CoChat
Sign Up
Cited by 8Second Language Acquisition and LearningEFL/ESL Teaching and LearningLanguage Development and Disorders

Syntactic Networks of Interlanguage Across L2 Modalities and Proficiency Levels

Open access
Hao, Yuxin, Wang, Xuelin, Wu, Meng, Liu, Haitao·2021·Frontiers in Psychology
DOI 10.3389/fpsyg.2021.643120PMC PMC8316831
Open this paper in CoChat — free accountOpen PDF

Abstract

Over time, interlanguage studies have shifted from early qualitative to quantitative studies of specific linguistic structures. However, the focus of these studies is usually on one aspect of an interlanguage instead of the whole system. The ideal object of interlanguage research is a second language (L2) learner language system, for only in this way can the entire L2 learning process can be examined. As a self-organizing and self-regulated system, the panorama of interlanguage can be revealed objectively through a complex network approach. In this study, we construct eight interlanguage dependency syntactic networks of varying proficiency levels and modalities, and conduct a quantitative study of respective network parameters. We find that all syntactic networks of Chinese L2 learners (English native speakers) initially present scale-free and small-world properties. Additionally, there is no sudden syntactic emergence in interlanguage with different modalities. This suggests varying regularities in the development of a syntactic network between interlanguage and native language acquisition. Moreover, the first language plays an important role in L2 development. The network parameters (<k>),L, C, ND, and NC can differentiate interlanguage modalities, and five quantitative parameters, <k>,C, ND, γ′, and NC, can indicate L2 proficiency.

AI snapshot

Written from the paper’s full text.

TL;DR

This study constructs eight Chinese interlanguage dependency syntactic networks from written and oral learner corpora across four proficiency levels, plus two native Chinese reference networks. It finds that all interlanguage networks are scale-free and small-world from the outset, with no sudden syntactic emergence, and that several network parameters differentiate modality and L2 proficiency.

Key findings

  • All eight CSL interlanguage syntactic networks fit power-law degree distributions (R² > 0.9) and show small-world properties: average path length L ranges from 3.008 to 3.331, and clustering coefficient C ranges from 0.162 to 0.333, versus 0.002–0.009 in corresponding random networks. No sudden syntactic emergence appears across written or oral interlanguage.
  • Modality differences are significant for <k>, L, C, ND, and NC. Oral interlanguage has higher average degree than writing (6.831 vs. 5.755), longer average path length (3.273 vs. 3.043), higher clustering coefficient (0.292 vs. 0.201), higher network density (0.0173 vs. 0.0068), and lower network centralization; written interlanguage NC is 0.057 higher than oral.
  • Five parameters can indicate L2 proficiency: <k>, C, and ND are negatively correlated with Chinese proficiency, while γ′ and NC are positively correlated. For example, written <k> regression is y = 6.474 − 0.302*x, p = 0.01, R² = 0.922; oral <k> is y = 7.3379 − 0.2243*x, p = 0.038, R² = 0.8077.
  • The authors conclude that L1 knowledge plays an important role in L2 development and that L2 syntactic network development differs from L1 acquisition, suggesting the innate language acquisition device may function differently in L2 acquisition.

Methods

  • Participants were English native-speaking Chinese L2 learners, aged 20–25, majoring in Chinese at a Beijing university and beginning Chinese at college. Writing and oral corpora were divided into four proficiency sub-corpora each (W1–W4, O1–O4), each containing 5,000 words after punctuation removal; native Chinese written and oral reference corpora (WN, ON) each contained about 5,000 tokens.
  • Dependency syntactic networks were built with nodes as word types and edges as syntactic dependency relations. Tagging used the Language Technology Platform (LTP), followed by manual proofreading and error labeling; syntactic errors were retained. Python scripts and Pajek were used to construct and analyze undirected networks.
  • Network parameters analyzed included γ′, R², <k>, C, L, ND, and NC. The study compared interlanguage networks with random networks and used regression and t-tests to assess modality and proficiency differences.

Limitations

  • Stated by the paper: the corpora came from examination production rather than natural daily-life language, which may have influenced objectivity; the study also did not investigate same-topic similarities and differences between native Chinese and L2 Chinese writing/oral production, which may have affected results and should be controlled in future studies.
  • Evident from Table 1: per-level sample sizes are small and uneven (e.g., W4 n = 8, O4 n = 12; W1 n = 46, O1 n = 34), which may limit statistical power and generalizability.
  • Evident from the design: proficiency levels are compared as separate corpora rather than following the same learners longitudinally, so developmental claims are cross-sectional.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.3389/fpsyg.2021.643120

Cites

100 of 100 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

22 of 102 cited sources checked · 14 supported · 0 flagged · accuracy not yet measured

  • Premisecitation

    In Jiang et al. (2019), no further statistical tests were conducted to determine which network parameters can predict interlanguage proficiency.

