What Meaning-Form Correlation Has to Compose With: A Study of MFC on Artificial and Natural Language

Timothee Mickus; Timothée Bernard; D. Paperno

doi:10.18653/v1/2020.coling-main.333

What Meaning-Form Correlation Has to Compose With: A Study of MFC on Artificial and Natural Language

Timothee Mickus, Timothée Bernard, D. Paperno

Research output: Contribution to conference › Paper › Academic

Abstract

Compositionality is a widely discussed property of natural languages, although its exact definition has been elusive. We focus on the proposal that compositionality can be assessed by measuring meaning-form correlation. We analyze meaning-form correlation on three sets of languages: (i) artificial toy languages tailored to be compositional, (ii) a set of English dictionary definitions, and (iii) a set of English sentences drawn from literature. We find that linguistic phenomena such as synonymy and ungrounded stop-words weigh on MFC measurements, and that straightforward methods to mitigate their effects have widely varying results depending on the dataset they are applied to. Data and code are made publicly available.

Original language	English
Pages	3737–3749
Number of pages	13
DOIs	https://doi.org/10.18653/v1/2020.coling-main.333
Publication status	Published - Dec 2020
Event	The 28th International Conference on Computational Linguistics (COLING) - Online Duration: 8 Dec 2020 → 13 Dec 2020 https://coling2020.org/

Conference

Conference	The 28th International Conference on Computational Linguistics (COLING)
Abbreviated title	COLING'2020
Period	8/12/20 → 13/12/20
Internet address	https://coling2020.org/

Access to Document

10.18653/v1/2020.coling-main.333Licence: CC BY

2020.coling-main.333Final published version, 284 KB

Cite this

@conference{16babf542d70431cacf76983e30559d8,

title = "What Meaning-Form Correlation Has to Compose With: A Study of MFC on Artificial and Natural Language",

abstract = "Compositionality is a widely discussed property of natural languages, although its exact definition has been elusive. We focus on the proposal that compositionality can be assessed by measuring meaning-form correlation. We analyze meaning-form correlation on three sets of languages: (i) artificial toy languages tailored to be compositional, (ii) a set of English dictionary definitions, and (iii) a set of English sentences drawn from literature. We find that linguistic phenomena such as synonymy and ungrounded stop-words weigh on MFC measurements, and that straightforward methods to mitigate their effects have widely varying results depending on the dataset they are applied to. Data and code are made publicly available.",

author = "Timothee Mickus and Timoth{\'e}e Bernard and D. Paperno",

year = "2020",

month = dec,

doi = "10.18653/v1/2020.coling-main.333",

language = "English",

pages = "3737–3749",

note = "The 28th International Conference on Computational Linguistics (COLING), COLING'2020 ; Conference date: 08-12-2020 Through 13-12-2020",

url = "https://coling2020.org/",

}

TY - CONF

T1 - What Meaning-Form Correlation Has to Compose With

T2 - The 28th International Conference on Computational Linguistics (COLING)

AU - Mickus, Timothee

AU - Bernard, Timothée

AU - Paperno, D.

PY - 2020/12

Y1 - 2020/12

N2 - Compositionality is a widely discussed property of natural languages, although its exact definition has been elusive. We focus on the proposal that compositionality can be assessed by measuring meaning-form correlation. We analyze meaning-form correlation on three sets of languages: (i) artificial toy languages tailored to be compositional, (ii) a set of English dictionary definitions, and (iii) a set of English sentences drawn from literature. We find that linguistic phenomena such as synonymy and ungrounded stop-words weigh on MFC measurements, and that straightforward methods to mitigate their effects have widely varying results depending on the dataset they are applied to. Data and code are made publicly available.

AB - Compositionality is a widely discussed property of natural languages, although its exact definition has been elusive. We focus on the proposal that compositionality can be assessed by measuring meaning-form correlation. We analyze meaning-form correlation on three sets of languages: (i) artificial toy languages tailored to be compositional, (ii) a set of English dictionary definitions, and (iii) a set of English sentences drawn from literature. We find that linguistic phenomena such as synonymy and ungrounded stop-words weigh on MFC measurements, and that straightforward methods to mitigate their effects have widely varying results depending on the dataset they are applied to. Data and code are made publicly available.

U2 - 10.18653/v1/2020.coling-main.333

DO - 10.18653/v1/2020.coling-main.333

M3 - Paper

SP - 3737

EP - 3749

Y2 - 8 December 2020 through 13 December 2020

ER -

What Meaning-Form Correlation Has to Compose With: A Study of MFC on Artificial and Natural Language

Abstract

Conference

Access to Document

Fingerprint

Cite this