General Linguistics Colloquium (SoSe 2022)

Immediately preverbal focus placement, common especially in V-final languages, has been shown to result from different syntactic configurations cross-linguistically. In this paper, we offer a unified account, based on the prosodic requirements of preverbal foci. We do so by bringing together two independent existing proposals, Focus-as-Alignment (Féry 2013) and flexible Intonational Phrase (ɩ)-mapping (Hamlaoui & Szendrői 2015) and applying the resulting analysis to a number of unrelated Eurasian languages.

This talk discusses the ideas proposed in Wolfe (2015) about the nature of superficially V1 clauses in Old Romance. I will focus on data from Old Italian, which has been claimed to display a relaxed V2 grammar along with other Old Romance varieties (Benincà, 2013; Poletto, 2014). Here, V2 is defined as a combination of two syntactic operations:

• Finite verb movement to C at least in main clauses;
• Obligatory presence of an EPP-like diacritic on C.

Wolfe claims that the attestation of V1 main declarative clauses in Old Romance is still compatible with the idea that its V2 grammar includes an obligatory EPP feature on C. Following ideas already put forth in the literature, he proposes that V1 main clauses result from merging a null (topical) element in first position. This marked syntactic option is also sometimes available in Germanic V2 languages (see for instance Sigurðsson, 2011). To support his proposal, Wolfe seeks to show empirically that V1 declarative clauses are restricted to three functional types. Given theoretical considerations and a data collection from corpora, I will try to show that the category of Topic Continuity V1 is particularly problematic for Old Italian, and that (at least a subset of) V1 clauses should be accounted for without postulating any null element in the left periphery, hence also without an obligatory EPP feature on C. This returns a simpler theory, a better descriptive account of the data and one less potential bias for the future research on Old Romance word order.

References
Benincà, P. (2013). Caratteristiche del V2 romanzo: Lingue romanze antiche, ladino dolomitico e portoghese. In E. Bidese & F. Cognola (Eds.), Introduzione alla linguistica del mòcheno (pp. 65–84). Rosenberg & Sellier.
Poletto, C. (2014). Word order in Old Italian. Oxford University Press.
Sigurðsson, H. Á. (2011). Conditions on argument drop. Linguistic Inquiry, 42(2), 267–304.
Wolfe, S. (2015). Verb-initial orders in Old Romance: A comparative account.Revue roumaine de linguistique, LX(2–3), 148–172.

In this talk, I look at the distinction between mere degree and measurement in questions. “Standard” degree questions tend to be ambiguous between the two readings (e.g. the English how-type), whereas in (dialectal) Norwegian and Icelandic, unambiguous measurement questions form a special category, subject to particular semantic and syntactic restrictions. In this regard, my main focus will be on one subtype of this set, a wh-less question form unambiguously asking for precise degrees (i.e. measurements):
   (1) Er  dem   gammel    (sånn    circa)?
       are  they   old             so         circa
       “How old are they (approximately)?”
I discuss some new data points in order to explore how these might shed light on what the morpho-syntactic makeup of this question form is.

Adnominal pronoun constructions (e.g. English "we linguists") display crosslinguistic variation in various dimensions, but detailed empirical studies on the phenomenon are less commonly found than on other aspects of nominal syntax. This session brings together investigations on Danish, German and English APCs.
Andrea Schröter is going to discuss results from her BA thesis on the properties of APCs in Danish, making use of elicited and corpus data.
Georg Höhn will present some initial results from a study on the distribution of APCs in two corpora of German (DWDS) and English (BNC), both around 100 million tokens. There will be space for discussion of methodological considerations and challenges involved in the data collection.

