In this paper I show that Swedish has a type of relative clause that doesn’t modifynominal expressions, contrary tomost descriptions/definitions of relative clauses.Instead this type of relative clause modifies evaluative predicates. The relativeclause has similarities to both control clauses and attributive relative clauses.I point out some issues that theoretical accounts of these relative clauses musttake into consideration, and also how current generative analyses fail to meetthese considerations. A promising route in such a generative framework seemsto be one where the head of the relative clause and the relative clause itself areindependently merged in the structure; the head as subject in the matrix clauseand the relative clause as an adjunct/complement to the predicate it modifies.What makes this relative clause different from NP modifying relative clauses isthe relation between the gap and the head.

This paper investigates tenseless finite clauses in Swedish. In certain contexts thefinite perfect auxiliary, ha ‘have’, is optional. These contexts are finite non-V2clauses and V2 clauses in which the V2 position is filled by a modal adverb, forexample kanske ‘maybe’. The analysis of these tenseless clauses is presented inthe constraint based lexicalist theory, lexical functional grammar. The analysisbuilds on, and develops, the one presented by Sells (2007).

This short communication is concerned with long object shift of reflexives in Swedish. Only 3rd person reflexives can shift across their antecedent. For some reason this is possible even if the antecedent is 1st or 2nd person as well, but certain requirements on the antecedent are necessary. This paper shows that neither a purely syntactic nor a purely semantic analysis can account for all the facts. Instead the best analysis seems to be one that makes use of Bonet's (1995) post-syntactic morphological processes: feature delinking, feature erasure and feature insertion.

This paper investigates the apparent near-optionality of the perfect auxiliary ha (‘have’) in Swedish. Restrictions on the near-optionality of ha-omission are typically recognized in the previous literature as lexical or structural. The paper scrutinizes the viability of these restrictions by means of corpus data. Most of these restrictions are shown not to hold. Instead, the paper defends the view that the optionality of the perfect auxiliary ha in Swedish is linked to whether the clause it occurs in requires a finite feature or not, and whether ha is the only element that can supply a value for this feature. The phenomenon of ha-omission is accounted for in the syntactic frame work LFG. It is shown that ha can only be omitted if another element in the clause can provide a value for the finite feature.

Vilka konstruktioner som Äar mÄojliga eller inte mÄojliga i ett sprºak kan tyckas varaen enkel frºaga som har ett objektivt svar. I praktiken visar det sig att det inte allsÄar sºa lÄatt att kategorisera konstruktioner som grammatiska eller ogrammatiska.Till att bÄorja med sºa verkar en tvºadelad klassi¯cering av uttryck i grammatisk{ogrammatisk vara alldeles fÄor grov. Dessutom har infÄodda talare olika uppfattningom vad som ska klassi¯ceras som grammatiskt och ogrammatiskt. Det ¯nns ocksºaen hel del andra faktorer som pºaverkar en talares bedÄomning av grammatikalitet.Det kan rÄora sig om sºa enkla faktorer som enskilda ord eller lÄangden pºa en mening.I en studie av passiva konstruktioner i ¯nska ¯ck experimentets deltagare, ¯nskamodersmºalstalare, rangordna 50 meningar efter grammatikalitet. Metoden somanvÄandes kallas 'magnitude estimation' och insamlingen av data skedde med hjÄalpav ett webbaserat datainsamlingsprogram, webexp. I det hÄar fÄoredraget kommerjag att presentera dels metoden, som ursprungligen anvÄants inom psykofysik fÄoratt mÄata fÄorsÄokspersoners upplevelser av fysisk stimuli, sºasom ljusfÄorÄandringaroch smÄarta, och dels det webbaserade datainsamlingsprogrammet. Jag kommerocksºa att diskutera de faktorer som verkar pºaverka deltagarnas uppfattning omgrammatikalitet i experimentet.

