-- KerstinFischer -- 14 Jan 2005

Workshop on Constructions and Language Change

• Organizers: GabrieleDiewald and AlexBergs
• Framework: ICHL in Madison

INTRODUCTION

Studies in diachronic linguistics increasingly acknowledge that linguistic change is highly context-dependent. Especially in its initial stages, linguistic change is tied not only to particular text types or registers, but also to specific morphosyntactic and semantic environments, i.e. to specific recurring patterns of co-present linguistic features, that is to say ''constructions''.

This workshop investigates and highlights the role of ''constructions'' in linguistic change. In doing so, the term ''constructions'' is deliberately understood to have a broad extension, i.e. to include, but not be limited to Construction Grammar proper. Thus, any constructional approach to language and language change is welcome.

Suggestions for topics to be addressed in this workshop include:

• The role of constructions as source(s) of linguistic change
• The role of constructions as product(s) of linguistic change
• Mechanisms of change within constructions
• Constructions and grammaticalization
• Constructions, frequency, and linguistic change
• Cross-linguistic constructional phenomena in linguistic change
• The definition and delimitation of the terms ''construction'', ''context'', etc.

Workshop programme:

CONSTRUCTIONS AND LANGUAGE CHANGE (XVII ICHL, Madison 2005)

Conveners Alexander Bergs, HHU Düsseldorf bergs@phil-fak.uni-duesseldorf.de

Gabriele Diewald Gabriele.diewald@fbls.uni-hannover.de

PAPERS (in alphabetical order)

1. Geert Booij (Vrije Universiteit Amsterdam) The progressive construction in Dutch: recursivity of grammaticalization patterns

2. Corrien Blom (Vrije Universiteit Amsterdam) The Grammaticalization of the Particle-Verb Construction

3. Wallace Chafe (University of California, Santa Barbara) Syntax as a Repository of Historical Relics

4. Gunther De Vogelaer (Ghent University) Construction genesis as degrammaticalisation? On doubled pronouns in Dutch

5. Martin Hilpert (Rice University, Texas) Where did this future construction come from? The case of Swedish komma att V

6. Mirjam Fried (Princeton University) Transpositional morphology and Construction Grammar: a diachronic perspective

7. Suzanne Kemmer & Martin Hilpert (Rice University, Texas) Constructional Grammaticalization in the Make-Causative

8. Marianne Mithun (University of California, Santa Barbara) Multi-phrasal constructions as borrowable starting points for grammaticalization

9. Tatiana Nikitina (Stanford University) Mixed categories and word order change

10. Jan-Ola Östman (University of Helsinki) Formulaic expressions in context: from minimal variation to language change

11. Peter Petré and Hubert Cuyckens (University of Leuven) Bedusted, yet not beheaded: The role of be-’s constructional semantics in its conservation

12. Malcolm Ross (Australian National University) Negative verbal clause construction in Puyuma

13. Elizabeth Traugott (Stanford University) Bits/?shreds of evidence for the grammaticalization of negative and positive polarity constructions: a Radical Construction perspective

14. Nina Azumi Yoshida (University of California, Los Angeles) The role of the topic-comment construction in the reanalysis of “things” in the Japanese modals -monoda and -kotoda

ABSTRACTS

The Grammaticalization of the Particle-Verb Construction

Corrien Blom c.blom@let.vu.nl

In (1) three examples of Dutch particle-verb constructions are given.

(1) PV main clause a. de schoenen inlopen Jan liep zijn schoenen in. 'to break in the shoes' 'John broke in his shoes.' (lit. in-walk) b. over de film napraten We praatten over de film na. 'to talk about the movie afterwards' 'We talked about the movie afterwards.' (lit. after-talk) c. het publiek toespreken Jan sprak het publiek toe. 'to address/talk to the audience' 'John addressed the audience.'(lit. to-speak)

According to most synchronic analyses, particle verbs (PVs), being syntactically separable (cf. (1)a-c), are phrases. Concerning their diachrony, PVs are generally assumed to be grammaticalizations of constructions with resultative phrases. Under these analyses, however, the characteristics of PVs given in (2) remain puzzling.

(2) a. PVs have conventionalized properties (e.g. particles have construction-specific meanings), but are nevertheless productively formed. b. PVs have divergent argument-structural and lexical-aspectual properties: (1)a is transitive and telic, (1)b is intransitive and atelic, and (1)c is transitive and atelic.