Switch reference (SR) systems are defined as morphological paradigms of forms obliging a choice between morphemes requiring identity or non-identity between a clause's subject and that of another clause. While not unquestioned in prior work, SR has often been regarded as being functionally motivated by the need for reference resolution, though admittedly having often become generalized beyond this. In this talk, I present a new sort of challenge to this picture through a case study of SR in A'ingae (an endangered isolate of Amazonian Ecuador/Colombia). We identify three uses of the A'ingae SR morphemes same-subject -pa SS and different subject -si DS, exploring the syntax and semantics of each: (i) bridging clause linkage in the sense of Guérin & Aiton 2019, (ii) clause-chaining, and (iii) causal and temporal adverbial clauses. We show that these three constructions are syntactically and semantically diverse, but that they are unified by the kinds of discourse coherence relations they establish. SR in A'ingae therefore serves not simply to help resolve reference, but to signal particular kinds of coherence relations as well as interrelated information regarding reference. Beyond providing a unified understanding of these SR morphemes within A'ingae, the proposed generalization provides a novel hypothesis regarding the motivation of SR cross-linguistically (though we leave testing this hypothesis to future work).

Dialectometry can be seen as a subdiscipline of both dialectology and computational linguistics. The term was coined by Jean Seguy in the early seventies when he was the director of the Atlas linguistique de la Gascogne. Literally it means 'the measure of dialect'. By using dialectometry patterns in the dialect landscape can be revealed by measuring linguistic distances among local dialects by considering the realizations of a large number of linguistics phenomena. In this talk we focus on the Levenshtein distance as a technique for measuring aggregated distances in the sound components. Furthermore, we discuss some statistical techniques that help to reveal (geographic) patterns once the linguistic distances are found. The several techniques are demonstrated by using LED-A, a web app that makes dialectometry accessible to every linguist.

Saami people are the indigenous population of Scandinavia. The original Saami settlement area stretches to Central Scandinavia and overlapped with the assumed Proto North Germanic homeland. In 2000 Jurij Kusmenko and Michael Rießler published a study on the Saami origin of a few phonological innovations in Eastern Norwegian and Northern Swedish dialects, which evolved after the Common Scandinavian period as a result of Saami cultural assimilation and language shift-induced substratum interference. More potential substrate features where identified and described in later paper publications, including an older layer of contact influence affecting Common Scandinavian. A monograph on the topic was published by Kusmenko in 2008. Based on two substrate features, metaphony and preaspiration, my presentation outlines the research historical context of this hypothesis, which deals with prominent North Germanic phonological and grammatical features but doesn’t seem to be widely known.

Two main directions in Optimality Theory account for syntax-prosody interface: Match and Align. According to Match, the mapping from syntactic structure to prosodic structure always considers both the beginning and the end of the syntactic and prosodic constituents at once. This approach predicts that the recursive structure of syntax is reflected in prosody and that both edges are equally important. Align only considers one edge in each constraint; the other edge may be inserted by the symmetric constraint or by a constraint asking for exhaustive parsing. Align disfavours embedding of two constituents of the same level into each other in prosody and favours asymmetry of constituent's edges. I will first sum up the conclusion of Bellik, Ito, Kalivoda & Mester (2022), who show with phrases of three and four words that both Align and Match are needed in Japanese. For German, I will use my own experimental data to show that the right boundary in left-branching structures elicits a higher f0 register (upstep) than left edges that only elicit unmarked downstep. In other words, Align-XP-Right is needed if this asymmetric behaviour is to be reflected in the prosodic constituency. Thus, Match cannot replace Align, both are needed in German as well, though for different reasons than Japanese.

