Language Evolution: The Windows Approach addresses the question: "How can we unravel the evolution of language, given that there is no direct evidence about it?"
The LINGUIST List is dedicated to providing information on language and language analysis, and to providing the discipline of linguistics with the infrastructure necessary to function in the digital world. LINGUIST is a free resource, run by linguistics students and faculty, and supported primarily by your donations. Please support LINGUIST List during the 2016 Fund Drive.
Review of Adverbial Clauses, Main Clause Phenomena, and Composition of the Left Periphery
In this book, Liliane Haegeman sets out to provide a syntactic account of the internal structure of adverbial clauses, in particular, to explain the compatibility of only some adverbial clauses with so-called main clause phenomena (MCP) on syntactic grounds. The specific approach taken follows the tradition of cartographic analysis of syntax, or the articulated left periphery. Haegeman first introduces the specific MCP on which she bases her analysis, i.e., argument fronting in English. After extensive discussion of the data reported in the literature and some novel observations, the author presents her syntactic account for the distribution of MCP in main clauses. As the main contribution of the book, she then carries over this cartographic analysis to adverbial clauses, arguing that adverbial clauses fall into two distinct classes, which she calls ''central'' and ''peripheral'' adverbial clauses. MCP are ruled out in central adverbial clauses through intervention effects, according to Haegeman. On the other hand, MCP are allowed in peripheral adverbial clauses because of their different syntactic structure. After arguing this case for temporal and conditional clauses, Haegeman tentatively extends the movement account to other types of embedded clauses, namely complements of factive verbs, complements of N, and subject clauses.
The book consists of six chapters. The first chapter provides an overview of the history of the left periphery and its analysis in generative grammar. The presentation is based on illustrative examples from the extensive literature on the topic. Already in this first chapter, the considerable variation concerning crucial grammaticality judgments, even in the reported literature, becomes obvious. Throughout, Haegeman conscientiously points out such variation and acknowledges it, even when it cannot be explained within the scope of the book.
Chapter Two contains detailed data on argument fronting in English, both for topicalized as well as focalized arguments. The presented data is aggregated from the literature, with the addition of differing judgments in some cases. The data discussed is completed by attested examples. Contrasting arguments and adjuncts in the left periphery of a clause, Haegeman shows that fronted adjuncts are generally allowed, while fronted arguments are ruled out in most embedded clauses, and give rise to intervention effects. The author then compares English argument fronting with Romance clitic left-dislocation (CLLD), a construction used for expressing a topic-comment structure which contains a resumptive clitic for the dislocated constituent. The collected data shows that Romance CLLD generally patterns with English fronted adjuncts by being permitted even in embedded contexts and not triggering intervention effects. Haegeman calls the different behavior of English arguments vs. adjuncts and English argument fronting vs. Romance CLLD the ''double asymmetry'' of fronted constituents. It is the main empirical finding on which the analyses in the remainder of the book are based.
The cartographic approach to the left periphery, as presented in Chapter One, makes available a range of distinct syntactic positions which can be filled by fronted constituents, including at least a position for focalized phrases and several topic positions. However, not all positions can be filled simultaneously. Chapter Three presents two possible analyses to account for the unavailability of certain sentences in English, in particular, sentences where a lower topic is dominated by a fronted argument in a focus, topic, or wh-position. After discussing a strictly positional account, which only explains some of the data, Haegeman introduces a feature-based proposal and argues that the unavailability of certain combinations in the left periphery is reduced to effects of the locality condition on movement.
Chapters Four and Five address the central topic of the book, the distribution of MCP, and most notably, argument fronting in adverbial clauses. In Chapter Four, Haegeman introduces the distinction between central and peripheral adverbial clauses (p. 160). She notes that temporal adverbials, such as those introduced by 'while' or 'when', have an additional concessive reading. Haegeman claims that the temporal reading modifies the event denoted by the main clause and calls those adverbials ''central'' adverbials. The concessive uses, she argues, are discourse-structuring, i.e., ''peripheral'' adverbial clauses. The distinction between central and peripheral adverbial clauses is crucial to the following discussion, and is extended to other types of conjunctions, such as conditional and causal conjunctions. The remainder of the chapter contains an empirical study of the syntax of the two types of adverbial clauses. Peripheral adverbial clauses are argued to be less closely associated with the main clause; they are outside the scope of matrix operators, and are not temporally subordinated. In contrast, peripheral adverbial clauses are shown to allow MCP, such as English argument fronting.