In my paper I illustrate that (2)a is accounted for by assuming PVs to be instantiations of partly lexicalized phrasal templates; constructions. These PV templates are part of a multiple-inheritance network, in which generalizations are stated at different levels. My hypothesis is that the diachrony of PVs accounts for (2)b: I assume that various elements showing up left-adjacently to the verb in older stages of Dutch (resultative phrases, adverbial phrases, postpositions) have been reanalyzed together with the verb as syntactic units. Subsequently, these syntactic units developed their own semantic and pragmatic properties, thus grammaticalizing into productive PV patterns; PV constructions. The different participant-licensing properties of the hypothesized particle sources (resultative phrases, adverbial phrases, postpositions) are assumed to account for the differences between the resulting PV constructions mentioned in (2)b. I present historical data supporting this diachronic hypothesis and discuss the structural and semantic changes involved in this grammaticalization development. In particular, I show how this development led to the formation of different PV subconstructions, which represent specific instantiations of a more general (dominating) PV construction.

The progressive construction in Dutch: recursivity of grammaticalization patterns

Geert Booij ge.booij@let.vu.nl

One of the ways in which progressive aspect can be expressed in Dutch is the aan het Infinitive-construction, as in

Jan is aan het fietsen John is at the cycle-INF ‘John is cycling’

This originally locative construction of the form ‘be + PP’ has developed into a periphrastic construction for progressive aspect. Similar constructions have developed in a number of Germanic languages. This construction may be qualified as a constructional idiom in which one position, the verbal infinitive, is still an open slot whereas the choice of preposition and determiner is fixed. The constructional idiom aan het V-inf is also part of larger constructions such as aan het V-inf slaan ‘to start V-ing’ in which the verb slaan ‘to hit’ has acquired a purely aspectual, inchoative meaning. I will show that Dutch features a range of such progressive constructions which support the claim that grammaticalization of lexical items into more grammatical ones takes place in the context of specific constructions. The periphrastic role of the progressive construction is clear from the fact that present participles in Dutch can only be used in attributive, not in predicative position:

De fietsende man ‘the cycling man’ De man *is fietsend / is aan het fietsen ‘The man is cycling’

Since periphrasis is usually associated with inflection, periphrasis is the phenomenon par excellence that shows how grammaticalization is dependent on constructions. The Dutch progressive construction is also interesting because the variable position can not only be filled by verbs but also by particle verbs and other varieties of separable complex verbs, as illustrated by:

Jan is zijn moeder aan het opbellen John is his mother at the up-phone-INF ‘John is calling his mother’

Particle verbs are themselves cases of grammaticalization of syntactic constructions, and thus this is another case of recursivity of grammaticalized constructions. Thus, the aan het Inf-construction can be used as a test for the construction status of certain word combinations. Finally, I will also present phonological evidence for the grammaticalization of the progressive construction.

Syntax as a Repository of Historical Relics

Wallace Chafe chafe@linguistics.ucsb.edu

I assume a view of language in which speakers begin with thoughts, which are organized into semantic structures, which are in turn modified by historical changes to become syntactic structures. Such changes include the creation of metaphors, collocations, and idioms, along with processes of grammaticalization and phonological change. I will illustrate the ways in which these processes interact with the English sentence "That's gonna bring down the house." Underlying this sentence is the thought of an event in which a performance of some kind will elicit an enthusiastic response from an audience. This event was categorized semantically as an instance of an agent performing a specific kind of action, and it was oriented as something that would occur in the future. Its syntactic structure resulted from successive applications of the following historical processes: the adoption of "house" as a metaphor for a theater audience, the establishment of "bring down" as a familiar collocation, the creation of the idiom "bring down the house" to express the idea of causing a theater audience to react with enthusiasm, the use of "be ... ing" to express the progressive aspect, and subsequently the use of "be going to" (later replaced by "be gonna") as a way of expressing future tense. The syntax of this sentence, and in fact of all sentences, can usefully be seen as a museum of such changes, each serving in its own way to distance syntax from semantics and thus from the thoughts being expressed and understood.

Construction genesis as degrammaticalisation? On doubled pronouns in Dutch

Gunther De Vogelaer gunther.devogelaer@ugent.be

In southern Dutch dialects, pronominal subjects can be doubled by another pronoun (De Vogelaer & Neuckermans 2002), giving rise to ‘subject doubling’-constructions such as (1).

(1) a. Ga-de gullie naar Brussel? go.2pl-youweak you(pl)strong to Brussels ‘Are you going to Brussels?’

b. Ge-gaat gullie naar Brussel. youweak-go.2pl you(pl)strong to Brussels ‘You are going to Brussels.’

c. Gullie gaat gullie naar Brussel. you(pl)strong go.2pl you(pl)strong to Brussels ‘You are going to Brussels.’