Background: As is well-known, Latin underwent a diachronic change from being a language that allowed as specific kind of hyperbaton –Left Branch Extraction (LBE)– to the modern Romance languages, which systematically disallow LBE, i.e., are subject to the Left Branch Condition (LBC, cf. i.a. Ledgeway 2012: 424). (1-a), from Latin, illustrates LBE of an AP, and it is likewise attested for possessive adjectives, demonstratives, wh-determiners (cf. (2)), and others, cf. Devine & Stephens (2006, 542ff). (1-b) from Italian showcases the LBC being operative.
(1) a. qui summami inter eos habet [ti auctoritatem ] (Caes. B.G. 6.13.8)
who highest.ACC.FSG among them.ACC has power.ACC.FSG
b. chi (*[più alta]) fra di loro ha [la più alta autorità ] (Italian)
who more high among of them has the more high authority
‘who has the highest authority among them’
(2) a. meoi tu epistulam dedisti [ti servo]
my.DAT you.NOM letter.ACC gave.2PERF servant.DAT
‘you gave the letter to my servant?’ (Plautus, Pseudolus 1203)
b. hanci cum habeat [ti praecipuam laudem ]
this.ACC since has.3SUBJ particular.ACC merit.ACC
‘since he has this particular merit’ (Cicero, Brutus 261)
c. quisi umquam [ti Graecus ] comoediam scripsit
which.NOM ever Greek.NOM comedy.ACC wrote.3PERF
‘Which Greek ever wrote a comedy?’ (Cicero, pro Flacco 27.65)
Analytically, this transition has been discussed in terms of the NP-DP-parameter (Bošković 2005a, 2005b, 2008), of “a change in the head-directionality parameter with a concomitant shift from specifier- to head-oriented syntax in the domain of functional structure on the other” cast in terms of roll-up derivations (Ledgeway 2012), and of phonological movement in conjunction with several other phenomena (Agbayani & Golston 2016). I offer a novel approach using recent developments in syntactic theory (Chomsky 2013, 2015/POP(+)).
Proposal and analysis: The intuition behind the current approach is the classical one (cf. Schwegler 1990) that morphological richness of nominal inflection underlies its LBE-characteristic in Latin and that the loss of morphological richness of nominal inflection yielded the LBC that characterizes modern Romance. Specifically, analogous to POP+’s notion of “strong” and “weak T”, I propose that the functional nominalizing head n (cf. Borer 2005) comes in two kinds, strong and weak, correlating with rich nominal inflection on the one hand, and poor nominal inflection on the other. This is captured in the following hypothesis:
(3) The Nominal Strength Parameter
a. strong n/nstr: Latin
b. weak n/nwk: Italian, Spanish, French …
(3) is a new instantiation of the classical Borer-Chomsky conjecture which localizes syntactic variation in properties of functional heads. The analysis is couched in the POP/+ framework in which the set-forming operation Merge applies optionally (i.e., freely), whilst phase-by-phase transferred syntactic representations meet 3rd factor principles of efficient computation (Minimal Search) and interface conditions. One of the latter is that every syntactic object requires a label. POP proposes that this requirement is achieved in a computationally efficient manner by the Labeling Algorithm LA. The first step in the derivation involves a category-neutral root R and a categorizer K (POP: 47) introducing an asymmetry: While R does not, K bears grammatical features and is thus identified by the LA. Thus, a nominal phrase comprises the nominalizing head n and R (cf., e.g. Borer 2005) yielding {n, R}=. By (3) richness and poverty of nominal inflection is labeling-relevant: {nstr, R} is instantly labeled by nstr, while {nwk, R} cannot be labeled but requires “support” by an XP, where the heads n and X share relevant features which are found by the LA (cf. POP+ on weak T). Following Sag, Wasow & Bender (2003), Chomsky (2007: 25-26), and, roughly, Leu (2014), I assume that determiner categories are internally complex, i.e., phrasal, here represented as AP, i.e. XP=AP. (3) delivers the following six predictions:
(4) a. *[=? nwk R]
(5) a. [=nP nstr R]
b. [=⟨f, f⟩ AP [nwk R]]
b. [=⟨f, f⟩ AP [nstr R]]