    “In their study, no further statistical tests were conducted to determine which network parameters can predict interlanguage proficiency.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Jiang et al. (2019) source
  • Premisecitation

    The properties of small world and scale-free in syntactic networks can serve as metrics to measure the emergence of a syntactic network.

    “Therefore, the properties of “small world” and “scale-free” in syntactic networks can serve as metrics to measure the emergence of a syntactic network (Jiang et al., 2019).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Jiang et al., 2019 source
  • Premisecitation

    Jiang et al. (2019) did not directly relate the development of syntactic network parameters to the language proficiency of learners.

    “Jiang et al. (2019) focused on the small-world and scale-free properties in English interlanguage, but they did not directly relate the development of syntactic network parameters to the language proficiency of learners.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Jiang et al. (2019) source
  • Premisecitation

    Dependency analysis is an effective way to construct a syntactic network for analyzing the development of syntactic structures among language learners.

    “Previous studies have shown that dependency analysis is an effective way to construct a syntactic network for analyzing the development of syntactic structures among language learners (Corominas-Murtra et al., 2009; Jiang et al., 2019).”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Corominas-Murtra et al., 2009 source
    • cites Jiang et al., 2019 source
  • Interpretationreasoning

    The written vocabularies of L2 learners are richer than their oral vocabularies.

    “This indicates that the written vocabularies of L2 learners are richer than their oral vocabularies, as has been claimed by Kormos and Trebits (2012) and Zalbidea (2017).”
    Quote in the paperSubstantial claim1 of 2 sources checked
    • cites Kormos and Trebits (2012) source

      The results (Insert Table 2 around here) report more varied vocabulary in written than oral versions of both tasks. The paper measures lexical variety in the output of 44 upper-intermediate learners, however, not whether L2 learners generally have richer written vocabularies than oral vocabularies. (LLM, conf=0.85)

      Experimental — accuracy not yet measured
    • cites Zalbidea (2017) source
  • Classified as probably not a substantial claim. 53
  • May combine two claims. 29
  • The quote only partly states this claim. 4

  • Classified as probably not a substantial claim.

    Premisecitation

    The distribution of degrees of vertices in a complex network generally follows the power-law distribution.

    “The distribution of degrees of vertices in a complex network and a random network generally follows the power-law distribution and Poisson distribution, respectively (Barabási and Albert, 1999; Newman, 2003, 2010).”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Barabási and Albert, 1999 source
    • cites Newman, 2003 source

      The Degree distributions section (p. 13) says many networks have power-law tails, but some have exponential tails. The Scale-free networks section (pp. 13–14) also describes exponential and truncated power-law degree distributions. Thus the paper supports power laws as common in complex networks, not as their general degree-distribution form. (LLM, conf=0.87)

      Experimental — accuracy not yet measured
    • cites Newman, 2010 source

  • Classified as probably not a substantial claim.

    Premisecitation

    The distribution of degrees of vertices in a random network generally follows the Poisson distribution.

    “The distribution of degrees of vertices in a complex network and a random network generally follows the power-law distribution and Poisson distribution, respectively (Barabási and Albert, 1999; Newman, 2003, 2010).”
    Quote in the paperSubstantial claim2 of 3 sources checked
    • cites Barabási and Albert, 1999 source

      The passage on the Erdős–Rényi (ER) random graph model says its vertex degrees follow a Poisson distribution. It does not establish this for random networks generally: the paper also describes many large random networks, including its own growth model, whose degree distributions follow a power law. The supplied copy labels these passages as Abstract and provides no page number. (LLM, conf=0.87)

      Experimental — accuracy not yet measured
    • cites Newman, 2003 source

      In “Degree distributions” (p. 12), Newman says the Erdős–Rényi random graph has a binomial degree distribution that becomes Poisson in the large-graph limit. “Random graphs” (p. 20) also describes random-graph models with other degree distributions. The claim is supported for the standard large-graph model, but not for random networks generally. (LLM, conf=0.86)

      Experimental — accuracy not yet measured
    • cites Newman, 2010 source

  • Classified as probably not a substantial claim.

    Premisecitation

    In a scale-free network, only few linguistic units have the strong general syntactic capacity to combine with other linguistic units (words), while most other linguistic units have weaker combining ability.