Both diachronic Middle Dutch data (Van Helten 1887; Vanacker 1963) and synchronic dialect-geographical data (SAND) lead us to believe that constructions such as (1a) are older than (1b) and (1c). (1a) can be considered the result of re-analysis of the pronominal clitic de as an agreement marker on the verb, making subject doubling a typical example of grammaticalisation. Some more recent developments, however, have caused de to enter in a complementary distribution with the weak pronoun ge (1b) and the strong pronoun gullie (1c). The choice of a particular form is governed by pragmatic factors (Nuyts 1995). Hence an originally morphosyntactic phenomenon has turned into a discourse phenomenon, providing an example of what could be described as ‘construction genesis’. In my talk, I will try to account for the genesis of constructions such as (1b) and (1c), and discuss to what extent these data are compatible with the supposed unidirectionality of grammaticalisation.

References De Vogelaer, G. & A. Neuckermans (2002). Subject doubling in Dutch: a dialect phenomenon in cross-linguistic perspective. In: Sprachtypologie und Universalienforschung (STUF) 55, 234–258. Nuyts, J. (1995). Subjectspronomina en dubbele pronominale constructies in het Antwerps. In: Taal & Tongval 47, 43-58. SAND = Barbiers, S., H. Bennis, G. De Vogelaer, M. Devos & M. van der Ham (2005). Syntactic Atlas of Dutch Dialects. Volume 1: Pronouns, Agreement and Dependencies. Amsterdam: Amsterdam University Press. Vanacker, V.F.(1963). Syntaxis van gesproken taal te Aalst en in het land van Aalst in de XVde, de XVIde en de XVIIde eeuw. Belgisch interuniversitair centrum voor neerlandistiek. Van Helten, W.L. (1887). Middelnederlandsche spraakkunst. Groningen: Wolters.

Transpositional morphology and Construction Grammar: a diachronic perspective

Mirjam Fried mfried@princeton.edu

One of the defining features of Construction Grammar (CxG) is its assumption that grammar consists of networks of overlapping grammatical patterns organized around shared features (formal, semantic, pragmatic, prosodic, etc.); Fillmore 1988, Goldberg 1995, Croft 2001, Fried & Östman 2004. This paper tests the conceptual and representational apparatus of CxG by tracing diachronic relationships across constructions. Specifically, I will address the mechanisms of change within a particular (morphological) construction and the role of emerging abstract constructional patterns in grammatical change. The illustrative material is provided by a special participial form in Slavic as manifested in Old Czech (‘present active participial adjective’, PA; e.g. hledající ‘(the) one seeking’). As a particular type of a hybrid inflectional category (an internally verbal but externally nominal form within a regular verbal paradigm), the PA presents all the representational challenges that are inherent in ‘transpositional’ morphology. This concerns especially the form’s functional and categorial status, which also plays a central role in the form’s historical development: the PA underwent significant changes in its internal morphological structure, syntactic function, and syntactic behavior, moving from a richly polyfunctional and context-dependent category (cf. the ambiguity between predicative and attributive usage in ex. 1) to expressing primarily a modification function (2b) and marginally serving as an actor noun (3). I show that the overall shift can be best captured as proceeding along two mutually reinforcing dimensions, both calling for the notion of ‘construction’ as a conventionalized association between form and function: (i) re-calibrating the relative prominence of the verbal (2a) vs. nominal (2b) morphological features within the PA form vis-à-vis its syntactic function and (ii) the simultaneous emergence of a modification construction as a general grammatical pattern; the latter developed independently of PA but contributed to the strengthening of the PA’s modificational potential in certain contexts (shown in 2b). Using the tools provided by CxG (particularly the external/internal distinction and the manipulation of the attribute-value configurations, cf. Fried & Östman 2004), I suggest ways for a systematic representation of the diachronic shifts, which involve semantic, formal, and textual factors. Overall, the PA development illustrates how relatively complex grammatical patterns can make connections to other patterns, thus opening up paths toward reorganizing form-meaning associations along specific criteria in a motivated manner.

(1) nebo v duši < žádajúcí zlého > (nevende duch milosti) for into soul.ACC.SG.F desire.PA.ACC.SG.F evil.GEN.SG.N (...) (i) ‘(turn your mind toward God, seek him through good life…, for the spirit of mercy will not enter) into the souli if/when iti desires evil things’ (ii) ‘… into a soul which is desirous of evil things’ [a homily]

(2) a. uslyšel žáčka < dřéveřečený verš zpievajícieho > he.heard youth.ACC.SG.M aforementioned song.ACC.SG.M sing.PA.ACC.SG.M ‘(and when he again secretly entered the church on Friday,) he heard a youth sing that aforementioned song’ [popular entertainment] b. v tom žádajúcím a sčastném kostele in that.LOC.SG.M desire.PA.LOC.SG.M and happy.LOC.SG.M church.LOC.SG.M ‘(his body was laid to rest) in that beloved and happy church [of his]’ [chronicle] (3) slyšechme debs chodicieho ale nemohli sme yžádného vidieti we.heard noise walk.PA.GEN.SG.M but could.not AUX.1PL none see ‘we heard the noise of [one] walking around but we could not see anybody’ [autobiography]

References: Croft, William. 2001. Radical Construction Grammar. Syntactic theory in typological perspective. Oxford: Oxford University Press. Fillmore, Charles J. 1988. The mechanisms of ‘Construction Grammar.’ BLS 14: 35-55. Fried, Mirjam and Jan-Ola Östman. 2004. Construction Grammar: a thumbnail sketch. In M.Fried & J-O. Östman (eds.), Construction Grammar in a cross-linguistic perspective,11-86. Amsterdam: John Benjamins. Goldberg, Adele E. 1995. A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press.

