Non-canonical comparatives: Syntax, semantics, & processing

Transcription

1 Non-canonical comparatives: Syntax, semantics, & processing ESSLLI 2018 Roumyana Pancheva, Alexis Wellwood University of Southern California August 16, / 34

2 processing nominal and verbal comparatives how do we categorize entities as objects or substances? quantity judgements with mass and plular NPs 2 / 34

3 objects vs. substances How do we characterize an entity as an object or substance? Objects are entities whose structure - the manner in which their constituents are organized - we infer to be non-arbitrary; their forms are at least partly constitutive of what they are. In contrast, we understand ubstances to have arbitrary structure. (Prasada et al. 2002) Figure 1: table vs. amount of solid wood? When the entity is understood to be a table, we take it to have the structure it does because it is a table. Not any arrangement of the same material would be categorized as a table. 3 / 34

4 objects vs. substances Prasada et al. 2002: experimental investigation Something (one or more) was taken out of a box, and then participants had to choose how they d describe it, where the options included a novel noun embedded in either mass or count syntax, (1). (1) Prasada et al (2002) forced choice linguistic options a. There is blicket / There is a piece of blicket. [mass] b. There is a blicket. [count] 4 / 34

5 objects vs. substances regularity of shape Are regularly shaped entities more likely to be construed as objects than are irregularly shaped entities? Figure 2: example of the irregularly and regularly shaped entities (Prasada et al. 2002) Significantly more count noun responses for the regularly shaped items (72% of trials) than the irregularly shaped items (23% of trials). 5 / 34

6 objects vs. substances repeatability Are pluralities of equally shaped (irregular) entities more likely to be construed as objects than single instances of the same entities? Figure 3: Example of stimuli in the multiple instance condition (Prasada et al. 2002) Significantly more count noun responses on the repeated trials (65% of trials) than on the unrepeated trials (44% of trials), and on trials with multiple identical items (60% of trials) than trials with multiple nonidentical items (28% of trials). 6 / 34

7 objects vs. substances Exp. 2: the same manipulations but in a between-subject design Figure 4: Count noun responses and arbitrariness of structure ratings (Prasada et al. 2002) 7 / 34

8 object vs. substances; events vs. activities Wellwood et al. 2018b: participants describe pictures and animations A star is cut into natural and unnatural parts A star traverses a path (in the form of a flower) and pauses at the center of the flower (natural breaks) or at random points along the path (unnatural breaks). (2) Wellwood et al 2018b forced choice linguistic options a. Static images (regular noun syntax) i. There is some gleeb. [mass] ii. There are some gleebs. b. Dynamic scenes (deverbal noun syntax) [count] i. The star did some gleebing. [mass] ii. The star did some gleebs. [count] 8 / 34

9 object vs. substances; events vs. activities Figure 5: Sample item: static image (Wellwood et al. 2018b) 9 / 34

10 object vs. substances; events vs. activities Figure 6: Sample item: two dimensional rendering of an animation (Wellwood et al. 2018b) 10 / 34

11 object vs. substances; events vs. activities An asymmetry in participants selections between the scenes with natural and unnatural divisions. Figure 7: results (Wellwood et al. 2018b) 11 / 34

12 object vs. substances; events vs. activities Wellwood et al. 2018a: replication of results with VP descriptions for animations A star traverses a path (in the form of a flower) and pauses at the center of the flower (natural breaks) or at random points along the path (unnatural breaks). (3) Wellwood et al 2018a forced choice linguistic options a. The star gleebed around a little. [non-distributive] b. The star gleebed again and again. [distributive] 12 / 34

13 object vs. substances; events vs. activities Wellwood et al. 2018a: replication of results with VP descriptions for animations Figure 8: results (Wellwood et al. 2018a) 13 / 34

14 quantity comparison: experimental evidence Recall the experiments of Barner & Snedeker 2005: quantity comparison judgment tasks with children (4;1-4;6) and adults scenes/pictures with two characters and two quantities of objects plural vs. object mass vs. sustance mass Figure 9: set up of Exp. 1 (Barner & Snedeker 2005) 14 / 34

15 quantity comparison: experimental evidence Barner & Snedeker 2005: quantity comparison judgment tasks with children (4;1-4;6) and adults scenes/pictures with two characters and two quantities of objects the role of plural vs. mass syntax, while keeping the lexical predicate the same Figure 10: set up of Exp. 3: flexible nouns (Barner & Snedeker 2005) 15 / 34

16 results: quantity comparison Barner & Snedeker 2005: plural syntax (more stones) results in cardinality comparison mass syntax (more stone) results in non-cardinality comparison Figure 11: results of Exp. 3: flexible nouns in count or mass syntax (Barner & Snedeker 2005) 16 / 34

17 categorization and quantity comparison: experimental evidence Barner & Snedeker 2006: categorization and quantity comparison judgment tasks with children (3;0-3;6) and adults scenes/pictures with an object, introduced with a novel word in count or mass syntax (4) Oh look, this is some/a fem. Have you ever seen any fem(s) before? Do you think you have some fem(s) at home? That is some/a nice fem, isn t it? Figure 12: set up of Exp. 1 (Barner & Snedeker 2006) 17 / 34

18 categorization and quantity comparison: experimental evidence Figure 13: example of simple and complex entities (Barner & Snedeker 2006) 18 / 34