    “A network with the power-law distribution of degrees has the property of scale-freeness, showing that only few linguistic units have the strong general syntactic capacity to combine with other linguistic units (words), while most other linguistic units have weaker combining ability (Barabási and Albert, 1999).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Barabási and Albert, 1999 source

      The passage labeled Abstract (page not provided) describes scale-free networks with power-law connectivity and highly connected vertices, supporting the general idea that connectivity is uneven. It does not discuss linguistic units or establish that a word’s number of connections measures its syntactic capacity to combine with other words. (LLM, conf=0.86)

      Experimental — accuracy not yet measured

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Small-world and scale-free properties are universal features of human language on a macro scale.

    “Consequently, small-world and scale-free properties are seen as universal features of human language on a macro scale (Ferrer i Cancho et al., 2004; Cong and Liu, 2014).”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites Ferrer i Cancho et al., 2004 source

      The Results and Global Versus Sentence-Level Patterns sections report scale-free degree distributions and small-world properties emerging at the macroscopic scale in the networks studied. But the Discussion says these patterns are only candidates for linguistic universals and that more work is needed to establish them; the paper does not establish them as universal features of human language. Page numbers were not provided in our copy. (LLM, conf=0.88)

      Experimental — accuracy not yet measured
    • cites Cong and Liu, 2014 source

  • Classified as probably not a substantial claim.

    Premisecitation

    A network with the power-law distribution of degrees has the property of scale-freeness.

    “A network with the power-law distribution of degrees has the property of scale-freeness, showing that only few linguistic units have the strong general syntactic capacity to combine with other linguistic units (words), while most other linguistic units have weaker combining ability (Barabási and Albert, 1999).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Barabási and Albert, 1999 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Syntactic networks of different language types have the same small-world and scale-free properties as other complex networks.

    “By analyzing complex networks, researchers have found that the syntactic networks of different language types have the same small-world and scale-free properties as other complex networks.”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites Ferrer i Cancho et al., 2004 source
    • cites Cong and Liu, 2014 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    English L2 learning is characterized by a gradual approximation to the target language instead of a sudden emergence at a point.

    “It found that unlike the L1 learning of native English children, English L2 learning is characterized by a gradual approximation to the target language instead of a sudden emergence at a point (Jiang et al., 2019).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Jiang et al., 2019 source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationown data

    This finding verifies and supplements Jiang et al. (2019), who investigated syntactic networks of English foreign language learners.

    “This finding verifies and supplements Jiang et al. (2019), who investigated syntactic networks of English foreign language learners.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Jiang et al. (2019) source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Highly connected words tend not to be interconnected.

    “Second, in terms of syntactic connectivity, highly connected words tend not to be interconnected (Ferrer i Cancho et al., 2004).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ferrer i Cancho et al., 2004 source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    L2 learners exhibit higher syntactic complexity in oral English and more lexical richness in writing.

    “Previous studies have argued that L2 learners exhibit higher syntactic complexity in oral English and more lexical richness in writing.”
    Quote in the paperNot a substantial claim1 of 3 sources checked
    • cites Kormos and Trebits (2012) source

      The Implications and limitations section reports that 44 upper-intermediate Hungarian learners of English used more varied vocabulary in writing than in speech. It also reports similar syntactic complexity across the two modes, rather than higher syntactic complexity in speech; the Differences between modes of performance section explicitly describes a lack of difference. (LLM, conf=0.90)

      Experimental — accuracy not yet measured
    • cites Zalbidea (2017) source
    • cites Cho (2018) source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationcitation

    The findings of this study are similar to those of Kormos and Trebits (2012), Zalbidea (2017), and Cho (2018).

    “The findings are similar to those of Kormos and Trebits (2012), Zalbidea (2017), and Cho (2018).”
    Quote in the paperNot a substantial claim1 of 3 sources checked
    • cites Kormos and Trebits (2012) source

      The Implications and limitations section reports Kormos and Trebits’s findings on written and spoken narrative performance. The citing sentence does not say which findings of the citing study are similar, so our copy cannot establish the claimed similarity. (LLM, conf=0.72)

      Experimental — accuracy not yet measured
    • cites Zalbidea (2017) source
    • cites Cho (2018) source

  • Probably not a substantial claim: may not pass S2 (checkable).

    Premisecitation

    L2 acquisition is the process of interaction and connection between input and existing representation.

    “L2 acquisition is the process of interaction and connection between input and existing representation (Ellis, 2002).”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Ellis, 2002 source

  • Classified as probably not a substantial claim.