Where did this future construction come from? The case of Swedish komma att V

Martin Hilpert hilpert@rice.edu

This paper is a diachronic study of the Swedish Future construction komma att V. The construction involves the motion verb komma ‘come’, the infinitive marker att, and a non-finite verb which denotes the predicted action, as in the following example. (1) Priserna för röntgenundersökning kommer också att öka. prices.the for X-ray.examination come also to rise ‘Prices for X-ray examinations will also increase.’

The construction has undergone a semantic change from ‘physical movement towards a goal’ towards the meaning of ‘prediction’ (Heine and Kuteva 2002:78). While there is broad agreement in the field that such a change must have occurred, there are different theories as to how it proceeded. This paper tests two hypotheses against historical corpus data. Bybee et al. (1994:270) state in their cross-linguistic survey of grammaticized future constructions that ‘all modal and movement future sources begin with human agents and move from the expression of the intentions of that agent to the expression of prediction’. This predicts that the earliest examples of the construction involve intentional human agents. The opposite is predicted by Dahl, who compares several European future constructions that derive from verbs of coming, finding that none of these involve the notion of intentionality: ‘At any rate, there is no evidence to suggest that the Germanic de-venitives ever expressed intention’ (2000:322). Drawing on corpus data from three periods of Swedish, this study tests whether intentionality used to be a semantic component of the komma att V construction. Early examples of the construction actually contain intentional subjects, albeit not in the modal construction that evolved into the modern future construction. The Old Swedish construction komma til at V was a minor use of the verb that referred to change rather than intentional movement of human agents. Thus there is evidence for Dahl’s claim. This paper explores the semantic development of the komma att V construction, and proposes a modified grammaticization cline (movement → change → future) to account for the Swedish komma att V construction, and possibly other de-venitive future constructions.

References Bybee, J.L, R. Perkins and W. Pagliuca. 1994. The evolution of grammar: Tense, aspect and modality in the language of the world. Chicago: University of Chicago Press. Dahl, Ö. 2000. ‘The grammar of future time references in European languages.’ In Ö. Dahl (ed.) Tense and Aspect in the Languages of Europe. EUROTYP 20-6. Berlin: Mouton de Gruyter. 309-328. Heine, B. and T. Kuteva. 2002. World lexicon of grammaticalization. Cambridge: CUP.

Constructional Grammaticalization in the Make-Causative

Suzanne Kemmer, Martin Hilpert kemmer@rice.edu

Diachronic corpus studies such as Carey (1994) and Israel (1996) provide evidence that the grammaticalization of lexical elements proceeds along with changes in these elements’ syntactic and semantic environments. Hence, grammaticalization and constructional change must be viewed as closely linked. Carey’s (1994) study of the English Perfect demonstrates how the grammaticalization of have into a perfect auxiliary corresponds to increasing co-occurrence with complement predicates of specific semantic classes. Using Traugott’s (1989) idea of localized pragmatic inferences in linguistic change, Carey argues that the grammaticalization of have is not solely a semantic change. Instead, conventionalization of contextual factors within local contexts brings about new formal, constructional specifications. Israel (1996) tracks the development of the English way-construction, finding not just the grammaticalization of way, but the emergence of a constructional pattern in response to changes in the frequencies of main verbs used in the construction. By licensing verbs of increasingly different types, a new construction emerges that is quite different from its source. Kemmer (1995, 2001) refers to such processes as constructional grammaticalization. In this study, we investigate the make-causative in the recent history of English. Using the Corpus of Late Modern English Texts (CLMET), we trace the changes that the construction has undergone in the past 300 years. Examples like She made him do it may suggest that the primary meaning of the construction is compulsion. However, we show that such examples appear rather late in the history of the construction and even today are still relatively infrequent. The construction begins as a contextually motivated extension of the resultative construction (e.g. make it white), being increasingly used to express mental states and processes (e.g. make him angry). This development goes along with formal changes; whereas early usages have mostly nominal and adjectival complements, verbal complements increasingly dominate in modern usage. The evidence suggests that the modern make-causative is not the general, ‘plain vanilla’ causative it seems; in fact, it carries a specific semantic/constructional profile defined by types of participants and types of caused events, whose characteristics have changed over time and distinguish it from all other causative constructions in English. References Carey, Kathleen. 1994. Pragmatics, Subjectivity, and the Grammaticalization of the English Perfect. Doctoral dissertation, UCSD. Israel, Michael. 1996. The Way Constructions grow. In Adele Goldberg, ed., Conceptual Structure, Discourse and Language. Stanford: CSLI Publications. Kemmer, Suzanne. 1995. Analogy in Syntactic Change: The Rise of New Constructions. Paper presented at ICHL 7, Manchester. Kemmer, Suzanne. 2001. Causative Constructions and Cognitive Models: The English Make Causative. First Seoul International Conference on Discourse and Cognitive Linguistics: Perspectives for the 21st Century, 803-846. Seoul: Discourse and Cognitive Linguistics Society of Korea. Traugott, Elizabeth C. 1989. On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 57: 33-65.

