ComSyn Lectures

With the new semester quickly approaching, we’re delighted to announce the program for the upcoming season of ComSyn talks. The timeslot remains Thursdays, 16:15-17:30 (CET). The first two talks are in Lipsius 1.28, the remaining six are in Lipsius 1.33.

The full program of speakers is as follows:

We’re looking forward to hosting this varied program of talks and hope to see many of you there!

Speaker: Thom Westveer (UvA)
Title: Partitive constructions meet gender agreement: Gender agreement mismatches in French and German partitives
Date: Thu 21 Dec
Location: Lipsius 2.17
Time: 16:15-17:30

In languages with overt gender agreement morphology, partitive constructions involving human referents may give rise to mixed agreement (cf. Corbett, 1991). In the superlative partitive in (1a-b), a female student is selected out of a set of female and male students.

(1)       a.         ??Le/La           plus     jeune   des       étudiants         est        Marie.

                        the.m/.f           comp   young  of.the   student(m).pl  is         Marie

            b.         ??Der/Die       jüngste            der       Studenten        ist        Marie.

                        the.m/.f           young.sup       of.the   student(m).pl  is         Marie

In French and German, this set of females and males is typically referred to using a generic masculine plural. Yet, should the superlative referring to the female student Marie take the feminine form, in accordance with its referent’s biological sex, or the masculine form, to agree with the grammatical gender of the set noun? Grammaticality judgements show that native speakers of both French and German prefer the superlative to take the feminine form, resulting in a gender mismatch between subset and set. 

            Interestingly, speakers of French do not accept a mismatch between set and subset in a quantified partitive (2a); they prefer the quantifier to take the masculine form. Speakers of German, by contrast, still prefer a mismatch (2b), although their acceptance seems to be lower than for superlative partitives.

(2)       a.         ??Une/Un       des       étudiants         est        Marie.

                        one.f/.m           of.the   student(m).pl  is         Marie

            b.         ?Einer/Eine     der       Studenten        ist        Marie.

                        one.m/.f           of.the   student(m).pl  is         Marie

These observations raise at least two questions that should be addressed by a theoretical account: (i) how to explain the contrast between quantified and superlative partitives, and (ii) how to explain the difference between French and German. 

Although the theoretical analysis of partitive constructions received considerable attention – at least for quantified partitives (e.g., Cardinaletti & Giusti, 2017) – gender agreement in partitives has, to the best of my knowledge, only been discussed by Sleeman & Ihsane (2016). They account for the contrast between quantified and superlative partitives in French by arguing that superlative partitives, but not quantified ones, have a more complex syntactic structure that allows for late insertion of a gender value, which can result in a gender mismatch between set and subset. While Sleeman & Ihanse’s (2016) analysis can explain the French data, it falls short when confronted with the German pattern. Furthermore, their analysis assumes that a partitive’s structure contains a PP, which seems inadequate for the German data, as canonical partitives in German do not contain a preposition, but exhibit genitive case marking of the set DP instead. 

In my talk, I will propose an alternative analysis of partitive constructions, which also attempts to account for gender agreement (cf. Westveer, 2021). The syntactic analysis of partitives builds on a proposal by Den Dikken (2006) for qualitative constructions involving of in English (e.g., an idiot of a doctor): I will assume both French de and German genitive case marking in partitives to be overt realisations of a relator element, which functions as a nominal copula. In partitives, this relator links subset to set, expresses a belong-type interpretation, and is realised as the head of a small clause (labelled PredP) (3a).

(3)       a.         [_PredP [_Spec e][_Pred de][_DP les étudiants]]

            b.         [_QP [_Q un][_PredP [_Spec e][_Pred de][_DP les étudiants]]] 

            c.         [_DP [_D le][_FP plus jeune][_PredP [_Spec e][_Pred de][_DP les étudiants]]] 

To account for the differences between quantified and superlative partitives, I will argue that in quantified partitives, a QP selects the small clause (3b), whereas in superlative partitives, it is selected by a more complex construction, involving at least a DP and some functional projection to host the superlative adjective (3c). Crucially, a superlative partitive is headed by a DP, a projection establishing a clear link to a referent in discourse, which increases the likelihood of taking into account the gender of the referent. In this, I follow Steriopolo & Wiltschko (2010), who claim D to be the location of what they label ‘discourse gender’.

