Over the past twenty years of cartographic study, it has been generally assumed that the complementizer system can be split into multiple layers, which provide fine-grained landing sites for the co-occurrence of multiple elements in the left periphery of the clause:
However, there seems to be a great challenge to the cartographic program when the strategy is extended to Verb Second (henceforth, V2) languages like German, which require the inflected verb to be in the second linear position. That there is only one element allowed to precede the inflected verb seems to imply that there is only one possible functional projection populated in the left periphery, which is, at first blush, in contradiction to the split-CP idea. The book under review argues that the same left periphery in (1) can be shared by V2 languages. Giuseppe Samo proposes a general peripheral criterion model, according to which the verb undergoes head movement through all the activated left peripheral positions and it halts at the highest one requiring Spec–Head configuration. Consequently, the verb occupies the second linear position, just after its specifier.
This book consists of seven chapters. Chapter 1 starts with a general introduction to the cartographic enterprise and properties of V2, then presents the research question that the book is going to address: if the C-system is really as rich as cartography suggests, then why does V2 not allow more than one element to occur before the inflected verb? In Chapter 2, Samo develops a general peripheral criterion model within the cartographic framework. This model is appealing in two aspects. Firstly, it can account for the canonical V2 order; secondly, it is able to predict micro-parametric variation among V2 languages. The following four chapters are the application of the model and confirmation of the prediction. Chapter 2 (Section 2.3) examines variation in the subject-initial configuration; Chapters 3, 4, and 5 investigate V3 orders in the root context; Chapter 6 is dedicated to V3 orders in the embedded context. Chapter 7 recaps the book. Below I would like to comment on how Samo defends and applies his central proposal, and make two suggestions, accordingly.
In Chapter 2, following the cartographic guidelines – including ‘one feature, one head’, the syntacticization of scope-discourse semantics, and the locality principle – Samo establishes a general peripheral criterion model which goes as follows:
According to this model, V2 phenomena can be accounted for: phrases carrying criterial features such as TOP, FOC, or WH are fronted to the specifiers of the corresponding heads, and the verb undergoes head-movement through all the activated criterial positions (both the left periphery and the low IP area, Belletti Reference Belletti2004) and it halts at the highest one which requires the Spec–Head configuration. It is because the highest head is realized by the inflected verb at the interface with the sound system that we ‘hear’ the V2 phenomenon. In so doing, one can see that the structure of the left periphery of V2 languages can be as rich as that of Italian, as given in (1) above.
Appealing though it is, I would like to point out one aspect which may need refinement. In terms of the motivation for the verb movement, Samo explicitly indicates that, ‘the verb moves to the activated criterial position because it bears the relevant features’ (45). These relevant features are features like {+TOP, +MOD, +FOC, +WH, etc.}, but why does the verb carry these features, and how could we know that the verb carries these features?
Since Samo develops this general criterion model based on Rizzi’s (Reference Rizzi, Belletti and Rizzi1996) WH-criterion, we can first have a look at how Rizzi argues for such a criterion. To account for Subject-Auxiliary inversion in English (what Rizzi called residual V2), Rizzi proposes that the interrogative C head, endowed with [+wh] feature, triggers head movement of the inflected verb, and attracts the wh-element to its specifier. Crucially, Rizzi indicates that the ability of a verbal inflection to carry the [+wh] specification is strongly suggested by the fact that in some natural languages, such as Chamorro, the verb manifests a special morphology in interrogatives. Following the same logic, we expect that, concerning the focus-criterion and the topic-criterion, the verbal inflection should carry the corresponding [+focus] or [+topic] specification, manifested by special morphemes. However, the clause-initial topic and focus in most V2 languages, such as German and Dutch, are followed by identical verbal forms. Therefore, I would like to suggest that if Samo can provide morphological evidence from V2 languages in which the verb inflects distinctly with topic and focus, the assumption that the verb bears criterial features would be more convincing and the criterion model, in turn, would be more persuasive.
Apart from accounting for the V2 phenomenon, Samo holds that this criterial model is also able to predict micro-parametric variation among V2 languages. Based on this criterial model, Samo further assumes that if the leftmost functional projection ‘does not require’ or ‘ceases to trigger’ the Spec–Head configuration with the verb, a V3 order would be expected since the ultimate landing site of the verb is already preceded by its specifier phrase. Samo confirms this prediction by examining three syntactic environments in the following chapters.