This paper presents the results from two studies on anaphoric reference to quantifying expressions (QEs) in Swedish, contributing to the current cross-linguistic discussion on this issue. For English it has been shown that the polarity of the QE (positive vs negative) determines the anaphoric set reference (to the referens set, REFSET, or to the complement set, COMPSET), while for Spanish it has been claimed that while REFSET interpretation is the default, the relative sizes of the two sets (REFSET and COMPSET) also matters. In Experiment 1, a semantic plausibility study. The results showed that for positive QEs, anaphoric reference can only be to the REFSET, while for negative QEs, it can only be to the COMPSET. Unlike in English and Spanish, REFSET continuations were categorically ruled out for negative QEs. To investigate whether the internal differences between QEs could be explained in terms of set size, we conducted Experiment 2, an estimation task. The results from this experiment showed that the size of the REFSET relative to the COMPSET was not a determining factor.

In this talk, we present the results from an Event Related Potentials (ERP) study on the processing of anaphoric reference to quantied expressions (QEs) in Swedish. QEs pick out proportions of possible members of some set for which a property holds. In (1a) and (1b), for example, some or few members of the set of students attended the lecture.

(1)  a. Some students attended the lecture.    b. Few students attended the lecture.

(2)  a. They found it very interesting.    b. They stayed at home instead.

Some and few differ in polarity: some is positive (upward entailing) while few is negative (downward entailing) (Peters and Westerstahl, 2006) and this is of importance when referring back to the QE using anaphoric expressions. The sentence in (1a) is naturally followed by (2a), which is about the students attending the lecture (the reference set, refset). The sentence in (1b), in contrast, is naturally followed by (2b), which is about the students not attending the lecture (the complement set, compset) (e.g. Moxey and Sanford, 1987). While (1b) can in fact be followed either by (2a) or (2b), (1a), cannot be followed by (2b).

Filik et al. (2011) is one of few studies of anaphoric reference to QEs in English using online measures (ERP). They report results for positive and negative QEs separately. Each type of QE shows refset and compset eects, as described above, on the disambiguating word. A larger N400 for compset vs. refset continuations for posivive QEs, and the opposite for negative QEs. However, they do not report any results for the contrast between positive and negative QEs in the compset condition. Since this is a very important condition and since it is known that QEs dier across languages (Nouwen, 2010; Tsai et al., 2014), we investigated this issue for Swedish.

160 experimental items of four sentences each were manipulated along two dimensions: polarity (positive vs negative quantifier, några vs få in (3)), and set (refset vs compset targeting disambiguating adjective, duktiga vs dåliga in (3)). The quantiers included were: några (`some’), få (`few’), många (`many’), inte många (`not many’), alla (`all’), inga (`no’), nästan alla (`almost all’), inte alla’ (`not all’).

(3) Några/Få studenter skrev bra på tentan    some/few students wrote well on the-exam

igår och att deCW var såyesterday and that they were so

duktiga/dåligaCW förbryllade professorn.good/bad confused the-professor

Unlike Filik et al. (2011) we found that positive QEs showed a pronounced positivity over the central region (FCZ, CZ, CPZ, PZ) in the compset condition relative to negative QEs, in the P600 time span (500{800 ms) after the onset of the critical word (the disambiguating adjective,`bad’). A linear mixed eects model analysis (LmerTest) showed a highly signicant main eect of polarity in the central region and the P600 time span above. We interpret this to mean that for positive QEs, a new discourse referent needs to be introduced following compset reference, while for negative QEs this discourse referentis already available (Burkhardt, 2007).

References

Burkhardt, Petra. 2007. The p600 reflects cost of new information in discourse memory. Neuroreport 18:1851 - 1854.Filik, Ruth, Hartmut Leuthold, Linda M. Moxey, and Anthony J. Sanford. 2011. Anaphoric reference to quantied antecedents: An event-related brain potential study. Neuropsychologia 49:3786 - 3794.

Moxey, Linda M., and Anthony J. Sanford. 1987. Quantiers and focus. Journal of semantics 5:189 - 206.Nouwen, Rick. 2010. What’s in a quantier? In The linguistics enterprise: from knowledge of language to knowledge in linguistics, ed. Martin Everaert, Tom Lentz, Hannah de Mulder, Øystein Nilsen, and Arjen Zondervan, 235 - 256. Amsterdam: John Benjamins.