Within the analysis I propose, two conditions mediate the possibility of a gender mismatch: (i) a structural difference between quantified and superlative partitives – as stipulated above – and (ii) the overtness of morphological gender cues on the set DP. That is, in the plural in German, gender is only marked on the noun, whereas in the plural in French, determiners and adjectives are marked for gender through agreement as well. This, I assume, may explain the difference in acceptability of gender mismatches between French and German. In a next step, I will discuss how the analysis does not only account for the general patterns, but can also accommodate (speaker) variation.

Selected references

Cardinaletti, Anna & Giuliana Giusti. 2017. Quantified expressions and quantitative clitics. In Martin Everaert & Henk C. van Riemsdijk (eds.), The Blackwell Companion to Syntax (2^nd edn.), 1-61. John Wiley & Sons. | Corbett, Greville G. 1991. Gender. Cambridge: Cambridge University Press. | den Dikken, Marcel. 2006. Relators and Linkers: The Syntax of Predication, Predicate Inversion, and Copulas. Cambridge MA: MIT Press. | Sleeman, Petra & Tabea Ihsane. 2016. Gender mismatches in partitive constructions with superlatives in French. Glossa: a journal of general linguistics 1 (1), 1-25. | Steriopolo, Olga & Martina Wiltschko. 2010. Distributed GENDER hypothesis. In Gerhild Zybatow, Philip Dudchuk, Serge Minor & Ekaterina Pshetoskaya (eds.), Formal Studies in Slavic Linguistics, 155-172. New York: Peter Lang. | Westveer, Thom. (2021). Gender mismatches in partitive constructions in French and German. Amsterdam: LOT.

Speaker: Elise Newman (MIT)
Title: When wh-phrases are their own interveners
Date: Thu 14 Dec
Location: Zoom & Lipsius 2.17
Time: 16:15-17:30

Much work on syntactic locality has shown that processes like wh-movement are subject to several kinds of locality restrictions. In addition to being sensitive to intervening wh-phrases, wh-movement must proceed successive cyclically through various points in the clause, and in some cases/languages, may not cross intervening arguments (see e.g. Branan and Erlewine (2022) for a recent overview). Sensitivity to intervening arguments is known to be quite fine-grained: according to insights from Keenan & Comrie (1977) and others, languages might differ with respect to what kinds of arguments count as interveners for a wh-element, and might also treat arguments vs. adjuncts differently.

In this talk, I propose that all of these locality restrictions and their various levels of granularity are interconnected. More specifically, I suggest that they reduce to a particular view of how selection influences the projection of category information from daughter nodes to their mothers. I show that by examining the nature of selection and projection, we can leverage the architecture of grammar to predict the requirement for wh-movement to be successive-cyclic: the projection rule makes it so that wh-phrases create their own barriers for extraction if their wh-features get too high, meaning they have to move outside the scope of their own features in order to extract. The theory entails that movement must be successive cyclic, but does not say through which positions. By varying the different allowed parameters in this theory, I show that it also captures variable sensitivity to the Keenan & Comrie hierarchy. Thus, the various locality requirements governing wh-movement can be reduced to basic principles governing selection and projection.

