Since Evers’ (Reference Evers1975) account of Verb Raising, verbal complexes (VCs)—clusters of multiple verb forms at the clausal right periphery—have become one of the most widely studied syntactic phenomena in the Continental West Germanic languages and dialects. While VCs in German have been studied primarily from a synchronic perspective with an emphasis on modern varieties, the considerable areal, sociolinguistic, and microparametric variation that characterize these constructions in earlier stages of the language have also garnered diachronic attention.

Building on previous diachronic research (for instance, Ebert Reference Ebert1981), this monograph investigates subordinate clause VCs in Middle High German (MHG), Early New High German (ENHG), and contemporary varieties. The study's main goals are to document VC word orders, describe their areal and temporal distributions, and determine which (socio)linguistic factors favor or disfavor the occurrence of particular serializations, thereby elucidating both diachronic and synchronic patterns. The following ENHG two-verb complexes illustrate the word order variation that forms the basis for the study (cited in Sapp Reference Sapp2011:3):

1. (1)

   

2. 

In ENHG, as in MHG and some modern dialects, the finite verb (indicated by the subscript 1) may appear directly before or after the nonfinite verb it selects (indicated by the subscript 2), yielding the 2-1 and 1-2 orders shown in 1a,b. In addition, nonverbal material may appear between the verbs (1-x-2) or to the right of the VC, as in 1c,d.

The book consists of acknowledgements, six chapters, a bibliography, and an index. Chapter 1, the introduction, briefly outlines the study's aims and rationale, gives an overview of previous scholarship on diachronic German verb order, and ends with a few words on VCs in contemporary German and an outline. Sapp fills a gap in the literature by examining a large corpus of VCs from a wide variety of text types in multiple dialect regions and by using sophisticated statistical tools to simultaneously analyze the effects of various factors on word order in both two- and three-verb complexes. While Sapp does not make the point himself, his study is also novel in its extensive treatment of focus phenomena and their effects on VC serializations. A focused constituent introduces new or contrastive information into the discourse context, as the examples in 2 illustrate.

1. (2)

   

In the context established by the question What did Karl give to the child yesterday?, the direct object in 2a, das Buch ‘the book’, represents new information and is focused. In 2b, framed by the question What happened yesterday?, not only the direct object, but the entire clause is the focused constituent.

Chapter 2 provides an analysis of VCs from 13 MHG prose texts written between 1070 and 1315 in seven dialect areas. The majority of these complexes comprise two verbs and display the 2-1, 1-2, and 1-x-2 orders, followed in some cases by extraposed constituents. Of these two-verb complexes, 28.7% occur with the ascending order 1-(x)-2.

Using GoldVarb X (Sankoff et al. Reference Sankoff, Tagliamonte and Smith2005), a statistics program that enables the researcher to measure the favoring or disfavoring effects of various factors on the occurrence of a dependent variable, Sapp determines for the two-verb complexes that five linguistic factors—syntagm type, category and weight of the word preceding the VC, extraposition, focus, and verbal prefix type—as well as four socio- and extralinguistic factors—occupation, genre, dialect, and date of composition—have statistically significant effects on the occurrence of the 1-2 versus 2-1 word orders. With GoldVarb’s step-up/step-down function, he also finds that the combination of the factors syntagm type, category and weight of the word preceding the VC, focus, occupation, genre, and dialect results in the statistically most significant model.

In particular, the analysis reveals that the 1-2 order is favored by infinitival constructions, stressed words preceding the VC, extraposed argument NPs and PPs, both new and contrastive focus (especially object focus), and the occurrence of a stressed separable prefix or no prefix on the lexical verb in the complex. Furthermore, the 1-2 order is slightly favored in texts written by clerics and is favored in sermons, while this order is strongly disfavored in chancery documents. Finally, the data show that besides a dip in the 12th century, the frequency of the 1-2 order remained stable over the entire period under investigation.

The three-verb complexes in Sapp's MHG corpus are attested with the 3-2-1, 3-1-2, 1-(x)-3-2, and 1-2-3 orders. Because of the small number of tokens, the results for the factors analyzed, namely, syntagm type, the category and weight of the word preceding the VC, extra-position, focus, prefix type, date of composition, dialect, and socio-linguistic factors, are inconclusive. Nonetheless, the results indicate that the 3-2-1 order is rare, and that 1-2-3 becomes much more frequent from the 12th to the 13th century.

Chapter 3 presents a corpus study of VCs from 30 ENHG texts composed between 1350 and 1600 in 10 dialect areas. The corpus consists primarily of two-verb complexes, but also three-verb complexes and, unlike Sapp's MHG corpus, a small number of four-verb complexes.

