The Geometry of Culture: Analyzing the Meanings of Class through Word Embeddings

The Geometry of Culture: Analyzing the Meanings of Class through Word Embeddings Austin C. Kozlowski Matt Taddy James A. Evans

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

The Geometry of Culture: Analyzing the Meanings of Class through Word Embeddings 2019 American Sociological Review / arXiv 10.1177/0003122419877135 1803.09288 kozlowski2019-geometry-culture kozlowski2019-geometry-culture.txt, the project's staged copy a616273f26e7bddc41b8842d65487dea41fd429d arxiv-open digital-humanities

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them method Quotation withheld because this record is not cleared for redistribution. cultural dimensions in word-embedding spaces Quotation withheld because this record is not cleared for redistribution. system-or-tool-building system-or-tool-building Quotation withheld because this record is not cleared for redistribution. mechanisms building cultural dimensions as averaged vector-differences between antonym word pairs, then projecting words onto them Quotation withheld because this record is not cleared for redistribution. The method needs a large corpus; subtle and complex cultural associations cannot be recovered from small bodies of text. what this limits: any application to a small corpus: WORD EMBEDDINGS ... REQUIRE A RELATIVELY LARGE BODY OF TEXT IF THE OUTPUT VECTOR SPACE IS TO CAPTURE SUBTLE AND COMPLEX ASSOCIATIONS Quotation withheld because this record is not cleared for redistribution. Race could not be traced historically because the terminology shifted across the century, so the historical analysis was restricted to gender and class. what this limits: the absence of race from the historical analysis: UNDERTAKING THIS TASK EFFECTIVELY WOULD REQUIRE A DEEPER ENGAGEMENT WITH LINGUISTIC SHIFTS IN RACIAL TERMINOLOGY THAN IS FEASIBLE WITHIN THE SPACE CONSTRAINTS - the dimension is omitted for a vocabulary reason Quotation withheld because this record is not cleared for redistribution. The validating survey used a Mechanical Turk sample that is not representative of the US population. what this limits: the survey validation specifically: OUR SAMPLE CANNOT BE SAID TO BE REPRESENTATIVE OF THE GENERAL POPULATION OF THE UNITED STATES Quotation withheld because this record is not cleared for redistribution. Embedding projections capture semantic association in text, not material or on-the-ground social realities. what this limits: every substantive claim about culture: WORD EMBEDDINGS MODELS, BY DISTILLING INFORMATION FROM LOCAL WORD CO-OCCURRENCES, DO NOT PROVIDE DIRECT INFORMATION ABOUT MATERIAL, ON-THE-GROUND REALITIES Quotation withheld because this record is not cleared for redistribution. The confidence intervals depend on treating books and newspapers as independent draws from a population of cultural observations, which is stated as a condition rather than shown. about this quote: THE QUOTE IS AN ASSUMPTION, NOT A FINDING - the sentence opens 'If we assume that the texts' before it, and a page number ('29') sits between that opening and the quote. · what this limits: the confidence or credible intervals reported for distances and projections here: they follow from assuming the underlying texts are i.i.d. observations drawn from a population of cultural observations Quotation withheld because this record is not cleared for redistribution. The method the paper displaces is criticised on reproducibility, which is the ground on which the new one is offered. what this limits: the comparison between close reading and embedding-based analysis: close reading is credited with advancing the field and criticised for clear limitations in reproducibility Quotation withheld because this record is not cleared for redistribution. the interpretation space sampled SYSTEMATICALLY - 1,775 antonym pairs taken from WordNet - rather than by selecting salient dimensions, with the reported dimensions restricted to those at the smallest angle to the focal ones eliminates: CHERRY-PICKING BY THE ANALYST. The paper names the charge it is answering: that researchers 'conveniently cherry-pick explanations to confirm…'. An unsupervised model yields far more readable structure than any paper reports, so the choice of what to report is where the freedom lives · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. word embeddings word embeddings Quotation withheld because this record is not cleared for redistribution. semantic relations semantic relations Quotation withheld because this record is not cleared for redistribution. cultural categories cultural categories Quotation withheld because this record is not cleared for redistribution. meaning - OPERATIONALISED AS GEOMETRY: semantic relations between words represented as geometric relationships between vectors, which the paper calls a relational model of meaning Quotation withheld because this record is not cleared for redistribution. semantic change - measured as CHANGE IN POSITION: a word's rate of semantic change taken to be the change in its overall position in the vector space Quotation withheld because this record is not cleared for redistribution. cultural categories - held by the intersectionality literature to be inseparable: their meaning changes as they overlap, so they cannot be isolated and understood independently Quotation withheld because this record is not cleared for redistribution. builds-on → word2vec (CBOW/skip-gram) and GloVe (from