Speaker: Danfeng Wu (University of Oxford)
Title: Allomorphy of ‘One’ and ‘Two’ in Mandarin Chinese
Date: Thu 7 Dec
Location: Zoom & Lipsius 2.17
Time: 16:15-17:30

In many languages numerals may appear in different forms depending on the context, which Greenberg (1978) called the contextual and absolute forms respectively. Greenberg made the universal generalization that if a numeral has two forms, then all the lower numerals also do. He then mentioned Mandarin Chinese as an exception because while its numeral ‘2’ has two forms, he claimed that the lower numeral ‘1’ doesn’t. I argue for a different view about Mandarin ‘1’ – that is, ‘1’ actually has two forms just like ‘2’, despite their segmental identity. Then I argue that the two forms of ‘1’ and ‘2’ are not distinguished by use as Greenberg claimed for ‘2’, but rather by the morphophonological context: the contextual form appears when followed by overt material at the point of vocabulary insertion of the numeral, otherwise the absolute form appears. This generalization, together with the key assumption that vocabulary insertion proceeds bottom-up, leads to a particular structure for enumerating numerals like liǎng gè nǚhái ‘two girls’, where the Cardinal head liǎng ‘two’ takes the Classifier Phrase gè nǚhái as its complement. I will also provide novel evidence based on ellipsis and tone sandhi that suggests that in Mandarin complex enumerating numerals like ‘125 pears’, the complex cardinal phrase ‘125’ merges with the Classifier Phrase, supporting He (2015) and challenging Ionin & Matushansky (2016, 2018).

Speaker: Luciana Sanchez Mendes (Universidade Federal Fluminense, CNPq, Faperj) and Ana Clara Polakof (Universidad de la República, SNI)
Title: Expressive interpretation of exclamative indefinites in Brazilian Portuguese and Rioplatense Spanish
Date: Thu 9 Nov
Location: Lipsius 2.17
Time: 16:15–17:30

The focus of this talk is to present a semantic analysis of exclamative indefinites in Brazilian Portuguese (BP) and in Rioplatense Spanish (RpS) in sentences like (1) and (2). Both data can be used in both languages.

(1)       O         Pedro   comprou          UMA               CASA.                                    (BP)
            the       Pedro  bought             ind.fem.sg      house
            ‘Pedro bought a house (a special one).’                           (Sanchez-Mendes, 2013)

(2)       ¡El       niño     es         UN      VIVARACHO!                                              (RpS)
            the      child    is          one      smart
            ‘How smart the child is!’(= ¡Qué vivaracho (que) es el niño!) (Masullo, 2017, 109)

We follow Castroviejo’s (2007, for Catalan) and Rett’s (2008, 2011, for English) treatment of overt exclamatives and Masulo’s (2007, for Spanish) proposal for covert exclamatives and assume that these data involve a degree construction. Additionally, we assume that the high degree expressed by them involves an expressive component that is associated with a conventional implicature (as Castroviejo, 2008 forwh-exclamatives) rather than a quantity implicature (Rett, 2015).

Our formal analysis is based on Masullo’s (2017) proposal for Spanish, that covert exclamatives have an extreme degree feature associated with a Degree Phrase. We assume that this extreme degree evokes an evaluative intonation that encodes expressive content (CI) (Castroviejo, 2008) that, in the case of indefinites, is due to the presence of uninterpretable and interpretable expressive features (à la Gutzmann, 2019). The uninterpretable expressive feature is provided by a scalar lexical category that is the complement of the extreme degree phrase that will be the gradable noun in cases such as vivaracho (RpS) and an omitted adjective in cases of non-gradable nouns, such as casa (BP). In the case of gradable nouns, the interpretable feature is located with the exclamative since the indefinite article emulates the role of an intensifier. In the case of non-gradable nouns, the interpretable feature is present in the indefinite itself that projects a DP.

Speaker:  Mieke Slim (Max Planck Institute for Psycholinguistics)
Title: Are all, every and each all the same? The trajectory of universality in the development of English universal quantifiers
Date: Thu 12 Oct
Location: Lipsius 2.17
Time: 16:15–17:30

Most languages have multiple universal quantifiers, like all, every, and each. These quantifiers share a common core of universality (that is, they all quantify over a full set), but differ along other dimensions. In particular, the use of each is more restricted to that of all. Since each is distributive, it can only express that a predicate separately applies to each individual member of the quantified set. The quantifier all, however, can also pair a predicate to the quantified set collectively.