Peters, Stanley, and Dag Westerståhl. 2006. Quantiers in language and logic. Oxford: Oxford University Press.Tsai, C.-Y. Edwin, Gregory Scontras, Kenneth Mai, and Maria Polinsky. 2014. Prohibiting inverse scope: An experimental study of Chinese vs. English. In Empirical Issues in Syntax and Semantics 10 , ed. Christopher Pi~non, 305 - 322. Paris: CSSP.

We present the results from an Event Related Potentials (ERP) study on the processing of anaphoric reference to quantified expressions (QEs) in Swedish. QEs pick out proportions of possible members of some set for which a property holds. In (1a) and (1b), for example, some or few members of the set of students attended the lecture.

(1) a. Some students attended the lecture.    b. Few students attended the lecture.

(2) a. They found it very interesting.    b. They stayed at home instead.

Some and few differ in polarity: some is positive (upward entailing) while few is negative (downward entailing) (Peters and Westerståhl, 2006) and this is of importance when referring back to the QE using anaphoric expressions. The sentence in (1a) is naturally followed by (2a), which is about the students attending the lecture (the reference set, REFSET). The sentence in (1b), in contrast, is naturally followed by (2b), which is about the students not attending the lecture (the complement set, COMPSET) (e.g. Moxey and Sanford, 1987). While (1b) can in fact be followed either by (2a) or (2b), (1a), cannot be followed by (2b).

Filik et al. (2011) is one of few studies of anaphoric reference to QEs in English using online measures (ERP). They report results for positive and negative QEs separately. Each type of QE shows REFSET and COMPSET effects, as described above, on the disambiguating word. A larger N400 for COMPSET vs. REFSET continuations for positive QEs, and the opposite for negative QEs. However, they do not report any results for the contrast between positive and negative QEs in the COMPSET condition. Since this is a very important condition and since it is known that QEs differ across languages (Nouwen, 2010; Tsai et al., 2014), we investigated this issue for Swedish.

160 experimental items of four sentences each were manipulated along two dimensions: polarity (positive vs negative quantifier, några vs få in (3)), and set (REFSET vs COMPSET targeting disambiguating adjective, duktiga vs dåliga in (3)). The quantifiers included were: några (‘some’) ,få (‘few’), många (‘many’), inte många (‘not many’), alla (‘all’), inga (‘no’), nästan alla (‘almost all’), inte alla’ (‘not all’).

(3) Några/Få studenter skrev bra på tentan igår och att de_CW var så duktiga/dåliga_CW förbryllade professorn.

some/few students wrote well on the exam yesterday and that they were so good/bad confused the professor.

There were four lists with 40 sentences from each condition. Each participant (29 in total, results reported below based on the first 13) only saw one sentence from each item, but saw all types of manipulation. In total, each participant read 400 sentences (160 test items, 240 fillers).

Unlike Filik et al. (2011) we found that positive QEs showed a pronounced positivity over the central region (FCZ, CZ, CPZ, PZ) in the COMPSET condition relative to negative QEs, in the P600 time span (500–800 ms) after the onset of the critical word (the disambiguating adjective, ‘bad’). A linear mixed effects model analysis (LmerTest) showed a highly significant main effect of polarity in the central region and the P600 time span above. We interpret this to mean that for positive QEs, a new discourse referent needs to be introduced following COMPSET reference, while for negative QEs this discourse referent is already available (Burkhardt, 2007).

