Retrieving and Processing the Syntax and Semantics of the Mass/Count Distinction Linnaea Stockall, E. Matthew Husband and Ashley Benatar 1 2 3

Hampshire College, School of Cognitive Science

Brown University, Cognitive Science Department

McGill University, Communication Sciences and Disorders

Background - representation and processing of mass and count nouns Question: How is the mass/count distinction encoded lexically?

Question: How is the mass/count distinction encoded morpho-syntactically?

Question: How are mass and count nouns processed in isolation?

Indefinite Determiner Plural Morphology Mass ∗eat a spinach ∗ eat spinaches eat spinach Count eat a noodle eat noodles ∗eat noodle Dual eat a zucchini eat zucchinis eat zucchini

[1] Quine 1960: Count nouns denote individuals. I Mass nouns are cumulative and homogeneous. I

[2] Krifka 2008:

I

[3] Gillon, Kehayia and Taler (1999) .

I

Question: How are mass and count nouns processed in sentential context? I

Mass nouns evoke slower lexical decision times than count nouns.

[5] Steinhauer, Pancheva, Newman, Gennari and Ullman (2001) .

[4] Taler and Jarema (2007) Partially replicated [3] finding for younger adults: low frequency mass nouns evoked slower LDTs than low frequency count nouns and dual nouns. No differences among high frequency items . For older adults (mean age 75), dual nouns evoked faster LDTs than mass and count, which were not different from each other

count nouns evoked larger frontal negativity ERP than mass nouns (this effect was distinct from an N400 response evoked by a plausibility manipulation)

.

Count Nouns are repre- Mass Nouns are represented sented by counting func- by measuring functions. tions.

Provisional Conclusions: I The mass/count distinction is encoded Count Noun: [DP the [#P the [ClP wines [NP wines ]]]] e.g. #: If #(x) = 1 and e.g. KG: If KG(x) = 1 and lexically, and mass nouns are more difficult [DP the [#P the [NP wine ]]] #(y) = 1, where x, y are KG(y) = 1, where x and y to retrieve/process than count nouns, Mass Noun: [ClP wines [NP wines ]] disjoint may overlap. although the effect seems to depend on age [NP wine ] and lexical frequency. Other accounts: Mass nouns do not consist of I count interpretation is linked to the presence of #P in the I Dual nouns either pattern with count nouns, individuals (Bloom, 1999), are inherently plural structure or may be associated with a polysemy (Chierchia (1998) or are specified only by world I mass interpretation arises when #P is absent from the knowledge (Gillon, 1996). advantage

I

[6] Barner and Snedeker (2005) .

adults and children (aged 4:1-4:6) judged a single large entity as ’more’ than multiple small entities for mass nouns, and vice versa for count nouns and object-mass nouns (furniture, silverware). For dual nouns, judgements depended on morpho-syntactic context.

Provisional Conclusions: The morpho-syntactic mass/count distinction has processing consequences - count nouns are associated with greater costs than mass nouns [5], and the interpretation of dual nouns is driven by grammatical context [6]

structure (Borer, 2005)

Experiment 1 - Single Word Lexical Decision

Experiment 2 - Self Paced Sentence Reading Question: How do roots combine with morphology and syntax to generate mass and count interpretations in sentences?

Question 1: Are mass nouns really associated with longer lexical decision latencies? Question 2: Are dual nouns lexically count nouns that become mass nouns in the right grammatical context (or vice versa), or are they lexically polysemous?

I

90 dual nouns . 2 x 2 determiner type x number morphology manipulation . matched for length, lexical frequency, and surface frequency .

MATERIALS:

I

MATERIALS:

24 native English speakers judged acceptability of 200 words in mass and count morpho-syntactic frames . we chose 120 words that ranged in length from 4-9 letters, in lexical frequency from 0.65 - 217/million (English Lexicon Project), and covered the mass/count spectrum evenly .