The two-verb complexes display the word orders 2-1, 1-2, and 1-x-2, with extraposition in some cases.Footnote ⁵ Of these, 24.5% occur with the ascending order 1-(x)-2. A GoldVarb analysis reveals that five factors affect word order: syntagm type, category of the word preceding the VC, extraposition, focus, and verbal prefix type. In addition, the socio- and extralinguistic factors occupation, genre, dialect, and date of composition are shown to have statistically significant effects. The factors determined by the step-up/step-down function as those resulting in the best model fit are syntagm type, category of the word preceding the VC, focus, verbal prefix type, occupation, genre, dialect, and date of composition.

As in MHG, the 1-2 order is favored by infinitival constructions, extraposed arguments, the occurrence of a stressed separable prefix or no prefix on the lexical verb in the complex, and both new and contrastive focus. In addition, Sapp determines that the lack of scrambling of NPs to the left of a negator favors the 1-2 order. Since new information and contrastive NPs tend not to scramble, Sapp interprets these findings as a possible confirmation of the effect of focus. In contrast to MHG, the stress of the word before the VC does not affect word order in the corpus as a whole, though individual dialects do prefer the 1-2 order after nouns, as Sapp Reference Sapp2006 and Ebert Reference Ebert1981 also show.

Generally confirming results in Bies Reference Bies1996 and Ebert Reference Ebert1981, Sapp finds that occupations with a higher status (for example, university officials and nobles) correlate with lower rates of 1-2, and that clerics show the highest rate. GoldVarb indicates furthermore that the most formal genre (chronicles) displays the lowest rate of 1-2, while religious writings and literature favor this order. Finally, all dialects evidence an overall decrease in the 1-2 order from the 14th to 16th century.

The three-verb complexes in Sapp's ENHG corpus display the 3-2-1, 3-1-2, 1-(x)-3-2, and 1-(x)-2-(x)-3 orders, alone or with extraposition. The GoldVarb results show that the factors syntagm type, focus, and verbal prefix type affect word order in the VC, though only few of the results are statistically significant because of the small number of tokens in this group. As for focus, no effect on the individual word orders 3-1-2, 1-3-2, and 1-2-3 can be determined, but the data indicate that new and contrastive focus favors these orders in combination, and that old information favors the 3-2-1 order. The effects of the socio- and extra-linguistic factors are complex and less clear than those for two-verb VCs.

To conclude chapter 3, Sapp demonstrates that the vast majority of occurrences of the 1-2 order in his ENHG corpus can be attributed to at least one prosody-related factor, including focus, which holds for his MHG corpus as well. However, since written historical texts form the basis for these corpora, the data do not provide direct insight into prosodic patterns. The primary goal of chapter 4 is, therefore, to investigate the effects of focus and prosody on word order in modern German to evaluate the findings for earlier stages. After reviewing the existing scholarship on two-, three-, and four-verb clusters as well as complexes with intervening constituents, Sapp presents the results of small experiments conducted with speakers of three German dialects (Swabian, Austrian German, Swiss German) and Standard German.

To examine the role of focus in Swabian and Austrian German independent of prosody, Sapp administered two grammaticality judgment tasks involving two-verb complexes. Although the findings do not clearly establish a link between VC word order and focus, speakers of both dialects deemed the 1-2 order more acceptable with an unscrambled direct object adjacent to the VC, similar to the favoring effect that the lack of scrambling has on the 1-2 order in ENHG. Sapp suggests, however, that this increased acceptability may be attributable not to focus, but rather to heavy stress on the object.

In a separate more methodologically sound study of two-verb complexes, Sapp tested speakers of Zurich German using a questionnaire format. A GoldVarb analysis indicates that the 1-2 order is favored when the object or VP is focused and, although not statistically significant, that 1-x-2 is favored when the finite modal verb is focused. Sapp notes that the favoring effect of object and VP focus in Zurich German, which is accompanied by sentential stress on the object, is similar to the improved acceptability of the 1-2 order when a direct object is unscrambled or stressed in Swabian and Austrian German.

Finally, Sapp conducted an experiment with speakers of Standard German, building on Schmid & Vogel Reference Schmid and Vogel2004, to determine the effect of subject, object, VP, lexical verb, and modal verb focus on the acceptability of the orders 3-2-1, 3-1-2, 1-3-2, and 1-2-3 with the syntagm infinitive + modal + werden (for example, dass er es machen₃ müssen₂ wird₁ ‘that he will have to do it’). For all conditions but focus on the modal, the 3-2-1 order showed a higher degree of acceptability than the 1-3-2 order. However, 1-3-2 was judged almost equally as acceptable as 3-2-1 under object focus, which corroborates the finding from chapter 3 that object focus has a favoring effect on the occurrence of the 1-3-2 order in ENHG.