This article reports the results from an ERP study on the processing of anaphoric reference to quantifying expressions in Swedish (e.g. Many students attended the lecture and that they were present was noted). Negative quantifiers (e.g. few) differ from positive quantifiers (e.g. many), in allowing anaphoric expressions to target either the ref(erence) set (‘students attending the lecture’) or the comp(lement) set (‘students not attending the lecture’), while positive quantifiers only allow Refset continuations. Results from the present study show that negative quantifiers give rise to an enhanced frontal negativity at the anaphoric pronoun in the negative condition, relative to positive quantifiers. At the critical word disambiguating between a Refset and Compset reading, we found P600 effects for the anomalous relative to the non-anomalous conditions. We interpret the frontal negativity found with negative quantifiers as an indication of referential ambiguity interfering in the processing of anaphoric reference.

Noun phrases involving relative clauses are assumed to universally comprise syntactic “islands” for extraction, but Swedish exists as a possible exception. Using eyetracking while reading, we investigated whether extraction from Swedish restrictive relative clauses (RCE) ([such old wheelbarrows]1 saw I a man that always washed __1 with benzine...) elicit similar processing costs as extractions from non-restrictive relative clauses, which are known to comprise strong islands (StrongIs); or if they pattern closer to extractions from non-island constructions (NonIs). We also examined to what extent non-linguistic variables (working memory WM, verb-object frequency, and pragmatic-fit) contribute to such differences. Results from a mixed models analysis of the embedded verb (washed) and spillover region (with...) suggest that in early measures, both RCE and NonIs show facilitation relative to StrongIs, but in late measures, RCE patterns closer to StrongIs as WM and pragmatic-fit increase, suggesting that Swedish RCE acceptability is partly dependent on non-linguistic factors.

Relative clause extractions generally yield so-called island effects (degraded judgments) across languages (Ross 1967), (1). However, Swedish and the other Mainland Scandinavian languages comprise famous exceptions to this pattern (e.g. Engdahl & Ejerhed 1982), cf. (2).

(1) *Those kind of books I know a girl that writes.

(2) Såna böcker känner jag en tjej som skriver.

The origin of island effects is under debate. Sprouse et al. (2012) found superadditive effects in judgments of island violations in English – combined effects greater than the sum of the individual costs for extraction and complexity (island structure). Because no correlation was found between participants' memory span (measured via serial recall and n-back) and the superadditive effects, they concluded that island effects must be derived from violations of syntactic constraints rather than processing limitations. In contrast, Hofmeister et al. (2014) did find a correlation between working memory span (measured via reading span) and superadditivity in a rapid serial visual representation experiment. They also found superadditive effects in fully grammatical, but hard to process sentences, suggesting that processing factors do play a role in superadditivity and hence in island effects.

In our talk, we report on the results from an acceptability judgment experiment that was designed to look for superadditive effects in judgments of relative clause extractions in Swedish, to help determine whether these structures – even though claimed to be acceptable – behave like other island violations with regard to superadditivity. If they do, then this would reduce some of the exceptionality of their status. In the experiment, we investigated the (super)additive effect of clause type (TC vs. RC) and extraction type (long vs. short) on acceptability ratings, and tested whether any such effects correlate with memory span, as measured via an n-back and a reading span task. Our expectation was that the latter test would provide a much better marker of WM than the former.

Det var Anna som anmälde en kille som snattade såna chokladkakor i godisaffären. RC + ShortExt Det var Anna som anmälde att en kille snattade såna chokladkakor i godisaffären. TC + ShortExt Det var såna chokladkakor som Anna anmälde en kille som snattade i godisaffären. RC + LongExt Det var såna chokladkakor som Anna anmälde att en kille snattade i godisaffären. TC + LongExt

This brief article investigates the restrictions on Mainland Scandinavian relative clause extraction that have figured in the literature on island constraints. The conclusion is that none of these restrictions can be regarded as constraints on relative clause extraction per se and therefore that the peripheral status standardly assigned to Mainland Scandinavian relative clause extraction cannot be maintained.

Both (1) and (2) say that some undefined, small number of students went to the lecture, but the sentences differ in what type of quantifier is used: Några (‘some’) is a positive (upward entailing) quantifier, while få (‘few’) is a negative (downward entailing) quantifier (Peters and Westerståhl, 2006).