    Premisecitation

    In a scale-free network, the closer the power exponent of the cumulative degree distribution (γ′) is to 1, the better a Zipfian distribution fits the data.

    “Regarding the power distribution of a scale-free network, because the value of its power exponent corresponding cumulative degree distribution parameter γ′ is closer to the value of 1, the more Zipfian distribution fits the data (Jiang et al., 2019).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Jiang et al., 2019 source

  • May combine two claims. The quote only partly states this claim.

    Findingcitation

    The γ′ value is initially closer to 1 for CSL learners, consistent with Jiang et al. (2019).

    “What is consistent with the findings of Jiang et al. (2019) is that the value of γ' is initially closer to 1.”
    Quote in the paperNot a substantial claim1 of 1 sources checked
    • cites Jiang et al., 2019 source

  • May combine two claims.

    Findingown data

    Compared with corresponding random networks, all syntactic networks of CSL learners have far greater clustering coefficients and smaller average path lengths.

    “Compared with corresponding random networks, the results show that all syntactic networks of CSL learners have far greater clustering coefficients and smaller average path lengths.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The degrees in corresponding random syntactic networks follow a binomial distribution, differing from the interlanguage networks.

    “This differs from the degrees in corresponding random syntactic networks following a binomial distribution.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given

    There is no sudden emergence in the syntactic development of writing or in the oral corpus of interlanguage.

    “Furthermore, we confirm that there is no sudden emergence in the syntactic development of writing, or in the oral corpus of interlanguage.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The average path length (L) of the corresponding eight random networks ranges from 3.703 to 5.79, which is higher than that of CSL learners at all levels.

    “the value of L in the corresponding eight random networks is in the range of 3.703–5.79, higher than that of CSL learners at all levels.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Interlanguage networks have small L and far greater C than the random network.

    “Furthermore, interlanguage networks have small L and far greater C than the random network”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    There is no pivot or emergence in syntactic networks across either written or oral Chinese interlanguage.

    “The results indicate that there is no pivot and emergence in syntactic networks, across either written or oral Chinese interlanguage.”
    • Also stated in: § The Properties of Syntax Networks of Interlanguage · boosted
      “Furthermore, we confirm that there is no sudden emergence in the syntactic development of writing, or in the”
    Quote in the paperSubstantial claim
    • Rests on an assumption: The eight proficiency levels correspond to successive stages in the course of L2 acquisition.

  • May combine two claims.

    Findingown data

    Neither written nor oral interlanguage ND reach the target language level at P4.

    “and neither reach the target language level at P4.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The NCs of written and spoken interlanguage both increase with the improvement in Chinese proficiency.

    “The NCs of written and spoken interlanguage both increase with the improvement in Chinese proficiency”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given

    Both written and oral interlanguage networks have small-world properties.

    “With regard to the interlanguage, both written and oral networks have small-world properties.”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingreasoning

    The probability that two vertices that are neighbors of a given vertex are neighbors of each other decreases as proficiency improves.

    “Additionally, the probability that two vertices that are neighbors of a given vertex are neighbors of each other decreases.”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingreasoning

    The clustering coefficient in interlanguage decreases as levels of learners improve.

    “The changes in the clustering coefficient in interlanguage decreases as levels of learners improve.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    ECSL learners can produce richer vocabulary in writing than CSL learners.

    “Meanwhile, ECSL learners can produce richer vocabulary in writing than CSL learners.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    Only few vertices have extremely high degrees.

    “only few vertices have extremely high degrees.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    Each of <k>, γ′, C, NC, and ND can be applicable as a syntactic indicator in measuring interlanguage development.

    “each can be applicable as a syntactic indicator in measuring interlanguage development”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In L1 acquisition, children's syntactic development progresses from pre-syntactic organization to a scale-free and small-world complex syntactic network by leaps and bounds.

    “This suggests that the syntactic development of L1 acquisition of children progresses from pre-syntactic organization to a scale-free and small-world complex syntactic network by leaps and bounds.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Corominas-Murtra et al. (2009) source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The power-law distribution of degrees in a scale-free network suggests that it follows "the principle of least effort".

    “In conclusion, the degrees of vertices in a scale-free network generally follow a power-law distribution, i.e., a Zipfian-like distribution, which suggests that it follows “the principle of least effort” (Zipf, 1949).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Zipf, 1949

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Classified as probably not a substantial claim.