NLP) Quotation withheld because this record is not cleared for redistribution. extends → interpretive text analysis of culture (adding reproducibility) Quotation withheld because this record is not cleared for redistribution. extends → Bolukbasi et al. (2016) and Caliskan et al. (2017) on bias in word embeddings Quotation withheld because this record is not cleared for redistribution. contradicts → Bolukbasi et al. (2016) proposal to de-bias word embeddings Quotation withheld because this record is not cleared for redistribution. extends → Pierre Bourdieu's relational field theory / correspondence analysis Quotation withheld because this record is not cleared for redistribution. extends → Osgood's semantic differential method Quotation withheld because this record is not cleared for redistribution. semantic relations —operationalised-as→ word embeddings semantic relations word embeddings strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. 20th — span of the gender/class coevolution analysis Quotation withheld because this record is not cleared for redistribution. 1970s — when LSA-type methods first appeared Quotation withheld because this record is not cleared for redistribution. 20,000 — example vocabulary size reduced by embeddings Quotation withheld because this record is not cleared for redistribution. projecting words onto these dimensions recovers widely shared cultural connotations Quotation withheld because this record is not cleared for redistribution. feminine and masculine occupations sit at opposite poles of a gender dimension Quotation withheld because this record is not cleared for redistribution. word embeddings capture more complex semantic relations than prior text methods about this quote: A PAGE NUMBER SITS INSIDE THIS QUOTE: the '2' between 'past modes of' and 'computational text analysis' is the page's number, not part of the sentence; read the quote with it deleted. Quotation withheld because this record is not cleared for redistribution. word embeddings word embeddings Quotation withheld because this record is not cleared for redistribution. Latent Semantic Analysis (LSA) Latent Semantic Analysis (LSA) Quotation withheld because this record is not cleared for redistribution. relational model of meaning relational model of meaning Quotation withheld because this record is not cleared for redistribution. Bayesian Bayesian about this quote: QUOTE RECOVERED FROM THE PAPER'S OWN CHARACTERS. The recorded quote was LETTERS-ONLY IDENTICAL to this stretch of the staged text but not an exact substring of it - spaces, hyphens, ligature glyphs or mathematical operators differed, so it could never anchor. This is the file's own text between the same two letters: not produced by a rule about the damage, copied out of the file. Round-trip checked (the recovered span is letters-only equal to the original) and length-checked. The text as originally recorded is kept in evidence as extracted. Quotation withheld because this record is not cleared for redistribution. bootstrap bootstrap Quotation withheld because this record is not cleared for redistribution. confidence-interval confidence-interval Quotation withheld because this record is not cleared for redistribution. correlation correlation Quotation withheld because this record is not cleared for redistribution. mean mean Quotation withheld because this record is not cleared for redistribution. p-value p-value about this quote: QUOTE RECOVERED FROM THE PAPER'S OWN CHARACTERS. The recorded quote was LETTERS-ONLY IDENTICAL to this stretch of the staged text but not an exact substring of it - spaces, hyphens, ligature glyphs or mathematical operators differed, so it could never anchor. This is the file's own text between the same two letters: not produced by a rule about the damage, copied out of the file. Round-trip checked (the recovered span is letters-only equal to the original) and length-checked. The text as originally recorded is kept in evidence as extracted. Quotation withheld because this record is not cleared for redistribution. standard-deviation standard-deviation about this quote: QUOTE RECOVERED FROM THE PAPER'S OWN CHARACTERS. The recorded quote was LETTERS-ONLY IDENTICAL to this stretch of the staged text but not an exact substring of it - spaces, hyphens, ligature glyphs or mathematical operators differed, so it could never anchor. This is the file's own text between the same two letters: not produced by a rule about the damage, copied out of the file. Round-trip checked (the recovered span is letters-only equal to the original) and length-checked. The text as originally recorded is kept in evidence as extracted. Quotation withheld because this record is not cleared for redistribution. statistical-significance statistical-significance about this quote: QUOTE RECOVERED FROM THE PAPER'S OWN CHARACTERS. The recorded quote was LETTERS-ONLY IDENTICAL to this stretch of the staged text but not an exact substring of it - spaces, hyphens, ligature glyphs or mathematical operators differed, so it could never anchor. This is the file's own text between the same two letters: not produced by a rule about the damage, copied out of the file. Round-trip checked (the recovered span is letters-only equal to the original) and length-checked. The text as originally recorded is kept in evidence as extracted. Quotation withheld because this record is not cleared for redistribution. variance variance Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1177/0003122419877135.