# Using the Relative Entropy of Linguistic Complexity to Assess L2 Language Proficiency Development

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

- (1)
- How distinct are the differences in language proficiency between L2 learners at a lower level and L2 learners at a higher level as compared to the differences between intermediate-level L2 learners and higher-level learners from the perspective of information gain?
- (2)
- Does the algorithm of relative entropy have advantages over the frequency-based algorithms for lexical and syntactic complexity in detecting development patterns of L2 language proficiency?

## 2. Background

#### 2.1. Linguistic Complexity and the Development of Language Proficiency in L2

#### 2.2. Relative Entropy

_{2}p(“take”|1820s) − log

_{2}p(“take”|1810s))

_{2}p(“take”|1820s) − ∑p(“take”|1820s)*log

_{2}p(“take”|1810s)

- p(“take”|1820s) = 556/1435 = 0.3875,
- p(“took”|1820s) = 327/1435 = 0.2279,
- p(“taken”|1820s) = 345/1435 = 0.2404,
- p(“taking”|1820s) = 144/1435 = 0.1003,
- p(“takes”|1820s) = 63/1435 = 0.0439.

_{2}p(“take”|1820s)

_{2}(0.3875) + 0.2279*log

_{2}(0.2279) + 0.2404*log

_{2}(0.2404) + 0.1003*log

_{2}(0.1003) + 0.0439*log

_{2}(0.0439)) = −2.041431

- p(“take”|1810s) = 728/1415 = 0.5145,
- p(“took”|1810s) = 168/1415 = 0.1187,
- p(“taken”|1810s) = 228/1415 = 0.1611,
- p(“taking”|1810s) = 133/1415 = 0.094,
- p(“took”|1810s) = 158/1415 = 0.1117.

_{2}p(“take”|1810s) = −2.1864

## 3. Materials and Methods

#### 3.1. Material

#### 3.2. Method

- (a)
- Relative entropy and the discrimination of information distribution.

_{h}” refers to L2 learners at a higher level, but “Level

_{l}” L2 learners at a lower level.

_{l}” is the distribution of linguistic phenomena that learners have encountered at a lower level and “Level

_{h}” is the new distribution that learners will encounter at a higher level. More importantly, the algorithm of relative entropy examines the information differences between the same linguistic units encoded by two groups of L2 learners. This avoids the problem that characterized previous studies, namely ignoring the weights of different units and simply placing them under the same category.

_{h})” by using an encoding optimized for “L2 learners at a lower level (Level

_{l})”. When applied to the comparison of sub-corpora of the EFCAMDAT2, the KLD serves as a strong indication of the degree of difference between two sub-corpora (representing two groups of L2 learners) measured in bits as well as of the linguistic units that are primarily associated with a difference. That is to say, the difference in the KLD indicates that linguistic units need high amounts of additional bits for encoding. We can find the KLD as an indicator of change after sliding over different groups of L2 learners’ lines in the EFCAMDATA2 and by comparing adjacent L2 learners’ groups.

- (b)
- Language units (measures).

- (c)
- Traditional approaches to lexical/syntactic complexity and stationary time series.

## 4. Results

#### 4.1. The Results from the KLD

#### 4.2. The Results from Syntactic and Lexical Complexity

## 5. Discussion

#### 5.1. Conflicting Results from Different Studies

#### 5.2. The Developmental Patterns of Language Proficiency in L2 Learners

#### 5.3. Consistency with the Other Measures

## 6. Conclusions

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Conflicts of Interest

## Appendix A. JSD Algorithm

_{h}||Level

_{l})= 1/2*KLD(Level

_{h}||(Level

_{h}+ Level

_{l})/2) + 1/2KLD(Level

_{l}||(Level

_{h}+ Level

_{l})/2).

## Appendix B. The Results from the JSD

Cross-Proficiency Levels of L2 | JSD of Grammar | JSD of Lexicon | |||||||
---|---|---|---|---|---|---|---|---|---|

POS-Trigram | Sub-Conj. | Token | Lemma | ||||||

A1→A2 | A1→(A2, B1, B2, C1) | 0.05 | (coef = 0.03, p = 0.003) | 0.09 | (coef = 0.007, p = 0.34) | 0.17 | (coef = 0.003, p = 0.23) | 0.06 | (coef = 0.02, p = 0.17) |