When children learn universal quantifier, they therefore need to figure out a quantifier’s universality but also their distributivity. Several studies suggest that children understand that quantifiers like all and each are universal from early on, and become sensitive to the differences in distributivity later on in development (e.g., Brooks & Braine, 1996; Pagliarini et al., 2012). In this talk, I will present the results of three experiments that paint a different picture.

Experiment 1 was a ‘Give-a-Quantifier’ task, in which English-speaking children were asked to give all/each/every/all/some/dax objects to a puppet. The results showed that 4- and 5-year olds consistently gave the full set of items, but even 7-year-olds do not consistently treat each as a universal quantifier.

In Experiment 2, we zoomed in on children’s understanding of each. Children were shown animations of Cookie Monster biting 0, 2, or 3 out of 3 cookies, and asked “Did Cookie Monster bite the/each/two/dax (of the) cookies?”. Again, this experiment showed that even seven-year-olds did not consistently give a universal reading to each.

Experiment 3 was another Give-a-Quantifier task, but now the quantifiers were put in object position. In this experiment, the children were asked to give “each/every/all/some/dax of the fish“ objects. Data collection of this experiment is still underway, but preliminary results show that children year-old children consistently assign a universal reading of each from the age of six.

Altogether, the results of our three experiments indicate that children acquire the universality of each later than they acquire the universality of all and every. This raises questions about how children represent the semantics of each, and how they end up acquiring full understanding of universal quantification.

Note: Data were originally collected by Roman Feiman (Brown University) and David Barner (University of California San Diego).

Speakers:  Sjef Barbiers & Irina Morozova (LUCL)
Title: How and why ONE is different
Date: Thu 28 Sep
Location: Lipsius 2.17
Time: 16:15–17:30

ONE has been regarded as different from the other cardinals by Borer (2005), Barbiers (2005), (2007), Her & Lai (2012), Kayne (2019), Corver (2021), among others. In this talk we aim to show that the distinct (morpho-)syntactic behaviour of ONE is the result of its lexical feature specification. We propose a novel structure for numerals containing nominal phrases, representing sets of units, where both Set and Unit are classifiers (1).

This way we aim for a unified treatment of quantitatives (Num + N) and partitives (Num of D) (cf. Martí-Girbau 2010 for a different implementation), building on the analysis of indefinite pronouns by Leu (2005) and Roehrs (2008). Starting with the observation that in languages like Russian or Finnish numerals in non-oblique environments assign genitive or partitive case to the nominal complement, we argue that the relation between the numeral and the noun is established by an (abstract) P. Crucially, ONE is different in not assigning case to N. By means of the structure in (1), we show that ONE is a classifier with an unvalued [Unit] feature, the value for which can be provided by the syntactic environment of ONE, e.g., [gender], [person], [location], [time], [kind], [entity]. Treating ONE as a classifier that individuates the noun and makes it countable builds on Borer (2005), Barbiers (2007), Her & Lai (2012), Kayne (2019), Corver (2021). Other cardinal numerals, on the other hand, are phrasal quantifiers with a specific value for the feature [Partitioning: Value]. They specify a classifier element which can be one of the classifiers in languages like Mandarin (Her & Lai 2012) or Hungarian (Csirmaz & Dékány 2014), gender inflection in Hebrew, plural inflection in Germanic languages like English or Dutch (Doetjes 1996/1998, Borer 2005, Corver 2021). We show that unlike Cardinals, Indefinite Numerals (IndNums) like MANY or FEW are classifiers generated in head positions and have the feature specification [Unit:_; Partitioning]. Additionally, we argue for an optional projection SizeP, right above UnitP, that hosts classifier-like elements denoting size (size adjectives in Mandarin (Luo et al. 2017) or Hungarian (Csirmaz & Dékány 2014), the diminutive suffix in Dutch (cf. De Belder 2011, Corver 2021, Borer 2005) or the suffix -om in Russian collective numerals) which allows us to explain the distribution of collective numerals like Dutch met z’n drietjes ‘with the three of us’ or Russian vpjaterom ‘in fivesome’. We show that the structure proposed neatly captures the distribution of ONE, Cardinals, IndNums with respect to their cooccurrence with prepositions (Dutch); and case assignment patterns as well as the availability of approximative inversion and functional classifier-like nouns (Russian).