    Study design and conductreasoning

    The study retained dependency relations of syntactic errors of L2 learners to ensure efficiency and accuracy of annotation and reliability of results.

    “We retained the dependency relations of syntactic errors of L2 learners to ensure not only the efficiency and accuracy of the annotation to reflect the real proficiency of CSL learners but also the reliability research results of this study.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The degrees of vertices in a scale-free network generally follow a power-law distribution, i.e., a Zipfian-like distribution.

    “In conclusion, the degrees of vertices in a scale-free network generally follow a power-law distribution, i.e., a Zipfian-like distribution, which suggests that it follows “the principle of least effort” (Zipf, 1949).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Zipf, 1949

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Premisecitation

    Written language is a revised and polished product.

    “written language is a revised and polished product”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Halliday, 1989

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Levelt, 1989 source
    • cites Kellogg, 2001 source

  • Classified as probably not a substantial claim.

    Premisecitation

    Previous studies have found differences in syntactic performance across modalities and genres in both first language systems and interlanguage systems.

    “Previous studies have found differences in the syntactic performance across modalities and genres in both first language systems and interlanguage systems (Biber, 1988; Kormos, 2014; Biber et al., 2016; Qin and Uccelli, 2016; Zalbidea, 2017; Bulté and Roothooft, 2020).”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Biber, 1988 source
    • cites Kormos, 2014
    • cites Biber et al., 2016 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Complex networks with the small-world property have a small average path length.

    “Complex networks with the small-world property have both a small average path length and far greater clustering coefficients compared with those of random networks (Watts and Strogatz, 1998; Cong and Liu, 2014).”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Watts and Strogatz, 1998 source
    • cites Cong and Liu, 2014 source

  • Probably not a substantial claim: may not pass S2 (checkable).

    Premisecitation

    Spoken language is a reflection of the process of language construction.

    “spoken language is a reflection of the process of language construction”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Halliday, 1989

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Levelt, 1989 source
    • cites Kellogg, 2001 source

  • Classified as probably not a substantial claim.

    Premisecitation

    Complex networks with the small-world property have far greater clustering coefficients compared with those of random networks.

    “Complex networks with the small-world property have both a small average path length and far greater clustering coefficients compared with those of random networks (Watts and Strogatz, 1998; Cong and Liu, 2014).”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites Watts and Strogatz, 1998 source
    • cites Cong and Liu, 2014 source

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationreasoning

    The first language plays an important role in L2 development.

    “Moreover, the first language plays an important role in L2 development.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: The complex network approach can objectively reveal the panorama of interlanguage.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Complex language networks are generally sparse, meaning that the probability of a certain relation (i.e., a syntactic dependency relation) between any two language units in a language subsystem is miniscule.

    “Complex language networks are generally sparse, which means that the probability of a certain relation (i.e., a syntactic dependency relation) between any two language units in a language subsystem is miniscule (Liu, 2017).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Liu, 2017

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Liu (2008) and Chen and Liu (2013) found variation in several important network parameters across modalities and genres.

    “From the perspective of complex networks, Liu (2008), and Chen and Liu (2013) also found variation in several important network parameters across modalities and genres.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites Liu (2008) source
    • cites Chen and Liu (2013)

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    All corpora in this study were produced by adult L2 learners.

    “All the corpora in this study were produced by adult L2 learners.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    English is an inflectional language, the opposite of Chinese in surface grammatical devices.

    “This is the opposite of what English, an inflectional language, does (Li and Thompson, 1981).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Li and Thompson, 1981

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Interpretationreasoning

    The innate language acquisition device may play a disparate role in L2 acquisition.

    “Moreover, the innate language acquisition device may play a disparate role in L2 acquisition.”
    Quote in the paperNot a substantial claim
    • Rests on an assumption: There is an innate language acquisition device.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Native language syntactic development in children exhibits emergence, marking a pivot from pre-syntactic organization to a scale-free and small-world syntactic network.

    “From the perspective of the complex system, the syntactic development of the native language of children witnesses emergence, signaling the pivot from the pre-syntactic organization to a scale-free and small-world syntactic network.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Corominas-Murtra et al. (2009) source

  • May combine two claims.

    Interpretationreasoning

    Neither English nor Chinese interlanguage shows sudden emergence of a syntactic system.

    “Therefore, neither English nor Chinese interlanguage shows sudden emergence of a syntactic system because the L1 knowledge of L2 learners provides much experience and reference for L2 processing.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    L2 learners begin L2 learning with a parasitic lexicon, parasitic phonology, and parasitic set of grammatical constructs.

