The secret life of focus exponents, and what it tells us about fronted verbal projections

The paper investigates a complex word order phenomenon in German and the interaction of syntax and information structure it exemplifies: the occurrence of subjects as part of a fronted non-finite constituent and particularly the so-called definiteness effect excluding (many) definite subjects from this position. We explore the connection between focus projection and the partial fronting cases and show that it is the subject of those verbs which allow their subject to be the focus exponent that can be included as part of a fronted verbal constituent. In combination with the observation by Webelhuth (1990) that fronted verbal constituents need to be focused, this provides a natural explanation of the definiteness effect in terms of the information structure requirements in these sentences. Interestingly, the generally ignored exceptions to the definiteness effect are predicted by our analysis; we show that they involve definite noun phrases which can bear focus, which allows them to be part of a fronted verbal constituent. Finally, building on the integrated grammatical architecture provided in De Kuthy (2002), we formulate an HPSG theory which captures the interaction of constraints from syntax, information structure and intonation. 1 The phenomenon Haider (1982, p. 13), Kratzer (1984, p. 45), and Grewendorf (1989, p. 23) observed that in German it is possible for ergative verbs to realize a subject as part of a fronted non-finite verbal constituent. This is exemplified in (1). (1) a. [Ein annom Fehler error unterlaufen] crept in ist is meinem my Lehrer teacher noch still nie. never ‘So far my teacher has never made a mistake.’ b. [Haare hairnom wachsen] grow können can ihm him nicht not mehr. anymore ‘His hair cannot grow anymore.’ Haider (1990) observed that this option also exists for unergative verbs, which is illustrated by (2).1 (2) [Ein annom Außenseiter outsider gewonnen] won hat has hier hier noch still nie. never ‘An outsider has never won here yet.’ Grewendorf (1989, pp.192f) also mentions Toman (1986) with an example in which a subject has been fronted with an unergative verb. (i) ? [Eine anom Lösung solution gefehlt] lacked hat has uns usdat schon already lange for a long time ‘We have lacked a solution for a long time already.’ Nevertheless, the occurrence of subjects as part of a fronted non-finite constituent exhibits a number of restrictions. Meurers (2000, ch. 10) pointed out that the option is only available for subjects of raising verbs: (3) a. [Ein anom Außenseiter outsider zu to gewinnen] win scheint seems hier here eigentlich actually nie. never ‘An outsider never actually seems to win here.’ b. * [Ein anom Außenseiter outsider zu to gewinnen] win versuchte tried hier here noch actually nie. never ‘An outsider never actually tried to win here.’ Kratzer (1984, p. 46), Grewendorf (1989, p. 24), and Haider (1990, p. 96) mention a definiteness effect disallowing definite subjects from occurring in this construction, an effect which has also received attention in the more recent literature (cf., e.g., Wurmbrand 2001). Using definite subjects for the sentences (1a) and (2) thus results in the ungrammatical sentences in (4). (4) a. * [Dieser thisnom Fehler error unterlaufen] crept in ist is meinem my Lehrer teacher noch still nie. never ‘So far my teacher has never made a mistake.’ b. * Der the Außenseiter outsider gewonnen won hat has hier here noch still nie. never There are, however, some clear counterexamples to this definiteness effect, which are rarely noted2 and are not accounted for by any analysis we know of. Some examples are shown in (5). (5) a. Das the Herz heart geklopft beaten und and geschaudert shuddered hat has dem the Kind. child (Wegener 1990, p. 98) ‘The child’s heart beat and it shuddered.’ b. Die the Hände hands gezittert trembled haben have ihm him diesmal this time nicht. not (Höhle 1997, p. 114) ‘This time his hands didn’t tremble.’ c. Das the Telephon telephone geklingelt rang hat has hier here schon yet lange long nicht not mehr. anymore ‘The telephone hasn’t been ringing here in a long time.’ In this paper, we want to discuss and explore an exciting parallelism between the restriction on what can be fronted as part of a non-finite verbal projection and the properties of focus projection in German. We will argue that the definiteness E.g., in fn. 88 on p. 283 of Müller (2002). effect should be viewed as reflecting the information structure requirements in such sentences, instead of stipulating it as a syntactic constraint—and we will see that the apparent counter-examples are predicted under such an information structurebased approach. 2 Information structure German is a so-called intonation language in which focused constituents are signaled by pitch accent (cf., Féry 1993). The syllable bearing the pitch accent is called the focus exponent. Only one syllable is stressed by a pitch accent, but through focus projection larger parts of a sentence can be focused. 2.1 Focus projection and its connection to fronting As illustrated in (6), the focus exponent in an all-focus sentence normally is one of the arguments of the main verb, but not the subject (Stechow and Uhmann 1986).3 (6) Was ist denn hier für eine Aufregung? / What’s all the excitement about? a. [[Ein anom Politiker politician hat has das the VOLK people belogen.]]F lied to b. # [[Ein anom POLITIKER politician hat has das the Volk people belogen.]]F lied to Interestingly, as discussed by Grewendorf (1989) and Uhmann (1991, p. 199ff.), in certain cases the subject can be the focus exponent. In addition to ergative subjects (7), this is also the case for the subjects of many intransitive unergative verbs (8). (7) Was ist denn hier für eine Aufregung? / What’s the matter here? a. [[Dem the Präsidenten president ist is ein a FEHLER mistake unterlaufen.]]F crept in ‘The president made a mistake.’ (8) Was ist denn hier für ein Lärm? / What’s all the noise about here? a. [[Ein a HUND dog bellt.]]F barks b. [[Ein a KIND child weint.]]F cries Connecting this state of affairs to the issue of fronted verbal constituents we started with, one makes an important observation: The examples (9)–(11) show that it is the subject of those verbs which allow their subject to be the focus exponent that can be included as part of a fronted verbal constituent. We write the word that bears the pitch accent in capital letters and mark the entire focus domain with [[. . . ]]F . Sentences which are not felicitous under the given context are marked with #. (9) a. # [[Ein anom POLITIKER politician hat has das the Volk people belogen.]]F lied to b. * [Ein a Politiker politician belogen] lied hat has das the Volk people noch still nie. never (10) a. [[Ein anom HUND dog bellt.]]F barks b. [Ein a Hund dog gebellt] barked hat has hier here noch yet nie. never (11) a. [[Dem thedat Präsidenten president ist is ein anom FEHLER mistake unterlaufen.]]F crept in b. [Ein an Fehler error unterlaufen] crept in ist is dem the Präsidenten president bisher so far noch still nie. never This observation turns out to be a rediscovery: In a surprisingly neglected paper discussing evidence for a VP-constituent in German, Webelhuth (1990, p. 53) connects the issue of focus projection to the issue of subjects as part of fronted non-finite constituents. He claims that the fronted constituent can only contain dependents which are capable of projecting focus. This general claim is only substantiated with one example pair, though, and unfortunately seems to have been overlooked by the later literature. Webelhuth (1990, p. 53) concludes that “we can explain these facts if we assume that a topicalized constituent containing a verb has to be focused” and relates this to a claim by Grewendorf (1989, p. 194, pp. 219f).4 In order to illustrate Webelhuth’s claim let us take a look at some focus-background structures of sentences with a fronted verbal projection. The question-answer pair (12) shows that a fronted verbal projection can be in the focus of an utterance. (12) Was ist hier noch nie passiert? / What has never happened here? [[[Ein an AUSSENSEITER outsider gewonnen]]]F won hat has hier here noch yet nie. never The examples (13) and (14) show, that it is not sufficient for part of the fronted verbal projection to be the focus. (13) Was ist hier noch nie einem Außenseiter passiert? / What has never happened to an outsider? # [Ein Außenseiter [[GEWONNEN]]F ] hat hier noch nie. Close reading reveals that Grewendorf (1989) apparently did not realize the full generality of the connection between focus projection and fronting in that he restricts it to ergative and theme verbs, explicitly excluding agentive subjects such as the one in Webelhuth’s example (i). (i) [Leute people getanzt] danced haben have hier here noch never nie. (14) Wer hat hier noch nie gewonnen? / Who has never won here? # [[[Ein AUSSENSEITER]]F gewonnen] hat hier noch nie. The three examples thus support Webelhuth’s claim that a fronted verbal projection has to be focused. Webelhuth’s conclusion leaves open whether the fronted verbal projection can also be part of a larger focus. This is tested in (15). (15) Was ist denn hier für eine Aufregung? / What’s all the excitement about? # [[[Ein AUSSENSEITER gewonnen] hat hier noch nie]]F The fact that the answer is infelicitous in such an all-focus context shows that only the fronted verbal constituent must be focused, not more. 2.2 The definiteness effect We are now ready to return to the definiteness effect and the counter-examples from the beginning of this paper. We saw that many definite subjects cannot be part of a fronted verbal projection (4), whereas other definite subjects can (5). The pattern becomes transparent when one considers the focus projection possibilities in these examples: Example (16a) and (17a) show that in the sentences disallowing the fronting, the subject cannot be the focus exponent, whereas it can be the focus exponent in the second class of sentences, (16b) and (17b). (16) Was ist denn hier für eine Aufregung? / What’s the matter here? a. # [[Dem the Präsidenten president ist is schon yet wieder again dieser this FEHLER mistake unterlaufen.]]F crept in b. [[Dem the Präsidenten president zittern tremble die the HÄNDE.]]F hands (17) Was ist denn hier für ein Lärm? / What’s all the noise here? a. # [[Der the AUSSENSEITER outsider gewinnt.]]F wins b. [[Das the TELEPHON telephone klingelt.]]F rings The last missing ingredient of an information-structure based explanation for the definiteness effect and its exceptions is a categorization of definite NPs in terms of their discourse potential. De Kuthy (2002, sec. 6.5) discusses that one can distinguish definite NPs which function anaphorically and thereby have to be part of the background of a sentence from a second class of definite NPs which are used deicticly, endophorically or as a semantic definite.5 This second class can occur as the focus of a sentence. A closer look at the examples where the subject can be the Following Löbner (1985), we use semantic definite to refer to cases where the definite article is combined with a noun that represents a concept that necessarily only applies to one object, such as the weather, the moon, the president, or body parts like the heart. focus exponent, (16b) and (17b), shows that the definite NPs in those sentences are indeed instances of semantic definites. In conclusion, given Webelhuth’s generalization that a fronted verbal constituent has to be focused, such a constituent can only contain those definite subjects that can be a focus exponent, namely definite NPs which are used deicticly, endophorically or as a semantic definite. 3 An HPSG analysis We couch our analysis in the encoding of information structure in HPSG developed in De Kuthy (2002, sec. 6.6). Her approach builds on the proposal of Engdahl and Vallduvı́ (1996) in which a focus-background structure for every sentence is build up compositionally from the focus-background structures of its subparts. The information structure is encoded in the attribute INFO-STRUC that is appropriate for signs and has the appropriate features FOCUS and TOPIC, with lists of so-called meaningful expressions (semantic terms, cf. Sailer 2000) as values. The background of a sentence in De Kuthy’s approach is defined to be that part of the logical form of the sentence which is neither in focus nor in topic. This characterization of background closely resembles the definition of background employed by the so-called structured meaning approaches to focus of Stechow (1981), Jacobs (1983), or Krifka (1992). The INFO-STRUC value of a simple sentence with the focus as indicated in (18) is thus structured as shown in figure 1. (18) Peter Peter [[liest reads ein a BUCH.]]F book