Multi-phrasal constructions as borrowable starting points for grammaticalization

Marianne Mithun mithun@linguistics.ucsb.edu

The potential effects of contact on grammatical change are becoming ever clearer. Though the literature on syntactic borrowing continues to grow, that on morphological borrowing remains sparse. Instances of the transfer of individual affixes are well known, yet there is relatively little discussion of the transfer of abstract morphological structure without the substance that would carry it. The gap is not surprising. Morphological structure is among the most automated, tightly integrated, systematic aspects of grammar. Nevertheless we find some intriguing parallel structures in two genetically unrelated languages of the Northwest Coast of North America. In both Coast Tsimshian (Tsimshianic) and Kwak’wala (Wakashan), grammatical relations are marked not necessarily on the predicate (classical head-marking), nor on the arguments themselves (classical dependent-marking), but on whatever word precedes each dependent.

Coast Tsimshian: Boas 1911a:356 Dm dzakda=sga ġibaw=ga FUTURE kill=COMMON.ABSENT.ERGATIV wolf=COMMON.ABSENT.ABSOLUTIVE haas-ga dog-DEMONSTRATIVE ‘The wolf [ERGATIVE] will kill the dog [ABSOLUTIVE].’

Kwak’wala enclitics: Boas 1911b:533 Dó:x’w-aLél=e Dzá:wadalalisa=xa élkwa. see-suddenly=PROPER.NOMINATIVE NAME=COMMON.ACCUSATIVE blood ‘Dzawadalalis [NOMINATIVE] saw the blood [ACCUSATIVE].’

The mystery is solved once we untangle the diachronic processes that create such morphological structure. In spontaneous connected speech in most languages, speakers tend to express just one new idea per intonation unit or prosodic phrase (Chafe 1994). A common device exploited to regulate the flow of information in many languages, particularly those with predicate-initial order, is a construction built on demonstratives. This demonstrative construction spans two prosodic phrases separated by a pause, the first consisting of a predicate followed by a cataphoric demonstrative, the second a larger referring expression. The sentence ‘They got theirs from the Hudson Bay Company’, for example, is rendered:

They.got.theirs.from.it this Hudson.Bay.Company.

Bilinguals easily carry such rhetorical options from one language into another. The construction involves constituents that exist in both languages: a predicate, a demonstrative, and a referring expression. Relative frequencies of stylistic choices are easily transferred as well. Once such stylistic tools become frequent, normal processes of phonological fusion easily lead to structures like those above.

References Boas, Frans 1911a. Tsimshian. Handbook of American Indian Languages Part.I. Bureau of American Ethnology Bulletin 40. Washington D.C.: Smithsonian Institution. 283-422 Boas, Frans 1911b. Kwakiutl. Handbook of American Indian Languages Part.I. Bureau of American Ethnology Bulletin 40. Washington D.C.: Smithsonian Institution. 423-558. Chafe, Wallace 1994. Discourse, Consciousness, and Time. Chicago: University of Chicago Press.