A1→B1 | 0.08 | 0.117 | 0.18 | 0.11 | |||||

A1→B2 | 0.1 | 0.1 | 0.18 | 0.13 | |||||

A1→C1 | 0.13 | 0.12 | 0.18 | 0.12 | |||||

A2→B1 | A2→(B1, B2, C1) | 0.03 | (coef = 0.2, p = 0.18) | 0.04 | (coef = 0.02, p < 0.001) | 0.11 | 0.08 | (coef = 0.005, p = 0.33) | |

A2→B2 | 0.04 | 0.06 | 0.12 | 0.09 | |||||

A2→C1 | 0.07 | 0.08 | 0.1 | 0.09 | |||||

B1→B2 | B1→(B2, C1) | 0.02 | 0.014 | 0.1 | 0.06 | ||||

B1→C1 | 0.03 | 0.017 | 0.09 | 0.06 | |||||

B2→C1 | B2→(C1) | 0.02 | 0.009 | 0.07 | 0.05 |

Cross-Proficiency Levels of L2 | JSD of Grammar | JSD of Lexicon | |||||||
---|---|---|---|---|---|---|---|---|---|

POS-Trigram | Sub-Conj. | Token | Lemma | ||||||

A1→C1 | (A1, A2, B1, B2)→C1 | 0.13 | (coef = −0.04, p = 0.043) | 0.12 | (coef = −0.04, p = 0.034) | 0.18 | (coef = −0.034, p = 0.09) | 0.11 | (coef = −0.02, p = 0.015) |

A2→C1 | 0.07 | 0.075 | 0.1 | 0.09 | |||||

B1→C1 | 0.03 | 0.017 | 0.09 | 0.06 | |||||

B2→C1 | 0.02 | 0.009 | 0.07 | 0.05 | |||||

A1→B2 | (A1, A2, B1)→B2 | 0.1 | (coef = −0.04, p = 0.26) | 0.1 | (coef = −0.043, p = 0.008) | 0.18 | (coef = −0.04, p = 0.18) | 0.13 | (coef = −0.035, p = 0.05) |

A2→B2 | 0.03 | 0.058 | 0.12 | 0.09 | |||||

B1→B2 | 0.02 | 0.014 | 0.1 | 0.06 | |||||

A1→B1 | (A1, A2)→B1 | 0.08 | 0.12 | 0.18 | 0.1 | ||||

A2→B1 | 0.03 | 0.04 | 0.11 | 0.08 | |||||

A1→A2 | (A1)→A2 | 0.05 | 0.09 | 0.17 | 0.06 |

Cross-Proficiency Levels of L2 | JSD of Grammar | JSD of Lexicon | |||||||
---|---|---|---|---|---|---|---|---|---|

POS-Trigram | Sub-Conj. | Token | Lemma | ||||||

A1→A2 | adjacent levels | 0.05 | (coef = −0.01, p = 0.087) | 0.09 | (coef = −0.027, p = 0.06) | 0.17 | (coef = −0.03, p = 0.046) | 0.06 | (coef = −0.005, p = 0.48) |

A2→B1 | 0.03 | 0.04 | 0.11 | 0.08 | |||||

B1→B2 | 0.02 | 0.015 | 0.1 | 0.06 | |||||

B2→C1 | 0.02 | 0.009 | 0.07 | 0.05 |

## Appendix C. The Data on Syntactic and Lexical Complexity

**Table A4.**Syntactic complexity for each L2 level in EFCAMDAT2 and the difference between different L2 levels.

Levels | MLS | MLT | MLC | C/S | VP/T | C/T | DC/C | DC/T | T/S | CT/T | CP/T | CP/C | CN/T | CN/C |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