(1) Några studenter gick på föreläsningen.    some students went to the-lecture

(2) Få studenter gick på föreläsningen.    few students went to the-lecture

When referring back to ‘the students’ in (1) and (2), a difference in which students are referred to can be detected. (1) is naturally followed by (3), which talks about the students attending the lecture (the reference set), while (2) is naturally followed by (4), which talks about the students not attending the lecture (the complement set) (e.g. Moxey and Sanford, 1987). Whereas (3) can also follow (2) (although not as easily as (4)), (4) is only acceptable following (2).

(3) De tyckte den var väldigt intressant.    they thougth it was very interesting

(4) De stannade hemma i stället.    they stayed home instead

Anaphoric reference to quantifiers has been investigated extensively and we know that upward/downward entailment is one factor that influences set-reference in English (see e.g. Moxey and Sanford, 1987; San- ford et al., 1996; Paterson et al., 1998; Moxey et al., 2001; Moxey, 2006; Filik et al., 2011). However, to our knowledge there are no investigations of set-reference in Swedish. As quantifiers do not always behave the same across languages (Nouwen, 2010; Tsai et al., 2014), we investigated this issue for Swedish in a semantic plausibility study where we tested whether quantified expressions gave rise to reference set or complement set interpretations.

The material was manipulated along two dimensions: positive vs negative quantifier (några vs få in (5)), and refset vs compset targeting disambiguating adjective (duktiga vs dåliga in (5)). The quantifiers included were: några (‘some’), få (‘few’), många (‘many’), inte många (‘not many’), alla (‘all’), inga (‘no’), nästan alla (‘almost all’), inte alla’ (‘not all’).

(5) Några/Få studenter skrev bra på tentan igår och att de var så duktiga/dåligasome/few students wrote well on exam yesterday and that they were so good/bad

förbryllade professorn.confused the-professor

The results were that positive quantifiers with anaphoric reference to the compset were judged as anomalous, and negative quantifiers with anaphoric reference to the refset were judged as anomalous, although there was more variation with regard to the negative ones. As expected, these results are in line with previous studies.

Referenser

Filik, Ruth, Hartmut Leuthold, Linda M. Moxey, and Anthony J. Sanford. 2011. Anaphoric reference to quantified antecedents: An event-related brain potential study. Neuropsychologia 49:3786–3794.

Moxey, Linda M. 2006. Effects of what is expected on the focussing properties of quantifiers: A test of the presupposition-denial account. Journal of Memory and Language 55:422–439.

Moxey, Linda M., and Anthony J. Sanford. 1987. Quantifiers and focus. Journal of semantics 5:189–206.

Moxey, Linda M., Anthony J. Sanford, and E. Dawydiak. 2001. Denials as controllers of negative quantifier focus. Journal of memory & language 44:427–442.

Nouwen, Rick. 2010. What’s in a quantifier? In The linguistics enterprise: from knowledge of language to knowledge in linguistics, ed. Martin Everaert, Tom Lentz, Hannah de Mulder, Øystein Nilsen, and Arjen Zondervan, 235–256. Amsterdam: John Benjamins.

Paterson, Kevin B., Anthony J. Sanford, Linda M. Moxey, and Eugene Dawydiak. 1998. Quantifier polarity and referential focus during reading. Journal of Memory and Language 39:290–306.

Peters, Stanley, and Dag Westerst ̊ahl. 2006. Quantifiers in language and logic. Oxford: Oxford University Press.Sanford, Anthony J., Linda M. Moxey, and Kevin B. Paterson. 1996. Attentional focusing with quantifiers in production and comprehension. Memory &

Cognition 24:144–155.Tsai, C.-Y. Edwin, Gregory Scontras, Kenneth Mai, and Maria Polinsky. 2014. Prohibiting inverse scope: An experimental study of Chinese vs. English.

In Empirical Issues in Syntax and Semantics 10, ed. Christopher Pin ̃o ́n, 305–322. Paris: CSSP.

