Constructional (n (no) ) synonymy: a usage-based analysis of f - PowerPoint PPT Presentation

Constructional (n (no) ) synonymy: a usage-based analysis of f Comple lete Path in in Polis lish Daria Bębeniec and Małgorzata Cudna Maria Curie- Skłodowska University, Lublin Linguistics Beyond and Within, 22-23 October 2015

Two ne near-synonymous Pol olis ish con onstr tructio ions exp xpressin ing the the con oncept of of Com omple lete Path th NP (GEN) NP (GEN) • od ______________ do ______________ NP (GEN) NP (ACC) • od ______________ po ______________

Ex Exam ample les of of use use • 1) ( … ) rzeczą kardynalną jest przestrzeganie każdego prawa, od najmniejszego do najważniejszego . [NKJP] • ‘(… ) it is fundamental that every law be obeyed, from the smallest to the most important. ’ • 2) Pogłoski ( … ) skusiły ludzi w różnym wieku – od licealistów po emerytów . [NKJP] • ‘The rumours ( … ) tempted people of different ages – from high school students to old age pensioners. ’

Com omplete Path an and its its ins instances in in Pol olis ish Complete Path cx Form: source-PP goal-PP Meaning: COMPLETE PATH od-do od-po od-ku z-do z-na …

Con onstructio ions • “ basic units of language ” (Goldberg 1995: 4) • “ conventional, learned form-function pairings at varying levels of complexity and abstraction ” (Goldberg 2013: 17) • Non-compositional meaning Goldberg 1995 • Unpredictable (constrained) form • Sufficient frequency of occurrence Goldberg 2006 • Special collocational preferences Gries et al. 2005, Hilpert 2008, 2014

Con onstructio ions ct ctd. • Constructions are connected via several kinds of inheritance links, including instance links , polysemy links, metaphorical extension links and subpart links (Goldberg 1995: 72-80)

• Generalizations are formed on the basis of both form and meaning/function (e.g., Boyd and Goldberg 2011) CP • Do we have this generalization over the specific constructions? od-do od-po • Principle of No Synonymy (Goldberg 1995: 67) • What usage patterns are associated with each of the two constructions?

Today’s foc ocus: : len length of of prepositional complements (LM (LM-phrases) • Length is one of possible operationalizations of complexity • the dative alternation (Bresnan 2007, Bresnan et al. 2007, Bresnan and Ford 2010, Wolk et al. 2013) • the genitive alternation (Rosenbach 2003, Hinrichs and Szmrecsanyi 2007, Wolk et al. 2013, Ehret et al. 2014)

Our ur pr predic ictions/expectatio ions • od-do is going to be associated with shorter LMs than od-po and its LMs are going to be elaborated by phrases at the same level of specificity and hence complexity (cf. Przybylska 2002: 483-487 on the differences in meaning between the prepositions do and po ) → explanation based on both iconicity and frequency • EXAMPLES: • … od okrutnego strachu do bezrównej odwagi ‘from terrible fear to unequalled courage ’ [NKJP] • … od jesieni 1920 do jesieni 1921 ‘from autumn 1920 to autumn 1921 ’ [NKJP]

Our ur pr predic ictions/expectatio ions • in both constructions LM1 is going to be shorter than LM2 (due to processing constraints, cf. Wasow 1997 on the principle of end- weight) • EXAMPLE: • … od najjaśniejszych po najciemniejsze strony życia ‘from the brightest to the darkest sides of life ’ [NKJP]

Data a an and meth thod • 529 instances of both constructions ( od-do : 291 , od-po : 238 ) • 300m balanced subsection of NKJP ( Przepiórkowski et al. 2012) • POLIQARP NKJP (Janus and Przepiórkowski 2007) • usage-feature approach (e.g., Glynn 2009, 2010) / behavioural-profile approach (e.g., Divjak and Gries 2006, Gries and Divjak 2009) • R statistical environment 2.15.2 (R Development Core Team 2008) • Univariate and multivariate methods: chi-square test, multiple correspondence analysis, cluster analysis, logistic regression

Co Comparing th the tw two con onstructions • LM1 – syllables • LM2 – syllables Szmrecsanyi 2004 • LM1 – words • LM2 – words • Difference between LMs – syllables • Strict difference between LMs – syllables Wolk et al. 2013 • Difference between LMs – words • Strict difference between LMs – words