A1 | 8.56 | 8.17 | 7.04 | 1.21 | 1.3 | 1.16 | 0.11 | 0.13 | 1.05 | 0.09 | 0.24 | 0.2 | 0.58 | 0.5 |

A2 | 11.59 | 10.09 | 7.58 | 1.53 | 1.33 | 1.33 | 0.21 | 0.29 | 1.15 | 0.23 | 0.32 | 0.24 | 0.76 | 0.57 |

B1 | 13.28 | 12.01 | 8.53 | 1.56 | 1.41 | 1.41 | 0.27 | 0.37 | 1.11 | 0.3 | 0.3 | 0.21 | 1.09 | 0.78 |

B2 | 14.85 | 13.6 | 8.93 | 1.66 | 1.5 | 1.52 | 0.33 | 0.5 | 1.09 | 0.34 | 0.32 | 0.21 | 1.34 | 0.88 |

C1 | 16.32 | 14.58 | 9.54 | 1.71 | 1.53 | 1.53 | 0.33 | 0.5 | 1.12 | 0.34 | 0.39 | 0.26 | 1.56 | 1.02 |

A1_C1 | 7.76 | 6.41 | 2.5 | 0.49 | 0.23 | 0.37 | 0.22 | 0.37 | 0.07 | 0.24 | 0.16 | 0.05 | 0.98 | 0.52 |

A2_C1 | 4.72 | 4.49 | 1.96 | 0.18 | 0.2 | 0.2 | 0.11 | 0.21 | −0.03 | 0.11 | 0.08 | 0.02 | 0.8 | 0.45 |

B1_C1 | 3.04 | 2.56 | 1.01 | 0.15 | 0.12 | 0.12 | 0.06 | 0.13 | 0.01 | 0.04 | 0.09 | 0.04 | 0.47 | 0.24 |

B2_C1 | 4.72 | 4.49 | 1.96 | 0.18 | 0.2 | 0.2 | 0.11 | 0.21 | −0.03 | 0.11 | 0.08 | 0.02 | 0.8 | 0.45 |

A2_A1 | 3.04 | 1.93 | 0.54 | 0.31 | 0.3 | 0.17 | 0.1 | 0.16 | 0.1 | 0.13 | 0.08 | 0.03 | 0.17 | 0.07 |

B1_A1 | 4.72 | 3.85 | 1.49 | 0.34 | 0.53 | 0.25 | 0.16 | 0.25 | 0.06 | 0.2 | 0.06 | 0.01 | 0.51 | 0.27 |

B2_A1 | 6.29 | 5.44 | 1.89 | 0.45 | 0.75 | 0.36 | 0.22 | 0.38 | 0.04 | 0.29 | 0.09 | 0.01 | 0.76 | 0.38 |

C1_A1 | 7.76 | 6.41 | 2.5 | 0.49 | 0.77 | 0.37 | 0.22 | 0.37 | 0.07 | 0.28 | 0.16 | 0.05 | 0.98 | 0.52 |

A2_A1 | 3.04 | 1.93 | 0.54 | 0.31 | 0.3 | 0.17 | 0.1 | 0.16 | 0.1 | 0.13 | 0.08 | 0.03 | 0.17 | 0.07 |

B1_A2 | 1.68 | 1.92 | 0.95 | 0.03 | 0.23 | 0.08 | 0.05 | 0.09 | −0.04 | 0.07 | −0.01 | −0.02 | 0.34 | 0.21 |

B2_B1 | 1.57 | 1.59 | 0.4 | 0.11 | 0.22 | 0.12 | 0.07 | 0.13 | −0.01 | 0.08 | 0.02 | 0 | 0.25 | 0.1 |

C1_B2 | 1.47 | 0.98 | 0.61 | 0.05 | 0.02 | 0 | 0 | −0.01 | 0.03 | −0.01 | 0.07 | 0.04 | 0.22 | 0.14 |

**Table A5.**Lexical complexity for each L2 level in EFCAMDAT2 and the difference between different L2 levels.

Levels | ld | ls1 | ls2 | vs1 | vs2 | cvs1 | ttr | msttr | cttr | rttr | logttr | uber | lv | vv1 | svv1 | cvv1 | vv2 | nv |
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|

A1 | 0.54 | 0.56 | 0.88 | 0.02 | 43.42 | 4.66 | 0.03 | 0.73 | 17.32 | 24.49 | 0.74 | 22.29 | 0.05 | 0.02 | 60.94 | 5.52 | 0.01 | 0.07 |

A2 | 0.53 | 0.52 | 0.86 | 0.04 | 71.49 | 5.98 | 0.04 | 0.75 | 17.55 | 24.82 | 0.75 | 22.32 | 0.08 | 0.04 | 106.91 | 7.31 | 0.01 | 0.11 |

B1 | 0.52 | 0.47 | 0.93 | 0.02 | 143.27 | 8.46 | 0.02 | 0.79 | 19.82 | 28.04 | 0.72 | 23.45 | 0.03 | 0.02 | 167.3 | 9.15 | 0.01 | 0.04 |

B2 | 0.51 | 0.48 | 0.96 | 0.01 | 180.81 | 9.51 | 0.01 | 0.79 | 20.61 | 29.14 | 0.71 | 23.98 | 0.02 | 0.01 | 198.99 | 9.97 | 0 | 0.03 |