Plurality Determiner Type Sentence Plural Definite The country vineyard supplied the wines for the auction. Plural Bare The country vineyard supplied wines for the auction. Singular Definite The country vineyard supplied the wine for the auction. Singular Bare The country vineyard supplied wine for the auction.

Table: Count/Mass Rating Scale

-6 -5 -4 whale maggot apple count

-3 shark

-2 cord

-1 onion

0 beer

1 soda

2 coal

3 wine

4 metal

5 meat

6 wool mass

I

PROCEDURE: 60 participants - students at Michigan State University . word by word self paced reading .

PROCEDURE:

I

44 participants - students of Concordia University and McGill University (aged: 18-32) . single word lexical decision .

I

RESULTS:

RESULTS:

I

3 Level Analysis:a Looking more closely at the dual nouns: main effect of noun type manipulation (pMCMC = 0.012) and fre- most count-like dual nouns are slower to respond to than the most quency (pMCMC = 0.0001); no interaction (pMCMC >.1) mass-like dual nouns

On head noun: Transition from N to N+1: main effect of determiner (F1(1,59)=7.632, p=0.008, main effect of determiner (F1(1,59)=27.091, p <0.001, F2(1,89)=8.117, p=0.005) and main effect of plurality F2(1,89)=20.546, p <0.001) (F1(1,59)=72.350, p <0.001, F2(1,89)=57.128, p <0.001) no interaction interaction between determiner and plural (F1(1,59)=3.599, p=0.063, F2(1,89)=3.258, p=0.074)

CONCLUSIONS:

I

Mass nouns significantly slower to respond to than either count or dual, but no differences between count and dual . Dual nouns overall are processed in lexical decision as if they have the same kind of semantic representations as unambiguous count nouns . No evidence of a polysemy advantage for dual nouns .

I

CONCLUSIONS: mass terms (bare singular and plural nominals) are costly compared to count terms . bare singular mass are the most costly - reflecting their fully homogeneous model interpretation .

a

analyses are linear mixed effect models with centered predictors and random intercepts for subject and item.

Discussion although there are distributional differences between count nouns, such that some are more acceptable in mass environments than others, there is no categorical difference between count nouns and dual nouns I although nouns such as stone and potato are judged equally acceptable in both mass and count environments, in an online processing task, we find that mass environments are more costly to process than count I Further thoughts and future directions: . bare plurals: finer grained distinction: difference between having atoms and not (Rothstein (2007), Macdonald (2006) - difference matters for telicity)[see poster 2.36: Husband, E.M. & Stockall, L. Lexical Telicity? Processing evidence for and against verbal telicity for more complete discussion] . coercion: even the most homogenous of mass nouns can be made count with the right context and morpho-syntax (My company sells several different sands to beach developers.) . cross-linguistic variation: this research suggests count semantics is the more basic, default case, and that mass semantics is more restricted and difficult to process. Is this a fact about English? I

Selected References 6 Barner, D.,Snedeker, J. (2005) Quantity judgments and individuation: evidence that mass nouns count. Cognition, 97, p 41-66 3 Gillon, B., Kehayia, E. and Taler, V. (1999). The Mass/Count Distinction: Evidence from On-Line Psycholinguistic Performance. Brain and Language 68, 205-211. 2 Krifka, M (2008) Different Kinds of Count Nouns and Plurals. Syntax in the Worlds Languages III Freie Universitt Berlin 1 Quine, W. V. O. (1960). Word and object. Cambridge, MA: MIT Press. http://www.faculty.hampshire.edu/∼lstockall

5 Steinhauer, K., Pancheva, R., Newman, A. J., Gennari, S. & Ullman, M. T. (2001) How the mass counts: An electrophysiological approach to the processing of lexical features. Cognitive Neuroscience and Neuropsychology.12, 517, 999-206 4 Taler, V., Jarema, G. (2007) Lexical Access in Younger and Older Adults: The Case of the Mass/Count Distinction, Canadian Journal of Experimental Psychology 61, 1, 21-34

[email protected] - matthew [email protected] - a [email protected]