In chapter 5, Sapp shifts from the description of word order in VCs to a discussion of their formal derivation and an explanation for the effects of focus and prosody established in the previous chapters. After outlining den Besten's (Reference Besten and Abraham1983) classical account of German clause structure, Sapp demonstrates, following Haegeman's (Reference Haegeman1992) analysis of the 1-2 and 1-x-2 orders as Verb Raising and Verb Projection Raising, respectively, that MHG and ENHG structures can be derived from an SOV base through movement. Since such structures can also be derived from an SVO base, as Zwart (Reference Zwart, Black and Motapanyane1996) shows, Sapp then compares the SOV and SVO approaches. He concludes that the former offers a slight empirical advantage because it more straightforwardly excludes the unattested (or marginal) 2-x-1 order.

However, both approaches are faced with the problems of motivation and optionality of movement. Therefore, Sapp seeks an alternative analysis, one that is not restricted to the narrow syntax and can account for the main linguistic findings from chapters 2–4. As most of these are focus-related, Sapp evaluates two formal syntactic approaches to focus and determines that neither the prosodic-based (Cinque Reference Cinque1993) nor the feature-based analysis (Koopman & Szabolcsi Reference Koopman and Szabolcsi2000) can account for all of the findings. This leads him to propose that the focus effects are a manifestation of prosody that can best be captured postsyntactically at the syntax-prosody interface.

Building on the observation in Behaghel Reference Behaghel1932:IV, 87 that the occurrence of the 1-2 versus 2-1 orders is attributable in part to prosody, Sapp hypothesizes that the study's prosody-related findings are conditioned by Selkirk's (Reference Selkirk1984) Principle of Rhythmic Alternation (PRA), which deals with the alternation of stresses at multiple levels, including the syllable and word levels:

1. (3) Principle of Rhythmic Alternation

2. Between two successive strong beats on a metrical level n there must intervene at least one (and at most two) weak beat(s) of the metrical level n. (Selkirk Reference Selkirk1984:48; cited in Sapp Reference Sapp2011:190)

In essence, this principle bars the occurrence of stress clashes and lapses at each level. With respect to syllables and words, it dictates that two stressed syllables or words may not occur directly adjacent to each other, nor can more than two unstressed syllables or words occur in succession.

To resolve violations of the PRA, Selkirk proposes mechanisms that alter the prosodic structure of utterances. One such mechanism is Beat Movement, which breaks up clashing adjacent stresses by shifting one stress to a position that is normally weak. The following examples from Zurich German illustrate the resolution of a clash caused by two adjacent stressed words:Footnote ⁶

1. (4)

   

In 4a, the PRA is violated because two stressed words are adjacent: the lexical verb schriibe ‘write’ and the direct object Roman ‘novel’, which also bears sentential stress. As shown in 4b, the movement of the stress on schriibe to the modal verb, which is normally unstressed, resolves the clash.

Sapp suggests that, in addition to Selkirk's mechanisms, clashes may be resolved through the 1-2 order in varieties of West Germanic that allow Verb Raising, and that this is the preferred strategy. This form of resolution is shown in 5.

1. (5)

   

With the unstressed modal verb between the stressed direct object and lexical verb, the word clash is resolved.

These examples illustrate how the PRA can explain one of the key findings from chapter 4, namely, that the 1-2 order is favored in Zurich German if the direct object is part of the focus. Assuming that the object bears sentential stress in this case, a clash will arise with the 2-1 order since the lexical verb, stressed by default, will occur adjacent to it. With the 1-2 order, however, this clash is avoided. Sapp demonstrates how the remaining prosody-related results from chapters 2–4 follow to some degree from sensitivity to the PRA.

Sapp formalizes his analysis by adopting Büring's (Reference Büring, Molnár and Winkler2006) non-syntactic, prosodic approach to the effect of focus on word order. On this approach, focus does not affect word order directly, but rather through its prosodic manifestation. To reconcile this prosodic account with a derivational model of syntax, Sapp argues, following a modified version of Hinterhölzl's (Reference Hinterhölzl, Fuß and Trips2004) analysis of scrambling, that nonoptional movement operations occur in the narrow syntax, leaving copies of a constituent in both its base position and at the target of movement. While both copies are spelled out, the prosodic structure of the utterance determines which copy will be pronounced. Sapp demonstrates that this option-copy approach to VCs, whereby the results of movement are filtered at the syntax-prosody interface, is superior to analyses involving PF-movement or optional movement exploited at the interface.