C1_A1 | 0.52 | 0.49 | 0.93 | 0.02 | 186.36 | 9.65 | 0.02 | 0.8 | 20.15 | 28.5 | 0.73 | 23.47 | 0.03 | 0.02 | 218.08 | 10.44 | 0.01 | 0.05 |

A2_A1 | −0.01 | −0.04 | −0.02 | 0.02 | 28.07 | 1.32 | 0.01 | 0.02 | 0.23 | 0.33 | 0.01 | 0.03 | 0.03 | 0.02 | 45.97 | 1.79 | 0 | 0.04 |

B1_A1 | −0.02 | −0.09 | 0.05 | 0 | 99.85 | 3.8 | −0.01 | 0.06 | 2.5 | 3.55 | −0.02 | 1.16 | −0.02 | 0 | 106.36 | 3.63 | 0 | −0.03 |

B2_A1 | −0.03 | −0.08 | 0.08 | −0.01 | 137.39 | 4.85 | −0.02 | 0.06 | 3.29 | 4.65 | −0.03 | 1.69 | −0.03 | −0.01 | 138.05 | 4.45 | −0.01 | −0.04 |

C1_A1 | −0.02 | −0.07 | 0.05 | 0 | 142.94 | 4.99 | −0.01 | 0.07 | 2.83 | 4.01 | −0.01 | 1.18 | −0.02 | 0 | 157.14 | 4.92 | 0 | −0.02 |

A2_A1 | −0.01 | −0.04 | −0.02 | 0.02 | 28.07 | 1.32 | 0.01 | 0.02 | 0.23 | 0.33 | 0.01 | 0.03 | 0.03 | 0.02 | 45.97 | 1.79 | 0 | 0.04 |

B1_A2 | −0.01 | −0.05 | 0.07 | −0.02 | 71.78 | 2.48 | −0.02 | 0.04 | 2.27 | 3.22 | −0.03 | 1.13 | −0.05 | −0.02 | 60.39 | 1.84 | 0 | −0.07 |

B2_B1 | −0.01 | 0.01 | 0.03 | −0.01 | 37.54 | 1.05 | −0.01 | 0 | 0.79 | 1.1 | −0.01 | 0.53 | −0.01 | −0.01 | 31.69 | 0.82 | −0.01 | −0.01 |

C1_B2 | 0.01 | 0.01 | −0.03 | 0.01 | 5.55 | 0.14 | 0.01 | 0.01 | −0.46 | −0.64 | 0.02 | −0.51 | 0.01 | 0.01 | 19.09 | 0.47 | 0.01 | 0.02 |