Chapter Five presents the main new theoretical contribution of the book, a movement analysis of adverbial clauses that explains the status of MCP in the two different types of adverbial clauses. Having established the argument/adjunct asymmetry in the left periphery as a diagnostic for movement in the previous chapters, Haegeman first reanalyzes temporal adverbial clauses as derived by movement. The analysis of when-clauses as free relative clauses is extended to other temporal clauses with 'since', 'before', etc. It is shown that in most cases, these clauses also allow for low construal of the time phrase (an indicator of movement), which can, however, be blocked by islands. A parallel analysis is then given for ''central'' conditional clauses based on Bhatt and Pancheva (2006). Again, the asymmetry between fronted arguments and adjuncts in English (adjuncts are allowed, but not arguments) and the possibility of CLLD in Romance are taken as evidence for the movement account of these conditional clauses.
Finally, Chapter Six investigates to what extent the movement and intervention approach proposed for adverbial clauses in the book can account for the absence of MCP in certain complement clauses. In each case, the author follows the recipe established in the previous chapters: first, the impossibility of MCP is shown, followed by a demonstration of the ''double asymmetry'' between fronted arguments and adjuncts in English and between English argument fronting and CLLD, which is taken as a diagnostic for the presence of movement. Then, a null operator is motivated, which triggers the movement. This analysis is applied to complements of factive verbs, complements of N, and subject clauses.
The book's value for the field goes beyond the particular syntactic framework and analysis championed here. Due to the data aggregation in the first two chapters alone, this book is a must-read for anyone concerned with syntactic phenomena of the left periphery in English. In 100 pages, Haegeman collects over 100 examples from the extensive literature on the topic and presents them in a logical way. The presentation is complicated by the sometimes considerable variation in the reported judgments, which the author conscientiously reports, and at times, supplements with additional differing judgments. These chapters include many long footnotes which, in some cases, span several pages. Even though the collection of data itself is valuable to experts and graduate students interested in the left periphery, the data chapters would have been easier to follow if the book had started with a clearer exposition of the structure of the book and the arguments in later chapters, or if some less crucial data points had been omitted. For example, the relevance of the discussion of verb phrase ellipsis in footnotes (39-44) is unclear to the reader at this point in the book, and thus, could have been left out. In some cases, the reader is confused by small errata such as mismatched references to examples or footnotes.
Methodologically, the proposals in the second part of the book are based on the data collected and analyzed in the first chapters. It is unfortunate that the concepts ''topic'' and ''focus'', which are centrally featured in the book, especially in Chapter 3, were not clearly defined beyond a reference to ''old information'' and ''new information''.
In Chapter Three, the impossibility of argument fronting is analyzed as a kind of intervention effect based on syntactic features; an entity with a richer feature set can move across an entity with a more impoverished feature set, but not the other way around. While the analysis leads to relatively good results, a freely chosen feature set would, of course, be able to account for any constraints on the left periphery. The proposal, therefore, depends on good independent motivation for the proposed features associated with different types of phrases. For wh-phrases, it has been previously argued that only a small subset of these phrases is ''D-linked'' (Pesetsky, 1987), which, in feature terms, means it contains a D-feature. In these cases, according to Haegeman, the wh-phrase blocks topicalization. Based on the observation that yes-no-questions (for some speakers) block topicalization but not focalization, Haegeman tentatively concludes that yes-no-questions contain a D-feature, i.e., are D-linked. This seems counter-intuitive. Such a claim could potentially be motivated or disputed by considering independent arguments for D-linking.