In this paper we report the results from two small-scale studies on monotone decreasing (negative) quantifying expressions (QEs) and their negative polarity properties. Some approaches to QEs view polarity item licensing and clause polarity (affirmation/denial) as distinct phenomena. Here, we investigate if this distinction holds for monotone decreasing QEa. In two studies (one on English and one on Swedish), we tested to what extent the abilities of QEs to license negative polarity items (NPIs) and to give rise to clause level denials go hand in hand. Native speakers of English and Swedish rated sentences with quantified subjects (e.g. few, not all) and NPIs (e.g. yet, until) or polarity sensitive material. The results show that other dimensions of negativity are subject to variation within each language but seem to follow each other to a high degree across the two languages, and that QEs with the same logical properties can still be negative to different degrees.

In this study we investigate anaphoric reference to negative and positive quantied expressions (QEs) in Swedish. Anaphoric reference to QEs has been extensively investigated in English. One important factor is the polarity of the QE [see e.g. 1, 2, 3, 4, 5]. Some and few, for instance, differ in polarity: some is positive (upward entailing) while few is negative (downward entailing) [6]. When referring back to the QE using an anaphoric pronoun (they below), the sentence in (1a), with some, is naturally followed by (2a), whichis about the students attending the lecture (the reference set, REFSET). The sentence in (1b), with few, in contrast, is naturally followed by (2b), which is about the students not attending the lecture (the complement set, COMPSET) [e.g. 1]. While (1b) can in fact befollowed either by (2a) or (2b), (1a), cannot be followed by (2b).

(1) a. Some students attended the lecture.b. Few students attended the lecture.

(2) a. They found it very interesting.b. They stayed at home instead.

Whether the COMPSET is available for anaphoric reference as default or is introduced via some kind of pragmatic inference is a matter of debate [e.g. 3, 4, 5, 7, 8]. Filik et al. [9] used ERPs to investigate reference to sets in English. The results were that both types of QEs showed REFSET and COMPSET effects. They found a larger N400 for COMPSET vs. REFSET continuations for positive QEs, and the opposite for negative QEs. Their conclusion was that following a negative QE there are more sets available for reference than following a positive QE. However, it should be noted that it is questionable if the effects that are reported are effects of a real N400-component or spurious negativities. As the authors admit, the experimental set up makes it diffcult to interpret some effects. In our study, 160 experimental items of four sentences each, as in (3), were manipulated along two dimensions: polarity (positive vs negative quantier, e.g. några vs få), and set (REFSET vs COMPSET targeting disambiguating adjective, e.g. duktiga vs dåliga).

(3) Några/Få studenter skrev bra på  tentan          igår      och att  de_CW1 var så     some/few students wrote well on  the-exam yesterday and that they were so     duktiga_CW2/dåliga_CW2 förbryllade professorn.     good/bad                    confused  the-professor

31 participants saw the 160 experimental items, distributed across four lists in Latinsquare design. ERPs were measured on two critical words the pronoun de ('they') (CW1) and a disambiguating adjective, (CW2) picking out either the REFSET (duktiga, 'good'), or the COMPSET (dåliga 'bad') [cf. 9].

We performed a linear mixed eects analysis of the data and obtained the following signicant results. On the pronoun, CW1, there was a sustained left anterior negativityin the time span 100-1500ms after CW-onset for negative QEs compared to positive QEs (set-focus is not determined at this point in the sentence). On the disambiguating adjective, CW2, there was an interaction of polarity and set in the posterior region in the P600 span (600-900ms) [10]: positive QEs showed more positivity in the COMPSET condition than in the REFSET; sentences with COMPSET focus (with the adjective dåliga 'bad' in (3)) showed more positivity in the positive QE condition than in the negative QE condition; sentences with REFSET focus (with the adjective duktiga 'good' in (3)) showed more positivity in the negative QE condition than in the positive QE condition.