c. *APi … [=? ti [nwk R]]
c. APi … [=nP ti [nstr R]]
 in (4-a) is unlabelable since both nwk and R are too weak to label. In (4-b) nwk receives support by AP such that  is labeled by the shared feature borne by A and nwk, given here as a pair of f. In (4-c)  remains unlabeled because both nwk and R are too weak to label, and AP’s trace is invisible (cf. POP, Epstein et al. 2020). All of this contrasts with (5), where nstr is strong to label  in (5-a). In (5-b)  is labeled by the shared feature borne by A and nstr. Finally, AP-LBE does not undermine the labelability of  in (5-c), because nstr is strong.
(4) represents the modern Romance pattern and (5) represents Latin. (3-a) thus correctly and uniformly captures that in Latin, determiners are optional (cf. Quintilian (Inst. 1.4.19)), cf. (5-a) and (5-b), and LBE is possible (5-c), shown in (1-a) and (2). Moreover, (3-b) correctly and uniformly captures that modern Romance obligatorily requires the presence of a determiner category, cf. (4-a) vs. (4-b), exemplified by (6) from Longobardi (1994: 612):
(6) *(Un/il) Grande amico di Maria mi ha telefonato.
(a/the) great friend of Maria me have called-up
‘(A/the) good friend of Mary called me up.’
Moreover, the modern Romance languages are subject to the LBC as exemplified in (1-b), captured in (4-c). Relatedly, placement of possessive adjectives can be prenominal (7-a) and postnominal (7-b) in Latin since nstr can invariably label . This contrasts with modern Romance, where possessives must show up in “SPEC”-nwk (SPEC for exposition only) to support nwk for the identification of ’s label, as exemplified by French (7-c).
(7) a. Caesar suas copias subducit (Lat., Caes. B.G. 1.22.3)
Caesar.NOM his.ACC.FPL troops.ACC.FPL withdraws
b. copias suas Caesar […] subducit (Lat., Caes. B.G. 1.24.1)
troops.ACC.FPL his.ACC.FPL Caesar.NOM withdraws
c. César retire ses troupes (*ses) (French)
Caesar withdraws his troops his
‘Caesar withdraws his troops’.
I tentatively suggest that f is Case, perhaps gender and number. The diachronic passage is thus:
(8) nstr > nwk
The current approach thus sheds new light on “Latin’s recourse to synthetic strategies, in contrast to those of a predominantly analytic nature in Romance” (Ledgeway 2012: 424): The historical development from a language with rich noun inflection (rich case paradigms and declension classes) to a set of languages with poor or no noun inflection is reflected in (8). While Merge applies freely, it has to abide by the interface condition that every syntactic object be labeled by Minimal Search. This can be met in nstr-languages like Latin without ado with concomitant optionality of determiners and the option of LBE, while this can be met in nwk-languages only if {n, R}= is accompanied by AP to yield {AP, {n, R}}, deriving the obligatoriness of determiners and LBC-effects. Loss of case morphology was thus a reflection of the parametric shift from nstr to nwk, the underlying cause of rather dramatic syntactic changes in the nominal domain. I will comment on the analytical options regarding the grammaticalization of definiteness, given (3)/(8).
An implication: At least as far as the diachrony of the syntax of the nominal structure in Romance goes, the current approach casts doubt on economy principles that attribute historical syntactic clines to preferences of head over phrase status (cf. van Gelderen 2004, 2007, 2019, 2022; Dadan 2019; Meelen & Roberts 2022).
Selected references: Borer, H. (2005) Structuring sense (vol. 1): In name only. OUP. Bošković, Ž. (2005) Left branch extraction, structure of NP, and scrambling in J. Sabel and M. Saito (eds.) The free word order phenomenon: Its syntactic sources and diversity 13-73. Mouton de Gruyter, Berlin. Epstein, S. D., H. T. Kitahara & D. S. Seely (2020) Unifying labeling under minimal search in 'single-' and 'multiple-specifier' configurations, Arizona Linguistics Circle, U. of Arizona, 2019; published in Coyote Papers, U. of Arizona. Longobardi, G. (1994) Reference and proper names: A theory of N-movement in Syntax and Logical Form. LI 25(4).