    “Of note, L2 learners begin L2 learning with a parasitic lexicon, parasitic phonology, and parasitic set of grammatical constructs (MacWhinney, 1997).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites MacWhinney, 1997

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Semantic case indicates deep structures.

    “semantic case indicates deep structures”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Fillmore, 1968

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    A sharp transition in syntax occurs around 24 months and is close to adult proficiency, although children's vocabularies are inferior to adults'.

    “According to Corominas-Murtra et al. (2009), although vocabularies of children are inferior to those of adults, a sharp transition in syntax, occurring around 24 months, is close to adult proficiency.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Corominas-Murtra et al. (2009) source

  • Classified as probably not a substantial claim.

    Premisecitation

    Modality has a significant effect on L2 performance.

    “and that the modality also has a significant effect on L2 performance (Cho, 2018).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Cho, 2018 source

  • Classified as probably not a substantial claim.

    Premisecitation

    Interlanguage is peculiar in being permeable.

    “Adjemian (1976) pointed out that interlanguage is peculiar in being permeable.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Adjemian (1976) source

  • May combine two claims.

    Interpretationown data

    The absence of sudden emergence in syntactic development applies to both English (inflectional) and Chinese (isolating) interlanguage.

    “This is applicable not only to the inflectional language of English as an interlanguage, but also to the isolating language of Chinese as an interlanguage.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    A small-world network has shorter average path length (L) and a higher clustering coefficient (C) than its random network.

    “A small-world network has shorter average path length (L) and a higher clustering coefficient (C) than its random network (Watts and Strogatz, 1998).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Watts and Strogatz, 1998 source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Chinese is an isolating language that relies on function words and word order rather than on rich morphological information.

    “Chinese is an isolating language that relies on function words and word order rather than on rich morphological information.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Li and Thompson, 1981

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Classified as probably not a substantial claim.

    Premisecitation

    The <k> of a syntactic network reflects the average combination ability of syntactic units.

    “The <k> of the syntactic network not only reveals the lexical richness but also reflects the average combination ability of syntactic units (Liu, 2009b, 2017).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Liu, 2009b, 2017

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In native Chinese speakers' production, degrees of de are higher than those of zai.

    “In addition, degrees of de are higher than those of zai, which is also found in native the production of Chinese speakers (Chen and Liu, 2016).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Chen and Liu, 2016 source

  • May combine two claims.

    Interpretationreasoning

    The working memory demands of CSL learners are higher when they complete oral tasks, whereas the working memory capacity required for writing tasks is slightly lower, and its average path length is also shorter.

    “This indicates that the working memory demands of CSL learners are higher when they complete oral tasks, whereas the working memory capacity required for writing tasks is slightly lower, and its average path length is also shorter.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Previous studies have found no congruence between speaking and writing in terms of cognitive processes; speaking includes conceptualization, modulation, and monitoring, while writing is divided into planning, formulation, and monitoring.

    “Previous studies have found no congruence between speaking and writing in terms of cognitive processes; the former includes conceptualization, modulation, and monitoring, while the latter is divided into planning, formulation, and monitoring.”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Levelt, 1989 source
    • cites Kellogg, 2001 source
    • cites Kuiken and Vedder, 2012

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisecitation

    Human languages tend to have a minimized average dependency distance, which is constrained by grammar and the human capacity for cognitive working memory.

    “Human languages tend to have a minimized average dependency distance, which is constrained by grammar and the human capacity for cognitive working memory.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Hawkins, 2004 source

  • Probably not a substantial claim: may not pass S1 (asserted), S3 (load-bearing).

    Premisecitation

    The existence of the shortest path between any two vertices in a syntactic network is related to the minimization of the dependency distance.

    “Liu (2008) has contended that the existence of the shortest path between any two vertices in a syntactic network is related to the minimization of the dependency distance.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Liu (2008) source

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationcitation

    Average dependency distance can serve as a measure of processing as well as syntactic complexity.

    “Thus, it can serve as a measure of processing as well as syntactic complexity (Hawkins, 2004).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Hawkins, 2004 source

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Syntactic structures can be even more complex in speech than they are in writing.

    “Additionally, Halliday (1989) focused on native language (L1), and also claimed that syntactic structures can be even more complex in speech than they are in writing.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Halliday (1989)

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In Chinese, function words, such as prepositions, conjunctions, and auxiliary and modal particles, are the most connected word types that do not form syntactic dependency among themselves and only act as intermediate words in a syntactic network.