We take the sustained left anterior negativity on CW1 (pronoun) to be an NRef effect [11, 12, 13]. This effect indicates referential ambiguity and thus indicates that negative QEs make both REFSET and COMPSET available for anaphoric reference. This contrasts with positive QEs that only make the REFSET available for anaphoric reference. On the adjective, CW2, there was no signicant distinction between negative QEs in the REF-and COMPSET conditions, indicating that either set can still be in focus. Positive QEs only make the REFSET available and it is therefore problematic to integrate an adjective focussing the COMPSET, as reflected in the P600-effect. Importantly, there were no differences between negative QEs with COMPSET focus and positive QEs with REFSET focus, on the one hand, and the reverse conditions (NEG-REF vs POS-COMP), on the other hand. We take this to indicate that COMPSET is the preferred focus for negative QEs in Swedish and that REFSET is the only possible focus for positive QEs. These results raise important questions about how negative QEs contribute to the construction and processing of the mental discourse model.

This paper is concerned with the SALIENCE of discourse referents from both a hearer and a speaker perspective in the context of negative quantifying expressions (e.g. not all, few) in Swedish. In three sentence completion studies we investigated what discourse referent is most salient from each of these perspectives and how salience is affected by the syntactic form/semantic type (entity vs. proposition) of the sentence topic in the continuation. We also investigated what morphosyntactic forms are used to refer back to referents of different information structural status. The studies were specifically set up to allow us to distinguish between the two perspectives on salience, unlike previous studies on English, which have only taken hearer-salience into account. The emerging picture of what discourse referent is most salient for negative quantifying expressions is therefore more complex than what previous studies have shown. (c) 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

In this talk, we present the results of a semantic plausibility study investigating systematic mismatches in anaphoric reference to quantified expressions (QEs) in Swedish. Sentences as (1)–(2) have a similar meaning but differ in their use of quantifier: some is a positive (upward entailing) quantifier, while few is a negative (downward entailing) quantifier [1].

(1) Some students attended the lecture.

(2) Few students attended the lecture.

Both (1) and (2) are sentences about students attending a lecture. However, when referring back to ‘the students’, a difference can be detected between the two sentences. (1) is naturally followed by a sentence like (3), which, like (1), is about the students attending the lecture (the reference set, REFSET). (2), on the other hand, is naturally followed by (4), which differs from (2) in being about the students not attending the lecture (the complement set, COMPSET) [e.g. 2]. While (2) can in fact be followed either by (3) or (4), is also a possible continuation of (2), the case where there is a mismatch, i.e. (4) following (2), is actually the preferred continuation [3]. For (1), the mismatch continuation ((4) following (1)) is not allowed.

(3)  They found it very interesting.

(4)  They stayed at home instead.

Anaphoric reference to QEs has been extensively investigated in English and one important factor influencing set-reference is the positivity/negativity of the quantifier [see e.g. 2, 3, 4, 5, 6]. As quantifiers do not always behave the same across languages [7, 8], we investigated this issue for Swedish in a semantic plausibility study where we tested whether quantified expressions gave rise to REFSET or COMPSET interpretations.

The material was manipulated along two dimensions: positive vs negative quantifier (några vs få in (5)), and REFSET vs COMPSET targeting disambiguating adjective (duktiga vs dåliga in (5)). The quantifiers included were: några (‘some’), få (‘few’), många (‘many’), inte många (‘not many’), alla (‘all’), inga (‘no’), nästan alla (‘almost all’), inte alla’ (‘not all’).

(5) Några/Få studenter skrev bra på tentan igår och att de var såsome/few students wrote well on the-exam yesterday and that they were soduktiga/dåliga förbryllade professorn.good/bad confused the-professor

A linear mixed model showed that positive quantifiers with anaphoric reference to the REF-SET were judged as semantically congruent, while they were judged as anomalous with ana- phoric reference to the COMPSET. For the negative quantifiers, the opposite pattern emerged: they were judged as congruent with anaphoric reference to the COMPSET but anomalous with the reference to the REFSET. There was also a difference between positive and negative QEs. The preferred continuation for positive QEs, the matching continuation, was rated as more con- gruent than the preferred continuation for negative QEs, the mismatched continuation. There were also internal differences within the groups of positive and negative quantifiers. More spe- cifically, sentences where the syntactic subject included the positive quantifier några (‘some’) were significantly different from sentences with the other positive quantifiers in the subject: when reference was made to the REFSET, the sentences were judged as semantically congruent to a lesser degree than for the other positive quantifiers. Similarly for the negative quantifiers få (‘few’) and inte alla (‘not all’): when reference was made to the COMPSET, the sentences were judged as semantically congruent to a lesser degree than for the negative quantifiers inga (‘no’) and inte många (‘not many’).