19 results: categorization Barner & Snedeker 2006: count syntax presentation (e.g., (a fem) results in categorization of novel words as count ( extension by shape ) however mass syntax presentation (e.g., some fem) also sometimes resulted in categorization of novel words as count (more so by children than adults) Figure 14: results of Exp. 1: children and adults (Barner & Snedeker 2006) 19 / 34

20 results: quantity comparison Barner & Snedeker 2006: count syntax presentation (e.g., more fems) results in comparison by cardinality however mass syntax presentation (e.g., more fem) also sometimes resulted in comparison by cardinality (more so by children than adults) Figure 15: results of Exp. 1: children and adults (Barner & Snedeker 2006) 20 / 34

21 quantity comparison with younger children Barner & Snedeker 2006: replication of quantity comparison judgment tasks from Barner & Snedeker 2005 with children (3;0-3;6) scenes/pictures with two characters and two quantities of objects plural vs. mass syntax, while keeping the lexical predicate the same Figure 16: set up of Exp. 2 (Barner & Snedeker 2006) 21 / 34

22 results: quantity comparison Barner & Snedeker 2006: plural syntax (more stones) results in cardinality comparison mass syntax (more stone) typically results in non-cardinality comparison, though children also give cardinality judgments, and more so than the older children in Barner & Snedeker 2006 Figure 17: results of Exp. 2: flexible nouns in count or mass syntax (Barner & Snedeker 2006) 22 / 34

23 quantity comparison and object mass nouns: complications Grimm & Levin 2012: object mass nouns allow non-cardinality comparison Exp. 1: participants evaluated which counted as more furniture: five chairs a sofa, two chairs, a coffee table, and a bookcase (five items) all participants unanimously chose the heterogeneous set comparisons need not be based on component parts, but may involve the fulfillment of function 23 / 34

24 quantity comparison and object mass nouns: complications Exp. 2: varying contexts (e.g., at a gala event): A is wearing two gold bracelets, a diamond tiara, and a ruby and emerald necklace. (4 items) B is wearing three gold rings, a pearl necklace and a silver bracelet. (5 items) Who has more jewelry? 24 / 34

25 quantity comparison and object mass nouns: complications results vary, but all object mass nouns allow non-cardinality comparison Figure 18: results of Exp. 2 (Grimm & Levin 2012) 25 / 34

26 quantity comparison Scontras et al. 2017: Exp. 1: notionally count and notionally substance mass presence or absence of linguistic cue Figure 19: set up of Exp. 1: example of the no linguistic cue condition (Scontras et al. 2017) 26 / 34

27 results Scontras et al. 2017: linguistic cue present: link between plurality and cardinality linguistic cue absent: some flexibility with dimension of comparison Figure 20: results of Exp. 1: presence vs. absence of linguistic cue (Scontras et al. 2017) 27 / 34

28 quantity comparison: acquisition Odic et al. 2013: Interactions between children s grammatical knowledge of more and their (prelinguistic) abilities to quantify number and area. Figure 21: two alternative theories of the acquisition of more (Odic et al. 2013) 28 / 34

29 quantity comparison: acquisition Odic et al. 2013: children (2;0-4;0) (5) a. Is more of the goo blue or yellow? b. Are more of the dots blue or yellow? (children point or say the color) Figure 22: two alternative theories of the acquisition of more (Odic et al. 2013) 29 / 34

30 results 30 / 34

31 results 31 / 34

32 results Odic et al found that success at answering these questions improved with age at the same rate for the two dimension - number and area, but that the degree to which each individual was sensitive to the two dimensions was uncorrelated. They conclude that children acquire its sensitivity to the mass/count distinction simultaneously, but the estimation abilities they recruit are independent. That is, area estimation appears to proceed via different channels than number estimation. 32 / 34

33 References I Barner, David & Jesse Snedeker Quantity judgments and individuation: evidence that mass nouns count. Cognition 97(1) Barner, David & Jesse Snedeker Children s early understanding of mass-count syntax: individuation, lexical content, and the number asymmetry hypothesis. Language Learning and Development Grimm, Scott & Beth Levin Who has more furniture? an exploration of the bases for comparison. Mass/Count in Linguistics, Philosophy and Cognitive Science Conference; École Normale Supérieure, Paris, France. Odic, Darko, Paul Pietroski, Tim Hunter, Jeffrey Lidz & Justin Halberda Young children s understanding of more and discrimination of number and surface area. Journal of Experimental Psychology: Learning, Memory, and Cognition 39(2) Prasada, Sandeep, Krag Ferenz & Todd Haskell Conceiving of entities as objects and as stuff. Cognition / 34

34 References II Scontras, Greg, Kathryn Davidson, Amy Rose Deal & Sarah Murray Who has more? the influence of linguistic form on quantity judgments. In Patrick Farrell (ed.), Proceedings of Linguistics Society of America 2, 41, Washington, DC: LSA. Wellwood, Alexis, Susan J. Hespos & Lance Rips. 2018a. How similar are objects and events? To appear in Acta Linguistica Academica. Wellwood, Alexis, Susan J. Hespos & Lance Rips. 2018b. The object : substance :: event : process analogy. In Tania Lombrozo, Joshua Knobe & Shaun Nicholas (eds.), Oxford Studies in Experimental Philosophy, vol. II, chap. 8, Oxford UK: Oxford University Press. 34 / 34