We’re looking forward to a varied programme of ComSyn talks in the upcoming season, running from September to December. All talks will be held at the usual timeslot of Thursdays, 16:15-17:30 CET. Here’s the list of speakers and dates:

As always, looking forward to seeing you there!
—Maarten (on behalf of the ComSyn organizing committee)

Speakers:  Anikó Lipták & Crit Cremers (LUCL)
Title: Left node not raising: Word part ellipsis revisited
Date: Thu 22 Jun
Location: Lipsius 0.01 (note room change)
Time: 16:15-17:30

Abstract:
In this talk, we describe a novel ellipsis phenomenon that deletes the second part of compounds, stranding the left side of the compound in sentence-final position:

We refer to this phenomenon as “left part stranding” (LPS for short) and show that it occurs both in Dutch and Hungarian in very similar ways and can affect the second part of an N-N or A-A compound, or adjectival compounds of the type shown in (1). In addition to the fact that LPS curiously violates Lexical Integrity and cannot be classified as any known exception to this condition (such as coordination reduction of the first part of a compound in a second coordinand, cf. onderzoek doelstelling of __ belangstelling, or the second part of a compound in a first coordinand, cf. ijs __ en bruine beren, see Booij 1985), LPS is also curious in that it has many unexpected properties, which are atypical of any process of coordination reduction or right node raising.

• It is only possible under clausal but not phrasal coordination;
• it preferably occurs in compounds that are the adjectival or nominal predicate of a clause;
• it preferably (in some cases obligatorily) co-occurs with gapping or TP-ellipsis (such as fragments);
• the stranded left part is necessarily contrastive.

After inventorising its morphosyntactic and semantic properties, we explain this set of constraints on LPS with reference to semantic conditions on ellipsis identity.

Speaker:  Claudia Pañeda (University of the Basque Country)
Title: Wh-island effects are similar in English and Spanish
Date: Thu 25 May
Location: Lipsius 2.23/Zoom [online talk]
Time: 16:15-17:30

Abstract:
A-bar extraction from an embedded wh-question (e.g. ‘What did the politician want to know when they would reject _?’) is thought to yield unacceptability or ‘island effects’ in English, but not in Spanish (Torrego 1984). This contrast has been used to argue that the linguistic principles responsible for island effects are parameterized (see also Rizzi 1982). However, it is unclear to what extent the contrast holds: previous empirical work suggests that there are island effects in both languages (see e.g. Sprouse et al. 2012 for English and Pañeda et al. 2020 for Spanish), but that they can be small in Spanish in some cases (Pañeda & Kush 2021). One limitation of most previous studies is that they have tested island effects in each language independently, with different materials (cf. Ortega-Santos et al. 2018). To better compare the two languages, we ran three acceptability experiments in each language with lexically-matched items. In each experiment, we obtained judgments from around 100 participants on one type of wh-island (introduced by when, whether or why). Our design, based on Sprouse et al. (2011 et seq.) manipulated the distance between the two members of the A-bar dependency (long, short) and the structure where it was established (island, non-island), identifying island effects as a structure × distance interaction. The three clauses yielded clear island effects in both languages, and, unexpectedly, these were slightly larger in Spanish. This indicates there are no relevant cross-linguistic differences in effect size and suggests that it may not be necessary to posit parametric variation—as previous work has done—to explain the data from both languages. Our wh-island effects may be attributed to the violation of simpler, non-parameterized grammatical principles, or to processing limitations.