In a follow-up study, we investigated whether the relative size of the sets, in terms of number of members, influence anaphoric set interpretation. According to Zulaica-Herna ́ndez [9], the set with the largest number of members is the one most easily referred to. In relation to our results, this would mean that några (‘some’) should pick out a smaller REFSET than the other three positive QEs, and få (‘few’) and inte alla (‘not all’) should pick out larger REFSET than inga (‘no’) and inte många (‘not many’). To test these claims, we carried out an investigation using a questionnaire where each participant was instructed to write down the number they thought a quantifier corresponded to, given a fixed total number and a specific context, as in the following example [cf. 10]:

(6) There were 100 students in the auditory. QE of them had been there before. How many do you think had been there before? ANSWER:

The results were that några (‘some’) was taken to pick out a significantly smaller REFSET than the other positive QEs, and inte alla (‘not all’) was taken to pick out a significantly larger REFSET than inte många (‘not many’). However, få (‘few’) was taken to pick out a smaller, rather than bigger, REFSET than inte många (‘not many’). In addition, få (‘few’) and inte alla (‘not all’) were taken to pick out REFSET of very different sizes. Thus, these studies show that anaphoric reference to QEs in Swedish behaves as in English when it comes to polarity and also, unexpectedly, that the relative size of COMPSET and REFSET plays a role in the focussing one of the two sets.

[1]  Stanley Peters and Dag Westersta ̊hl. Quantifi- ers in language and logic. Oxford University Press, Oxford, 2006.

[2]  Linda M. Moxey and Anthony J. Sanford. Quantifiers and focus. Journal of semantics, 5:189–206, 1987.

[3]  Anthony J. Sanford, Linda M. Moxey, and Kevin B. Paterson. Attentional focusing with quantifiers in production and comprehension.Memory & Cognition, 24(2):144–155, 1996.

[4]  Kevin B. Paterson, Anthony J. Sanford, Linda M. Moxey, and Eugene Dawydiak. Quantifier polarity and referential focus dur- ing reading. Journal of Memory and Lan- guage, 39(2):290–306, 1998.

[5]  Linda M. Moxey, Anthony J. Sanford, and E. Dawydiak. Denials as controllers of neg- ative quantifier focus. Journal of memory & language, 44:427–442, 2001.

[6]  Linda M Moxey. Effects of what is expec- ted on the focussing properties of quantifi- ers: A test of the presupposition-denial ac- count. Journal of Memory and Language, 55 (3):422–439, 2006.

[7]  Rick Nouwen. What’s in a quantifier? In Martin Everaert, Tom Lentz, Hannah de Mulder, Øystein Nilsen, and Arjen Zon- dervan, editors, The linguistics enterprise: from knowledge of language to knowledge in linguistics, pages 235–256. John Benjamins, Amsterdam, 2010.

[8]  C.-Y.EdwinTsai,GregoryScontras,Kenneth Mai, and Maria Polinsky. Prohibiting inverse scope: An experimental study of Chinese vs. English. In Christopher Pin ̃o ́n, editor, Empir- ical Issues in Syntax and Semantics 10, pages 305–322, Paris, 2014. CSSP.

[9]  Iker Zulaica-Herna ́ndez. Complement ana- phora in Spanish: Reference and discourse re- lations. Journal of psycholinguistic research, 43(2):449–466, 2018.

[10]  Linda M Moxey and Anthony J Sanford. Prior expectation and the interpretation of natural language quantifiers. European Journal of Cognitive Psychology, 5:73–91, 1993.