Mixed categories and word order change

Tatiana Nikitina tann@stanford.edu

Mixed categories are constructions which combine properties of two different categories but have a single head. Well-known examples are nominalizations, which have the distribution of noun phrases but often preserve certain verbal syntactic properties. Mixed categories have been analyzed syntactically as constructions composed of two different categories that share one head either at the functional level (e.g., extended head theory, Bresnan 1997) or at the structural level (e.g., lexical sharing, Wescoat 2002). This paper addresses the role of mixed categories and head sharing in word order change, resulting in consistent placement of heads across different categories. Based on an example of VO/OV variation in Niger-Congo, I suggest that constructions with a head sharing relationship between different categories (nominal and verbal) may show a tendency for adjacent placement of the shared heads, which is independent of the general preference for consistent head placement across categories. Adopting the nominal periphrasis hypothesis that relates the change from SVO to SOV word order in the Kru and Mande families of Niger-Congo to the use of nominalizations in periphrastic expression of aspectual meanings (Heine 1976; Claudi 1993), I analyze this change as motivated by the tendency for adjacent placement of shared heads in mixed categories (in this case, the nominal periphrastic constructions). The use of the periphrastic construction in Kru resulted in a change from head-initial to head-final ordering in infinitival VPs under the pressure from head-final nominal structures. I suggest that this change resulted in non-adjacent placement of shared heads at the I’ level (VP being head-final, and I’ remaining head-initial), which has led to a change in the positioning of finite verbs in Mande. I show that the analysis of this series of changes in terms of the preference for adjacency of shared heads explains a “quirk” of the Mande syntax—S-Aux-OV order in sentences with auxiliaries, combined with SOV in other sentences. I argue that constructions with mixed syntactic properties may function as triggers in the process of word order change, or channels through which analogical changes occur, and discuss typological implications of this analysis.

Formulaic expressions in context: from minimal variation to language change

Jan-Ola Östman Jan-Ola.Ostman@helsinki.fi

Language change and variability are particularly challenging for the universal aspirations of Construction Grammar (CxG; Fillmore 1989, Fried & Östman 2004) since they necessarily involve work on peripheral (hence potentially ‘irregular’) grammatical structures. I will briefly deal with four types of variability of formulaic expressions: (1) the varying realizations of Wellerisms in Finland Swedish (esp. the reinterpretation of a relative pronoun as a temporal conjunction: X, said Y who … > X, said Y when …); (2) the variability of certain correlative constructions in English (There’s networking, and then there’s NETworking > There’s NETworking, and then there’s networking); (3) the variability in the presence/absence of certain grammatical elements in block language (Screen will not stop child from falling out windows); and (4) the emergence of the use of the second person singular in a generic sense in Finnish. These constructions are either formulaic in themselves (1-2), or they have formulaic uses, i.e. they construe a formulaic interpretation by virtue of being used (3-4). In response to the general question of how constructions vary and spread and, in the process, change, and how this can be captured in CxG, I briefly deal with the causes for change in terms of linguistic contact but also with how we can model mechanisms of change. In the four cases, the same general mechanism is at work: a construction first takes on a new value for a given attribute (in terms of attribute-matrices) and, with time, the newly adopted value gives rise to a distinct new attribute. In all the four cases, however, the changes are due to different causes; the one in (1) is largely due to homonymy; the one in (2) is due to a shift in information structure typical in ads; the one in (3) is an analogical extension of headlinese; and (4) is due partly to language contact, and partly based on conversational implicatures. A constructional approach to language change has a number of serious implications for how to model language overall; for one, since the attributes are not a priori tied to modules like ‘syntax’ and ‘semantics’, the attributesthemselves are the sole pegs – on different levels – that need to be made reference to.

Bedusted, yet not beheaded: The role of be-’s constructional semantics in its conservation

Peter Petré and Hubert Cuyckens Hubert.Cuyckens@arts.kuleuven.ac.be

During the Old English (OE) period, inseparable prefixes (IPs) gradually disappeared (e.g. ā-, ge-, for-, tō-). This disappearance has often been attributed to the syntactic shift from OV to VO (e.g. Traugott 1982:250) and/or to the IPs’ semantic bleaching (cf. CHIL, I.377ff). These accounts, however, fail to explain why be-, for instance, has remained productive to this day (besaltified, unbesocked), whereas an IP such as tō- ‘asunder’, which has much more concrete and specific semantics than be-, has disappeared. We suggest that considering IPs as part of a construction (IPC), with its own constructional semantics, may help solve this problem. On the basis of diachronic corpus evidence, it is argued that what seems to lie at the basis of the different behavior/life span of IPs is their predicative vs. non-predicative origin – a predicative IP can be paraphrased with a secondary predicate (or particle) (e.g., hē TŌbræc þā clūsan ‘he broke the bars APART’; niht BEcumeþ ‘night comes BY’), whereas a non-predicative IP needs a paraphrase with a path PP (hie BErīdaþ hine ‘they ride COMPLETELY AROUND him = they surround him’) (cf. Blom 2004). Importantly, when during OE the shift to VO put IPs under pressure, a predicative IPC [IP-V (DO)] could easily be replaced by a functionally/conceptually equivalent construction with a particle, i.e., without changing its valency or constructional semantics – this factor of language change, termed ‘intraference’ in Croft 2000, applies more generally in language (see also the replacement of the that-clause with suasive verbs by the to-infinitive; cf. Los 1999). A non-predicative IPC [IP-V DO], however, could not do so without changing its semantics: replacing it by a construction with a PP would lose the semantic component ‘total affectedness’ (as Michaelis and Ruppenhofer 2001 and Dewell 1996 point out, non-predicative IPs in present-day German remain productive because any prepositional alternatives are inappropriate to convey saliency and affectedness of their complements). Initially, then, the conservative nature of the [be-V DO] construction lies with its specific constructional semantics, combining ‘path’ and ‘affected object’. These original [be-V DO] constructions were then subject to increasing generalization and host-class expansion (cf. Himmelman 2005), processes typical of grammaticalization. In the second part of the paper, then, we will describe the following development in detail: surrounding (deverbal, berīdan) > covering (begān ‘override’) > affecting (deverbal/denominal, behēawan ‘beat all over’) > furnishing (denominal/deadjectival, bispusen ‘provide with spouses’, bespectacled). At this point, then, the conservative nature of the [be-V DO] construction can be attributed to the fact that its type frequency has gradually increased and that the construction has therefore become more entrenched (cf. Bybee 2001 on the conservative effect of entrenchment). Our analysis goes beyond pointing out general bleaching tendencies and predicts that other OE IPs (as for instance over-) only remained productive if their constructions originally were similar to the [be-V DO] construction.

