Falk's discussion of the subject GF and types of PIV is an important contribution to the literature on grammatical functions, especially as concerns their place and representation in our model of the architecture of grammar. His view of the nature of the subject GF, on the one hand, and the PIV, on the other, is well articulated and clearly constitutes a viable alternative to other approaches. It must be noted, though, that Falk is quite selective when elaborating his view of subjecthood and positioning himself in relation to other approaches. He does not take into account the analyses of subjecthood within frameworks such as Functional Grammar, Systemic Functional Grammar, Cognitive Grammar or Construction Grammar, and does not always do full justice to the approaches that he does consider, most notably Role and Reference Grammar. With this caveat aside, I thoroughly recommend Falk's Subjects and Universal Grammar to anyone seriously interested in the topic of subjecthood.