Cod odin ing sch schema VARIABLE LEVELS COMMENTS LM1.Length.levels.syllables short, medium, long 1-3 syl, 4-8 syl, 9 syl and more LM2.Length.levels.syllables short, medium, long 1-3 syl, 4-8 syl, 9 syl and more LM1.Length.levels.words short, medium, long 1 word, 2-3 words, 4 words and more LM2.Length.levels.words short, medium, long 1 word, 2-3 words, 4 words and more LMs.Length.difference.syllables LM1 longer, LM2 longer, same length +/-2 syl LMs.Length.strict.difference.syllables LM1 longer, LM2 longer, same length LMs.Length.difference.words LM1 longer, LM2 longer, same length +/-1 word LMs.Length.strict.difference.words LM1 longer, LM2 longer, same length

the catdes function from the FactoMineR package

Categ egory ry des descriptio ion OD-DO OD-PO 1. LM1 in syllables and words: long 1. LM2 in words: short 2. LM2 in words: medium 2. LM2 in syllables: short and medium 3. LM2 in syllables: long 3. Difference between LMs in syllables: 4. Strict difference between LMs in same length words: LM1 longer 4. Difference between LMs in words: LM2 longer This is contrary to our expectations! New explanation needed.

Tak akin ing bo both constructio ions tog ogether • LM1-LM2 – syllables Chi-square test for • LM1-LM2 – words independence • LM1 – syllables • LM2 – syllables • LM1 – words • LM2 – words Chi-square test for • Difference between LMs – syllables given probabilities • Strict difference between LMs – syllables • Difference between LMs – words • Strict difference between LMs – words

Chi hi-square tes ests for independence (LM (LM1, LM2 in n the the whole le sam sample) VARIABLE p-value COMMENTS LM1.Length.levels.syllables 2.136e-33 √ LM2.Length.levels.syllables LM1.Length.levels.words 3.952e-34 √ LM2.Length.levels.words

Chi hi-square tes ests for given pr probabil ilit itie ies (th (the e who hole sam sample) VARIABLE p-value COMMENTS LM1.Length.levels.syllables 0.0003722 √ LM2.Length.levels.syllables 1.225e-06 √ LM1.Length.levels.words 9.993e-10 √ LM2.Length.levels.words 4.674e-11 √ LMs.Length.difference.syllables 2.2e-16 √ LMs.Length.strict.difference.syllables 4.32e-15 √ LMs.Length.difference.words 2.2e-16 √ LMs.Length.strict.difference.words 8.054e-09 √

Th The ten endencies con onfir irmed 1. LM1/LM2 in words and syllables: in most cases LM1 short = LM2 short , LM1 medium = LM2 medium , LM1 long = LM2 long 2. LM1 in words and syllables: short and medium in most cases 3. LM2 in words and syllables: medium and medium in most cases 4. Difference between LMs in words and syllables: same length in most cases 5. Strict difference between LMs in words and syllables: LM2 longer in most cases Our expectations are borne out this time.

Con onclusions • We have shown that even with a small set of complexity-related variables it is possible to identify some usage patterns associated with both constructions under analysis • We have found some usage-based evidence for the existence of the general constructional schema (CP) • We have compared two operationalizations of complexity (cf. Szmrecsanyi 2004)

TH THANK YOU! Daria Bębeniec Małgorzata Cudna daria@hektor.umcs.lublin.pl mcudna@hektor.umcs.lublin.pl

References (1 (1) • Boyd Jeremy K. and Adele E. Goldberg. 2011. “Learning what not to say: The role of statistical preemption and categorization in a- adjective production ” , Language 87 (1): 55-83. • Bresnan Joan. 2007. “ Is syntactic knowledge probabilistic? Experiments with the English dative alternation ”, in S. Featherston and W. Sternefeld (eds.), Roots: Linguistics in Search of Its Evidential Base , Berlin: Mouton de Gruyter, pp. 77-96. • Bresnan Joan, Anna Cueni, Tatiana Nikitina and R. Harald Baayen. 2007. “Predicting the Dative Alternation ”, in G. Boume, I. Kraemer and J. Zwarts (eds.), Cognitive Foundations of Interpretation , Amsterdam: Royal Netherlands Academy of Science, pp. 69-94. • Bresnan Joan and Marilyn Ford. 2010. “Predicting syntax: Processing dative constructions in American and Australian varieties of English ” , Language 86 (1): 168 – 213. • Divjak Dagmar and Stefan Th. Gries. 2006. “ Ways of trying in Russian: Clustering behavioural profiles. ” Corpus Linguistics and Linguistic Theory 2, 3-60. • Ehret Katharina, Christoph Wolk and Benedikt Szmrecsanyi. 2014. “Quirky quadratures: on rhythm and weight as constraints on genitive variation in an unconventional data set” , English Language and Linguistics 18 (2): 263-303.