References BYBEE, Joan. 2001. Phonology and Language Use. Cambridge: Cambridge University Press. BLOM, Corrien. 2004. ‘On the diachrony of complex predicates in Dutch: predicative and non-predicative preverbs’. Journal of Germanic Linguistics 16 (1). 1-75. Cambridge History of the English Language (CHIL), volume 1. 1992. Cambridge: Cambridge University Press. CROFT, William. 2000. Explaining language change. London: Longman. DE SMET, Hendrik. 2005. ‘A corpus of Late Modern English Texts’. To appear in ICAME Journal. DEWELL, Robert B. 1996. ‘The separability of German über-: A cognitive approach’. In: Dirven René and Putz Martin (eds.). The Construal of Space in Language and Thought (Cognitive Linguistics Research, 8), 109-133. Berlin: Mouton de Gruyter. HIMMELMANN, Nikolaus. 2005.‘Lexicalization: opposite or orthogonal’. In: B. Wiemer, W. Bisang and N. Himmelmann (eds.). What makes grammaticalization?, 21-44. Berlin: Mouton de Gruyter. LOS, Bettelou. 1999. Infinitival complementation in Old and Middle English. Den Haag: Thesus. MICHAELIS, Laura A. – RUPPENHOFER, Josef. 2001. Beyond Alternations. A Constructional Model of the German Applicative Pattern (Stanford Monographs in Linguistics). Stanford, CA: CSLI Publications. TRAUGOTT, Elizabeth Closs. 1982. ‘From Propositional to Textual and Expressive Meanings; Some Semantic-Pragmatic Aspects of Grammaticalization’. In: W. P. Lehmann and Y. Malkiel (eds.). Perspectives on Historical Linguistics (Current Issues in Linguistic Theory, 24), 245-271. Amsterdam: John Benjamins.

Corpora:: The York-Helsinki-Toronto Parsed Corpus of OE, the Penn Parsed Corpus of Middle English, The Helsinki Corpus and the Corpus of Late Modern English Texts (De Smet 2005)

Negative verbal clause constructions in Puyuma

Malcolm Ross Malcolm.Ross@anu.edu.au

Puyuma (Austronesian, Taiwan) has separate intransitive and transitive verbal clause constructions. The verb is marked for transitivity. Intransitive S and transitive P are in the nominative, i.e. subject, case, whilst the A noun phrase (if any) is in the oblique case but is crossreferenced on the verb by a genitive proclitic. Schematically:

(1) ItrVerb ItrSubject:nom (2) gen=TrVerb TrSubject:nom (TrAgent:obl)

(Since S and P are both marked as subject, one might say the language is ergatively aligned, but other features belie this epithet.) A first or second person subject appears as a nominative enclitic attached to the verb. In this circumstance, the two constructions are:

(3) ItrVerb=nom (4) gen=TrVerb=nom (TrAgent:obl)

In the negative intransitive construction, the negator aDi precedes the verb, which retains its intransitive form. If there is a nominative enclitic, it is attached to aDi:

(5) aDi=nom ItrVerb

Thus the nominative enclitic appears to be a second-position clitic, and from these constructions one might expect to be able to predict the negative transitive construction, but one can't. The negator aDi precedes the verb as expected, but the verb itself has a special form found only in negative transitive clauses and the nominative enclitic remains attached to this verb, as in (4), not to the negator, as in (5), giving the negative transitive construction in (6):

(6) aDi gen=NegTrVerb=nom (TrAgent:obl)

The paper shows that this puzzling constructional disharmony is easily accounted for. The corresponding Proto Austronesian (PAn) constructions are readily reconstructed on the basis of data from other languages, and appear to have been in harmony with each other. The steps leading from the PAn constructions to Puyuma can be reconstructed, and I show that (6) reflects a PAn construction, although it is the seemingly disharmonic member of the set in Puyuma. It has become disharmonic because of changes in (4) which have, so to speak, left (6) stranded. I conclude by asking why Puyuma has retained this disharmony, rather than bringing (6) into line with the other constructions and creating the ‘expected' but non-occurring construction in (7).