In chapter 6, Sapp lays out implications of his study for the origin of Standard German, the clause structure of German, the peripheral nature of VC phenomena, and syntactic change. First, socio- and extralinguistic findings from chapters 2 and 3, which lead Sapp to conclude that the 2-1 order was standardized through its use in the supra-regional chancery style, support Besch's (Reference Besch1968) supra-regional approach to the development of the standard. Second, the study's diachronic perspective bolsters the analysis of German as an SOV language. On this view, the decrease in the three types of rightward movement attested in Sapp's corpora, namely, Verb Raising, Verb Projection Raising, and extraposition, can be explained by the increasing restriction of such movement over time, but this development is not as straightforward if one assumes that German is SVO. Third, following Uriagereka Reference Uriagereka, Lakarra and Ignacio Hualde2006, Sapp argues that VC phenomena reside at the periphery of the grammatical system, not its core, and are therefore prone to linguistic change, independent of changes to the core grammar. Finally, Sapp interprets the results of his study as support for a nuanced view of language change that distinguishes between core grammatical parameters and peripheral features.

This well-researched monograph represents a major contribution to research on VCs in West Germanic. Among its strengths are the temporal and areal spread of the data, rigor of the statistical analyses, use of experimental data to elucidate historical patterns, and the prosodic account of VC variation, which not only unifies the main linguistic findings but is also compatible with a derivational model of syntax. The multiple perspectives that this accomplished study offers will make it attractive to scholars working in a range of research traditions, including historical, generative, and variationist linguistics as well as dialectology.

My few minor substantive criticisms stem from the occasional lack of detail in an overall meticulously written work. The first concerns the determination of the focus domain in chapters 2 and 3. To establish whether clauses contain new or contrastive information, Sapp examined them in the context of the section of text in which they occur or, if a work lacks sections, within the previous one or two pages. While the decision to define the focus domain as the section, whenever possible, makes intuitive sense, it is unclear whether there was a principled motivation for otherwise choosing one to two pages as the domain as opposed to some other number of pages or a set word count. The particular number of pages is not crucial, of course, as long as the same standard is applied across all texts. However, if the amount of text per page varies by work, as a survey of the online texts in the Bochumer Frühneuhochdeutschkorpus, the source of the ENHG data, suggests, the focus domains will also vary, which may have affected the results.

In his analysis of the main focus- and prosody-related findings in chapter 5, Sapp is primarily concerned with clashes at the word level. He convincingly argues that the preference for many patterns can be attributed to the avoidance of word clashes, but in several cases (for instance, p. 197, 49a: I ha gsäit, dass dä Martin en Roman schriibe MUES ‘I said that Martin must write a novel’) a syllable clash presumably remains. Though Sapp does briefly address syllable clashes in note 21 on p. 192, there is no further discussion of these in the context of the relevant data. These persisting clashes are unlikely to have major implications for Sapp's analysis; Selkirk (Reference Selkirk1984:63) notes, for example, that “clashes on the second metrical level [that is, syllable level] seem to be tolerated quite well.” Nonetheless, greater attention to syllable stress clashes would have been warranted since one important feature of the PRA is its operation on multiple levels simultaneously.

Finally, Sapp clearly demonstrates in chapter 5 that the optional-copy approach can, in principle, better account for both scrambling and VC word order alternations than movement at PF or optional movement at the interface. However, this approach still requires syntactic movement that must be triggered: In the case of two-verb clusters, Sapp assumes “that 1-2 is derived by the obligatory rightward movement of the nonfinite verb to check some (as yet unspecified) syntactic feature” (p. 204). Just what might such a feature be? As Wurmbrand (Reference Wurmbrand, Everaert and van Riemsdijk2006:284–294) concludes after considering the motivation for VC movement, there do not seem to be candidate features that follow from independent grammatical properties of a particular language. In light of this state of affairs, a more detailed discussion of syntactic triggers would have been welcome.

The text is very well edited with only a few word-choice and typo-graphical errors (for example, p. 20: affect → effect; p. 81, note 38: syntagms → syntagms.), though somewhat more attention could have been paid to the text-internal consistency of numbers and percentages in chapter 2. Sapp states on p. 20 that 1,109 two-verb subordinate clauses occur in his MHG corpus and that 320 (28.9%) display the ascending 1-2 order. However, the subsequent tables and discussions show the total number to be 1,106, with 317 (28.7%) displaying 1-2. Similarly, the total number of three-verb clusters in MHG is reported on p. 18 as 23, but 26 tokens are included in the analyses on pp. 36–43.

Despite these minor points, The Verbal Complex in Subordinate Clauses from Medieval to Modern German is an outstanding piece of scholarship that I recommend to anyone interested in VCs, interface phenomena, language variation, or syntactic change.