    “In Chinese, function words, such as prepositions, conjunctions, and auxiliary and modal particles, are the most connected word types that do not form syntactic dependency among themselves and only act as intermediate words in a syntactic network (Chen and Liu, 2016).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Chen and Liu, 2016 source

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given

    Interlanguage is a self-organizing and self-regulated system under the influence of its native language knowledge.

    “Interlanguage is a self-organizing and self-regulated system under the influence of its native language knowledge.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Interlanguage is a non-linear and dynamic complex system that gradually approaches the target language.

    “We find that interlanguage is a non-linear and dynamic complex system that gradually approaches the target language.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    The corpora used in this study are collected from examination production, which deviates from the natural corpus of daily life.

    “At the same time, the corpora used in this study are collected from examination production, which deviates from the natural corpus of daily life.”
    Quote in the paperNot a substantial claim

  • May combine two claims. The quote only partly states this claim.

    Interpretationcitation

    The finding that CSL learners do not reach the target language at P4 is consistent with Borodkin et al. (2016), who found that L2 lexical networks display greater local connectivity and less modular community structure than native language networks, even among proficient bilinguals.

    “This is consistent with Borodkin et al. (2016) who found that L2 lexical networks display greater local connectivity and less modular community structure than the network in the native language, even among proficient bilinguals.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Borodkin et al. (2016) source

  • May combine two claims.

    Interpretationreasoning

    CSL learners use less limited language structures, increasing language structure difficulty and making their syntactic network sparse.

    “CSL learners use less limited language structures, increasing language structure difficulty and making their syntactic network sparse.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    The approximation of interlanguage network characteristics to the target language network can be attributed to increasing vocabulary diluting the edges.

    “This can also be attributed to the increasing vocabulary diluting the edges.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    Lexical richness and the usage rate of functional words are more evident in writing than in speaking.

    “This indicates that lexical richness and the usage rate of functional words are more evident in writing than in speaking.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    This study does not investigate the similarities and differences in the development of the parameters of syntactic networks between native Chinese and second-language Chinese writing on the same topic of the writing and oral production.

    “However, this study does not investigate the similarities and differences in the development of the parameters of syntactic networks between native Chinese and second-language Chinese writing on the same topic of the writing and oral production.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Most language complex networks are sparsely distributed because the probability of syntactic dependencies between any two language units in a language subsystem is low.

    “Most language complex networks are sparsely distributed because the probability of syntactic dependencies between any two language units in a language subsystem is low (Liu, 2017).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Liu, 2017

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Interpretationreasoning

    CSL writers usually aim to use more functional cohesive words and have a tendency to "talk written down."

    “Additionally, CSL writers usually aim to use more functional cohesive words and have a tendency to “talk written down.””
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    The omission of a same-topic native Chinese comparison may have influenced the results.

    “This may have influenced the results, and as such, should be controlled in future studies.”
    Quote in the paperNot a substantial claim
73 of 73

  • The paper relies on this without support.

    The complex network approach can objectively reveal the panorama of interlanguage.

    “As a self-organizing and self-regulated system, the panorama of interlanguage can be revealed objectively through a complex network approach.”

  • The paper relies on this without support.

    The syntactic dependency networks constructed from written and oral production are valid models of CSL learners' interlanguage syntax.

    “Figure 3 shows the syntactic network of ECSL learners across modalities and language proficiencies. The edges between vertices represent word types and dependency syntactic relations, respectively.”

  • The paper relies on this without support.

    The number of vertices (word types) in a syntactic network is a valid indicator of language proficiency.

    “This means that English native speakers produce more word types as their proficiency advances in both written and oral languages, but the language proficiency of CSL learners at P4 is still lower than that of Chinese native speakers.”

  • The paper relies on this without support.

    Native speaker syntactic networks represent the target language level.

    “the 10 sets of data of L2 learners and native speakers with two different modalities”

  • The paper relies on this without support.

    The eight CSL learners are representative of CSL learners at the corresponding proficiency levels.

    “the average path lengths extracted from networks of eight CSL learners are between 3.008 and 3.331”

  • The paper relies on this without support.

    The Chinese native speaker comparison values used in the study are valid and comparable to the CSL learner data.

    “the values of L for interlanguage across modalities are also less than those of Chinese native speakers.”

  • The paper relies on this without support.

    The eight proficiency levels correspond to successive stages in the course of L2 acquisition.

    “Throughout the course of an acquisition, Chinese interlanguage networks all exhibit the properties of scale-freeness and small-worldness”

  • The paper relies on this without support.

    There is an innate language acquisition device.