Chapter 2 (Section 2.3) investigates variation in the subject-initial context. Adopting the postulation by Cardinaletti (Reference Cardinaletti2004) that the subject position is a field that can be split with multiple nodes (Subj, EPP, AgrS), Samo proposes that the inflected verb may target different positions in the spectrum of the subject domain. Language variation is thus expected according to the height of verb movement. The prediction is that if the verb targets the Subj0 head, we would encounter V2 order; if the verb targets a lower functional head, a V3 order should be expected. This prediction is confirmed by the fact that in German, the subject must be adjacent to the verb, yielding V2; in Surselvan (a variety of Swiss Romansh), adverbs are allowed to intervene between the subject and the verb, forming the subject–adverb–V3 order, as seen in (3).
In Chapters 3, 4, and 5, Samo examines micro-parametric variation, namely, V3 order with the initial element in the pre-IP area. Specifically, Chapter 3 investigates V3 order involving temporal and locative items as the leftmost element. According to Rizzi (Reference Rizzi and Belletti2004a), IP internal temporal adverbs, when they are ‘highlighted’, will move to the ModP (modifier projection), which is higher than the SubjP but lower than other functional projections in the left periphery. Samo argues that whether or not this ModP requires the Spec–Head configuration leads to variation among V2 languages. That is, if the language requires the Spec–Head configuration in ModP, we would expect the initial temporal item to be adjacent to the verb, forming the V2 order. This expected pattern is observed in German, as seen in (4a) below. If the language does not require the verb to lexicalize the head of ModP, we would expect a ModP–subject–V3 order. This pattern is observed in the Swiss Romansch variety Putèr, in (4b).
Chapter 4 explores V3 orders involving two elements that are extracted from IP, before the inflected verb. In German, the V3 order involving a left-dislocated wh-element and a subject is ruled out, whereas in Norwegian this pattern is allowed. Samo attributes this contrast to whether or not the leftmost item requires the Spec–Head configuration. That is, in German the leftmost item was ‘what’ requires the verb to realize the head of its functional projection, hence the ungrammaticality of (5a); in Norwegian the highest element ka ‘what’ does not require the verb to lexicalize its functional head, hence the grammaticality of (5b).
Chapter 5 looks into V3 orders with the leftmost item (speech-act adverbs, periphery CPs, hanging topics, etc.) base-generated in the position higher than ForceP (usually named as FrameP). Samo assumes that these types of items do not require Spec–Head configuration with the verb. Under this assumption, we would expect that all V2 languages allow these base-generated elements to occur clause initially, forming the V3 order. This is exactly what has been observed in V2 languages, such as German and Kiezdeutsch:
Chapter 6 is dedicated to variation in the embedded context, as exemplified by (7), which is attributed to the mechanism of how the complementizer is merged.
In embedded V2 languages like Surmiran, the complementizer tgi ‘that’ is directly base-generated in Force0. The inflected verb vegia ‘have’ undergoes successive head movement through all the activated criterion positions, and it halts at the highest one, where the constituent dultschems ‘sweets’ to be focalized is attracted to its specifier, creating the Spec–Head configuration. As for the embedded non-V2 languages like German, the complementizer dass ‘yhat’ is generated IP internally and then extracted to Force0. Head-movement of the inflected verb to the left periphery would be blocked by the copy of dass in terms of Relativized Minimality. Consequently, the inflected verb is left far away from the highest criterion position, where the criterion head is not realized by the inflected verb.
Recall that the V3 in the above-mentioned three syntactic environments, i.e. subject-initial clause, root context, and embedded context, is explained under the assumption that the highest activated criterial projection does not require the verb to lexicalize its head. However, a further question, theoretically speaking, worth asking is why in V3 order the highest element ‘does not require’ or ‘ceases to trigger’ the Spec–Head configuration with the verb?
I would like to suggest that variation of V2 lies in the parameter of spelling out the functional head, more specifically, the highest functional head if more than one functional projection is activated. Format of spell-out of the highest head can be either null or a verb. When the highest functional head is overt, i.e. realized by the verb, we have V2 order; when the highest functional head is covert, i.e. phonetically null, we have V3 order. The appealing consequence of this suggestion is that we can reduce language variation to the covert/overt expression of the functional head (recall the Borer–Chomsky conjecture), which is generatively welcomed. Moreover, my suggestion can also maintain the fundamental spirit of the criterial approach, that is, the verb always halts at the highest head that needs Spec–Head configuration, be it V2 or V3 order.
Nevertheless, I think the general peripheral criterion model that Samo proposes is on the right track to the cartographic study of V2, and my two suggestions, providing morphological evidence to justify the postulation that the verb bears criterial features, and attributing the micro-parametric variation to the parameter of spelling out the functional head, are made in conformity with the criterial model and aim to refine the model.