In Chapter Four, the distinction between central and peripheral adverbial clauses is made based on the two readings available for many temporal conjunctions: one temporal reading, and one concessive or causal (Table 4.3, p. 164). However, the main arguments (especially temporal subordination) don't clearly carry over to the other types of adverbials. In particular, for 'because' and 'if', the central and the peripheral adverbial share large parts of their semantics, and can be distinguished from each other much less easily than the clearly distinct readings of the temporal and causal 'since' or the temporal and concessive 'while'. It is also surprising that concessive adverbial clauses are apparently always peripheral. The central/peripheral distinction for adverbial clauses does not seem to line up with the two kinds of adverbial clauses previous authors have distinguished in semantic and pragmatic literature. Here, starting with Rutherford (1970), a distinction is made between ''restrictive'' and ''non-restrictive'' subordinated clauses, of which the restrictive ones are more closely integrated with the main clause. This notion seems to partially correspond to the distinction between ''central'' and ''peripheral'' adverbial clauses. It is all the more surprising that the two sets of categories do not line up completely. For example, Rutherford already allowed for both restrictive and non-restrictive uses of concessive clauses, as in his examples (2a-b) (see also Sweetser, 1982, ex. (42)). In addition, causal 'since' can also be used either restrictively or non-restrictively. Haegeman does mention the existence of speech-act modifying adverbial clauses (Section 4.7), but claims they are orthogonal to the discussed central/peripheral distinction. In contrast, many authors seem to assume a major difference between propositional (central) uses, on the one hand, and epistemic (peripheral) or speech-act-related uses, on the other hand (Hooper & Thompson, 1973; Sweetser, 1982; Scheffler, 2013). For German 'weil' ('because'), Haegeman aligns the central/peripheral split with the internal syntax of the clause: central weil-clauses are verb-final (VF), while peripheral weil-clauses are verb-second (V2). The discussion of this phenomenon is simplified with respect to the extensive literature on the topic (Sohmiya, 1975; Rudolph, 1980; Thim-Mabrey, 1982; Pasch, 1983a,b; Küper, 1984; Blühdorn, 2006, and others). The literature concentrates on the cases where the two syntactic structures differ semantically, but in many cases, verb-final and V2 weil-clauses can be used interchangeably. Motivation for the claim that VF weil-clauses are central, whereas V2 weil-clauses are peripheral, is based on a range of semantic facts, such as the inability of VF weil-clauses to be in the scope of matrix operators and their inability to be bound into by matrix quantifiers. Alternative analyses of these facts appeal to semantic properties of the two kinds of 'weil' instead of syntactic ones (Scheffler, 2005). Finally, Haegeman claims (based on Antomo, 2009) that central (i.e. VF) weil-clauses are incompatible with speech act related modal particles (cf. ex. (70a)), a statement that is easily disproved by a corpus search. Examples with particles such as 'nämlich' or 'ja' are commonly attested:
(i) Manche sagen, er wäre durchgefallen, weil er ja gar kein italienischer Kandidat war, sondern der Kandidat von Merkel, Juncker und Sarkozy. (http://bit.ly/14z683S) Some say he failed because he JA is not at all an Italian candidate, but the candidate of Merkel, Juncker and Sarkozy.
For example, in (i), the weil-clause is clearly a ''central'' clause, since it not only exhibits verb-final word order, but is also clearly semantically embedded within the higher clause ('he failed'). These, and similar examples, suggest that either the alignment of clause-internal word order and the central/peripheral status of the adverbial clause are not as strict as assumed, or that syntactic and semantic factors (e.g. the semantic level of the expressed relation - propositional, epistemic or speech-act related) interact with each other to account for the data.
A similar problem with respect to the central/peripheral distinction arises for conditionals. Since no clear independent definition of the notions ''central'' and ''peripheral'' is given, the distinction is in danger of becoming circular. In the literature, many special subtypes of conditionals are distinguished based on their semantic and sometimes morphosyntactic properties (see Bhatt and Pancheva, 2006). In German, so-called relevance conditionals exhibit a particular external syntax; while regular (non-relevance) conditionals are integrated in the main clause as a phrase, and can be preposed as the (single) preverbal constituent in the main clause, relevance conditionals are less closely associated with the main clause. When they are preposed, another constituent follows to fill the preverbal position. In effect, while conditionals in English can be ambiguous between both readings, they are always unambiguous in German:
(ii) Wenn Du mich brauchst, bleibe ich den ganzen Tag zu Hause. If you need me, [only then] I will stay at home all day. (regular cond.) (iii) Wenn Du mich brauchst, ich bleibe den ganzen Tag zu Hause. If you need me, I'll be at home all day [anyway]. (relevance cond.)
However, while this distinction in German might be a reasonable interpretation of the syntactic difference between ''central'' and ''peripheral'' conditionals, this does not seem to be what Haegeman has in mind. Instead, judging from the examples and some of the exposition, what Haegeman calls peripheral conditionals seem to more closely line up with the subtype called factive or factual conditionals. In particular, she argues that peripheral conditionals usually ''echo'' a previous discourse proposition. For future research, it would be fruitful to more clearly define the central/peripheral distinction and establish how, if at all, it relates to other subclassifications of adverbial clauses, such as the content/epistemic/speech-act level distinction for causal clauses (Sweetser, 1982) or the subcategories of conditionals, such as factual and relevance conditionals (Bhatt and Pancheva, 2006), and others.