(7) **aDi=nom gen=TrVerb (TrAgent:obl)

Bits/?Shreds of evidence for the grammaticalization of the negative and positive polarity constructions: a Radical Construction perspective

Elizabeth Closs Traugott traugott@stanford.edu

It has become customary to include “constructions” as well as “lexical items” in definitions and characterizations of grammaticalization (see e.g., Lehmann 1995[1982], Bybee, Perkins and Pagliuca 1994, Traugott 2003), but “construction” has remained largely undefined in this literature. The development of Construction Grammar in its various versions (e.g., Goldberg 1995, Kay and Fillmore 1999, Croft 2001) has provided the opportunity to refine the relationship between constructions and grammaticalization (for some suggestions see Croft 2001). I show how aspects of Croft’s Radical Construction Grammar can be used to account for the development of modifier constructions not only like kind/sort of (e.g. Denison 2002), but also like a bit/shred of (all derived from [NP1 [of NP2]]). Some general hypothesis about the construction, its origins, and constraints on its development are discussed.

References Bybee, Joan L., Revere Perkins, and William Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: Univ. of Chicago Press. Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford and New York: Oxford University Press. Denison, David. 2002. Semantic pathways with sort of, kind of, type of, and their relation to grammatical gradience. Paper presented at Stanford University, May 2002; see also History of the sort of construction family, ICCG2, Helsinki (http://lings.ln.man.ac.uk/info/staff/dd/default.html). Goldberg, Adele. 1995. Constructions: A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. Kay, Paul and Charles J. Fillmore. 1999. Grammatical constructions and linguistic generalizations: The What’s X doing Y construction. Language 75:1-33. Lehmann, Christian. 1995 [1982]. Thoughts on Grammaticalization. München and Newcastle: LINCOM EUROPA. Traugott, Elizabeth Closs. 2003. Constructions in grammaticalization. In Brian D. Joseph and Richard D. Janda, eds., The Handbook of Historical Linguistics, 624-627. (Blackwell Handbooks in Linguistics.) Oxford and Malden, MA: Blackwell.

The role of the topic-comment construction in the reanalysis of “things” in the Japanese modals -monoda and -kotoda

Nina Yoshida nyoshida@ucla.edu

In Japanese, there exist two formal nouns for expressing the meaning "thing" of English—namely, mono and koto. Reference grammars commonly explain their distinction as being governed by such opposing semantic notions as "concrete" versus "abstract" (Martin 1975, McGloin 1989) or "tangible" versus "intangible" (Makino and Tsutsui 1989). Both mono and koto function frequently as a complementizer noun in Japanese, the common pattern being that of [complement] clause + mono / koto + copula da. Past studies (Teramura 1981, Agetsuma 1991, Tsubone 1994, Fujii 1999) have noted the -monoda construction’s pragmatic effects of conveying a wide range of speaker emotive affect, such as nostalgic reminiscences, conviction toward a natural truth, deep-seated desires, disbelief, indirect commands, etc. In contrast, -kotoda has less varied uses, being limited to marking mild disbelief and indirect commands. Previous analyses, however, have neglected to explain what specific features inherent to -monoda (but not -kotoda) endow the construction with such far-reaching pragmatic capabilities. This study presents an alternative analysis to account for the discourse modal functions of mono and koto in Modern Japanese. It does so by first proposing that the opposing semantics of a "physically perceived / unrationalized" existence versus a "cognitively conceived / rationalized" existence are the key, definitive features distinguishing mono from koto in instances when they are to take on a referential reading. This is derived through an etymological examination of mono and koto’s meanings in pre-modern Japanese, as revealed by such terms as mononoke (‘evil spirit’), kotomuku (‘rationalize’), etc. It moreover proposes that when mono and koto are employed in the -monoda and -kotoda constructions where they receive a non-referential reading, the semantics signaled by the two “shifts” by way of metaphorical inferencing from denoting an entity with a spatial orientation, to one with temporal persistence. This reanalysis is claimed to arise as a result of -monoda and -kotoda’s syntactic conformities to the topic-comment (A wa B da) construction in Japanese. Finally, it suggests that the "incomplete" state of the rationalization process signaled by mono's core meaning is what imparts a highly subjective, personal "voice" to discourse modal –monoda.