    “the innate language acquisition device may play a disparate role in L2 acquisition.”

  • The paper relies on this without support.

    Oral production provides evidence of the implicit knowledge of a learner, whereas written production allows for the use of explicit knowledge.

    “Oral production is generally considered to provide evidence of the implicit knowledge of a learner, whereas written production seems to allow for the use of explicit knowledge.”

  • The paper relies on this without support.

    Tasks involving vocal organs controlling, monitoring, and adjusting output compete for limited attention resources, and have trade-off effects in oral production.

    “Tasks involving vocal organs controlling, monitoring, and adjusting output compete for limited attention resources, and have trade-off effects in oral production.”

  • The paper relies on this without support.

    Natural language networks have a decreasing degree distribution.

    “in accordance with natural language networks”

  • The paper relies on this without support.

    L2 input is limited in the beginning stages of acquisition.

    “because of the limited L2 input in the beginning stages of acquisition”

  • The paper relies on this without support.

    Chinese is not rich in morphological inflection.

    “Chinese is not rich in morphological inflection”

  • The paper relies on this without support.

    Functional words are one of the most important syntactic devices in Chinese.

    “its functional words are one of its most important syntactic devices”

  • The paper relies on this without support.

    Native language acquisition provides a valid baseline for comparing interlanguage syntactic development.

    “We demonstrate that the syntactic development of interlanguage of L2 learners changes slower than that of their native language acquisition.”
15 of 15

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Conclusion reverses the oral-written average-degree comparison

    Major

    The conclusion says oral interlanguage has a lower overall average degree, but the discussion reports a higher average for oral than written interlanguage.

  • Oral path length moves away from the target-language value

    Minor

    The conclusion says all parameters approach target-language values as proficiency improves. For oral average path length, O3 is 3.290, close to ON's 3.279, whereas the more proficient O4 is farther away at 3.174.

  • Some learner path lengths exceed the native-speaker value

    Minor

    The claim that interlanguage path lengths are below those of native speakers does not hold for oral networks: O2 has L = 3.331, while ON has L = 3.279.

3 of 3

Cites

Flagged first, then by citations

Research works printed but not linked: 10

  1. Che, 2010, LTP: a Chinese language technology platform, 13
  2. Cutting, 2011, Spoken discourse, 155
  3. Ferrer i Cancho, 2005, The structure of syntactic dependency networks: insights from recent advances in network theory, 60
  4. Goldberg, 1995, Constructions: A Construction Grammar Approach to Argument Structures
  5. Lamb, 1966, Outline of Stratificational Grammar.
  6. Langacker, 1987, Foundations of Cognitive Grammar (Vol. I): Theoretical Prerequisites.
  7. Liu, Dependency Grammar: From Theory to Practice
  8. Liu, 2017, An Introduction to Quantitative Linguistics
  9. Petch-Tyson, 1998, Writer/reader visibility in EFL written discourse, 107
  10. Richards, 1996, Dictionary of Language Teaching and Applied Linguistics.

Cited by

Sorted by citations

Related

Authors
Hao, Y., Wang, X., Wu, M., & Liu, H.
Affiliations
  • Huaqiao University (CN)
  • Guangdong University of Foreign Studies (CN)
Corresponding author
Liu, Haitao
ORCID
Venue
Frontiers in Psychology
Publisher
Frontiers Media SA
Published
2021-07-14
Volume
12
Pages
643120-643120
ISSN
1664-1078 (ISSN-L)
DOI
10.3389/fpsyg.2021.643120
PMC
PMC8316831
OpenAlex
W3181093976
Semantic Scholar
2bf893f01b8e7e21933765c2f712ef01481213a1
Cited by
8
References
95
Citations by year
2026: 2 · 2025: 4 · 2023: 1 · 2021: 1
FWCI
0.72
Type
journal article
Language
en
Open access
Licence
Locations
  • Frontiers in Psychology (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion, cc-by)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
Funders
  • Huaqiao University
  • Fundamental Research Funds for the Central Universities
Topics
  • Social Sciences › Psychology › Developmental and Educational Psychology › Second Language Acquisition and Learning
  • Social Sciences › Arts and Humanities › Language and Linguistics › EFL/ESL Teaching and Learning
  • Social Sciences › Psychology › Developmental and Educational Psychology › Language Development and Disorders
Keywords
interlanguage, syntactic networks, dependency networks, second language acquisition, second language proficiency, modality effects, Chinese as a second language, scale-free networks, small-world networks, first language influence, network parameters

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-19.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-19.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review