In conclusion, the book presents a cartographic analysis of the availability and unavailability of main clause phenomena, in particular, argument fronting in English in adverbial clauses. The author proposes a featural account according to which adverbial clauses are derived by movement which gives rise to intervention effects, prohibiting argument fronting for so-called central adverbial clauses, but allowing it in peripheral clauses. This syntactic proposal is interesting, well-motivated and presented in a logical and clear structure. It opens up fruitful opportunities for further work on the interface to semantics, which has also concerned itself with many of the phenomena touched upon here. The book starts off with a large data section which aggregates and complements previously reported judgments on the composition of the left periphery in English. For this part alone, the volume is a must read for any linguist interested in phenomena of the left periphery. The author meticulously records disagreements and variation with regard to the reported judgments, which can lead to new research into syntactic variation and cross-linguistic studies.
Antomo, Mailin. 2009. Interpreting embedded verb second. Paper presented at ConSOLE XVII, Nova Gorica, January 16-18.
Bhatt, Rajesh and Roumyana Pancheva. 2006. Conditionals. In M. Everaert and H. van Riemsdijk, The Blackwell companion to syntax. Oxford: Blackwell.
Blühdorn, Hardarik. 2006. Kausale Satzverknüpfungen im Deutschen. In: Pandaemonium Germanicum. Revista de Estudos Germanisticos 10, 253-282. Sao Paulo:vFFLCH-USP.
Hooper, J. and Sandra A. Thompson. 1973. On the applicability of root transformations. In: Linguistic Inquiry 4, 465-498.
Küper, Christoph. 1984. Zum sprechaktbezogenen Gebrauch der Kausalverknüpfer denn und weil: Grammatisch-pragmatische Interrelationen. In: Linguistische Berichte 92, 15-30.
Pasch, Renate. 1983a. Die Kausalkonjunktionen “da”, “denn” und “weil”: drei Konjunktionen - drei lexikalische Klassen. In: Deutsch als Fremdsprache 332-337.
Pasch, Renate. 1983b. Untersuchungen zu den Gebrauchsbedingungen der deutschen Kausalkonnektoren da, denn und weil. In: Linguistische Studien des ZISW 104.
Pesetsky, David. 1987. Wh-in-Situ: Movement and unselective binding. In E. Reuland and A. ter Meulen (eds.), The Representation of (In)definitness. Cambridge: MIT Press. 98-129.
Rudolph, Elisabeth. 1980. Bemerkungen zur konnektiven Partikel denn. In: E. Weigand and G. Tschauder (eds.), Perspektive textintern. Akten des 14. Linguistischen Kolloquiums, Bochum 1979, vol. 1, 249-261. Tübingen: Niemeyer.
Rutherford, William E. 1970. Some Observations concerning Subordinate Clauses in English. In: Language 46.1, 97-115.
Scheffler, Tatjana. 2005. Syntax and semantics of causal denn in German. In: P. Dekker and M. Franke (eds.), Proceedings of the Fifteenth Amsterdam Colloquium.
Scheffler, Tatjana. 2013. Two-dimensional Semantics: Clausal Adjuncts and Complements. Berlin, Boston: De Gruyter Mouton.
Sohmiya, Yoshikazu. 1975. Syntax und Semantik der kausalen Konjunktionen, denn und weil. In: Doitsu Bungaku 42, 108-119.
Sweetser, Eve E. 1982. Root and Epistemic Modals: Causality in Two Worlds. In: Proceedings of BLS 8, 484-507.
Thim-Mabrey, Christiane. 1982. Zur Syntax der kausalen Konjunktionen weil, da und denn. In: Sprachwissenschaft 7, 197-219.
Cosmas II. Cosmas II corpora, Institut für Deutsche Sprache, Mannheim. http://www.ids-mannheim.de/cosmas2/
ABOUT THE REVIEWER:
Tatjana Scheffler is a post-doc in the Linguistics Department at the University of Potsdam, Germany. She received her PhD in 2008 from the University of Pennsylvania for her dissertation on 'Semantic Operators in Different Dimensions', the revised version of which was just published as Volume 549 in the series Linguistische Arbeiten, titled 'Two-dimensional Semantics. Clausal Adjuncts and Complements'. Her research interests are in semantics, pragmatics, the syntax-semantics interface, computational linguistics, discourse, and dialog.