<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="mccoy-etal-2020-syntax">
    <titleInfo>
        <title>Does Syntax Need to Grow on Trees? Sources of Hierarchical Inductive Bias in Sequence-to-Sequence Networks</title>
    </titleInfo>
    <name type="personal">
        <namePart type="given">R</namePart>
        <namePart type="given">Thomas</namePart>
        <namePart type="family">McCoy</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Robert</namePart>
        <namePart type="family">Frank</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <name type="personal">
        <namePart type="given">Tal</namePart>
        <namePart type="family">Linzen</namePart>
        <role>
            <roleTerm authority="marcrelator" type="text">author</roleTerm>
        </role>
    </name>
    <originInfo>
        <dateIssued>2020</dateIssued>
    </originInfo>
    <typeOfResource>text</typeOfResource>
    <genre authority="bibutilsgt">journal article</genre>
    <relatedItem type="host">
        <titleInfo>
            <title>Transactions of the Association for Computational Linguistics</title>
        </titleInfo>
        <originInfo>
            <issuance>continuing</issuance>
            <publisher>MIT Press</publisher>
            <place>
                <placeTerm type="text">Cambridge, MA</placeTerm>
            </place>
        </originInfo>
        <genre authority="marcgt">periodical</genre>
        <genre authority="bibutilsgt">academic journal</genre>
    </relatedItem>
    <abstract>Learners that are exposed to the same training data might generalize differently due to differing inductive biases. In neural network models, inductive biases could in theory arise from any aspect of the model architecture. We investigate which architectural factors affect the generalization behavior of neural sequence-to-sequence models trained on two syntactic tasks, English question formation and English tense reinflection. For both tasks, the training set is consistent with a generalization based on hierarchical structure and a generalization based on linear order. All architectural factors that we investigated qualitatively affected how models generalized, including factors with no clear connection to hierarchical structure. For example, LSTMs and GRUs displayed qualitatively different inductive biases. However, the only factor that consistently contributed a hierarchical bias across tasks was the use of a tree-structured model rather than a model with sequential recurrence, suggesting that human-like syntactic generalization requires architectural syntactic structure.</abstract>
    <identifier type="citekey">mccoy-etal-2020-syntax</identifier>
    <identifier type="doi">10.1162/tacl_a_00304</identifier>
    <location>
        <url>https://aclanthology.org/2020.tacl-1.9</url>
    </location>
    <part>
        <date>2020</date>
        <detail type="volume"><number>8</number></detail>
        <extent unit="page">
            <start>125</start>
            <end>140</end>
        </extent>
    </part>
</mods>
</modsCollection>

%0 Journal Article
%T Does Syntax Need to Grow on Trees? Sources of Hierarchical Inductive Bias in Sequence-to-Sequence Networks
%A McCoy, R. Thomas
%A Frank, Robert
%A Linzen, Tal
%J Transactions of the Association for Computational Linguistics
%D 2020
%V 8
%I MIT Press
%C Cambridge, MA
%F mccoy-etal-2020-syntax
%X Learners that are exposed to the same training data might generalize differently due to differing inductive biases. In neural network models, inductive biases could in theory arise from any aspect of the model architecture. We investigate which architectural factors affect the generalization behavior of neural sequence-to-sequence models trained on two syntactic tasks, English question formation and English tense reinflection. For both tasks, the training set is consistent with a generalization based on hierarchical structure and a generalization based on linear order. All architectural factors that we investigated qualitatively affected how models generalized, including factors with no clear connection to hierarchical structure. For example, LSTMs and GRUs displayed qualitatively different inductive biases. However, the only factor that consistently contributed a hierarchical bias across tasks was the use of a tree-structured model rather than a model with sequential recurrence, suggesting that human-like syntactic generalization requires architectural syntactic structure.
%R 10.1162/tacl_a_00304
%U https://aclanthology.org/2020.tacl-1.9
%U https://doi.org/10.1162/tacl_a_00304
%P 125-140

Markdown (Informal)

[Does Syntax Need to Grow on Trees? Sources of Hierarchical Inductive Bias in Sequence-to-Sequence Networks](https://aclanthology.org/2020.tacl-1.9) (McCoy et al., TACL 2020)

• Does Syntax Need to Grow on Trees? Sources of Hierarchical Inductive Bias in Sequence-to-Sequence Networks (McCoy et al., TACL 2020)

ACL

• R. Thomas McCoy, Robert Frank, and Tal Linzen. 2020. Does Syntax Need to Grow on Trees? Sources of Hierarchical Inductive Bias in Sequence-to-Sequence Networks. Transactions of the Association for Computational Linguistics, 8:125–140.