## References

- Bulté, B.; Housen, A. Defining and operationalising L2 complexity. In Dimensions of L2 Performance and Proficiency: Complexity, Accuracy and Fluency in SLA; Housen, A., Kuiken, F., Vedder, I., Eds.; John Benjamins: Amsterdam, The Netherlands, 2012; pp. 23–46. [Google Scholar]
- Ortega, L.; Iberri-Shea, G. Longitudinal research in second language acquisition: Recent trends and future directions. Annu. Rev. Appl. Linguist.
**2005**, 25, 26–45. [Google Scholar] [CrossRef] - Skehan, P. Modelling second language performance: Integrating complexity, accuracy, fluency and lexis. Appl. Linguist.
**2009**, 30, 510–532. [Google Scholar] [CrossRef] - Pallotti, G. CAF: Defining, refining and differentiating constructs. Appl. Linguist.
**2009**, 30, 590–601. [Google Scholar] [CrossRef] - Housen, A.; De Clercq, B.; Kuiken, F.; Vedder, I. Multiple approaches to complexity in second language research. Second Lang. Res.
**2019**, 35, 3–21. [Google Scholar] [CrossRef] - Crossley, S. Linguistic features in writing quality and development: An overview. J. Writ. Res.
**2020**, 11, 415–443. [Google Scholar] [CrossRef] - Crossley, S.A.; Skalicky, S. Examining lexical development in second language learners: An approximate replication of Salsbury, Crossley & McNamara (2011). Lang. Teach.
**2019**, 52, 385–405. [Google Scholar] - Lu, X. Automatic analysis of syntactic complexity in second language writing. Int. J. Corpus Linguist.
**2010**, 15, 474–496. [Google Scholar] [CrossRef] - Lu, X. The relationship of lexical richness to the quality of ESL learners’ oral narratives. Mod. Lang. J.
**2012**, 96, 190–208. [Google Scholar] [CrossRef] - Miestamo, M. Implicational hierarchies and grammatical complexity. In Language Complexity as an Evolving Variable; Sampson, G., Gil, D., Trudgill, P., Eds.; Oxford University Press: Oxford, UK, 2009; pp. 80–97. [Google Scholar]
- Wagner, R.K.; Puranik, C.S.; Foorman, B.; Foster, E.; Wilson, L.G.; Tschinkel, E.; Kantor, P.T. Modeling the development of written language. Read. Writ.
**2011**, 24, 203–220. [Google Scholar] [CrossRef][Green Version] - De Clercq, B.; Housen, A. A cross-linguistic perspective on syntactic complexity in L2 development: Syntactic elaboration and diversity. Mod. Lang. J.
**2017**, 101, 315–334. [Google Scholar] [CrossRef] - Ellis, R. The differential effects of three types of task planning on the fluency, complexity, and accuracy in L2 oral production. Appl. Linguist.
**2009**, 3, 474–509. [Google Scholar] [CrossRef] - Housen, A.; Kuiken, F. Complexity, accuracy, and fluency in second language acquisition. Appl. Linguist.
**2009**, 30, 461–473. [Google Scholar] [CrossRef][Green Version] - Peters, J.; Mulling, K.; Altun, Y. Relative entropy policy search. In Proceedings of the Twenty-Fourth AAAI Conference on Artificial Intelligence, Atlanta, GA, USA, 11–15 July 2010; pp. 1607–1612. [Google Scholar]
- Vedral, V. The role of relative entropy in quantum information theory. Rev. Mod. Phys.
**2002**, 74, 197. [Google Scholar] [CrossRef][Green Version] - Sayood, K. Information theory and cognition: A review. Entropy
**2018**, 20, 706. [Google Scholar] [CrossRef][Green Version] - Joe, H. Relative entropy measures of multivariate dependence. J. Am. Stat. Assoc.
**1989**, 84, 157–164. [Google Scholar] [CrossRef] - Klingenstein, S.; Hitchcock, T.; DeDeo, S. The civilizing process in London’s Old Bailey. Proc. Natl. Acad. Sci. USA
**2014**, 111, 9419–9424. [Google Scholar] [CrossRef][Green Version] - Trotzke, A.; Zwart, J. The complexity of narrow syntax: Minimalism, representational economy and simplest Merge. In Measuring Grammatical Complexity; Newmeyer, F., Preston, L., Eds.; Oxford University Press: Oxford, UK, 2014; pp. 128–477. [Google Scholar]
- Arnold, J.; Wasow, T.; Losongco, A.; Ginstrom, R. Heaviness vs. newness: The effects of structural complexity and discourse status on constituent ordering. Language
**2000**, 17, 28–55. [Google Scholar] [CrossRef] - Dahl, O. The Growth and Maintenance of Linguistic Complexity; John Benjamins: Amsterdam, The Netherlands, 2004. [Google Scholar]
- Givón, T. The Genesis of Syntactic Complexity: Diachrony, Ontogeny, Neuro-Cognition, Evolution; John Benjamins: Amsterdam, The Netherlands, 2009. [Google Scholar]
- Kusters, W. Complexity in linguistic theory language learning and language change. In Language Complexity: Typology, Contact, Change; Miestamo, M., Sinnemäki, K., Karlsson, F., Eds.; John Benjamins: Amsterdam, The Netherlands, 2008; pp. 3–22. [Google Scholar]
- Szmrecsanyi, B.; Kortmann, B. Introduction: Linguistic complexity: Second language acquisition indigenization contact. In Linguistic Complexity: Second Language Acquisition, Indigenization, Contact; Kortmann, B., Szmrecsanyi, B., Eds.; De Gruyter: Berlin, Germany, 2012; pp. 6–34. [Google Scholar]
- Juola, P. The time course of language change. Comput. Humanit.
**2003**, 37, 77–96. [Google Scholar] [CrossRef] - Hughes, J.M.; Foti, N.J.; Krakauer, D.C.; Rockmore, D.N. Quantitative patterns of stylistic influence in the evolution of literature. Proc. Natl. Acad. Sci. USA
**2012**, 109, 7682–7686. [Google Scholar] [CrossRef] [PubMed][Green Version] - Unsworth, S. Comparing child L2 development with adult L2 development: How to measure L2 proficiency. In Current Trends in Child Second Language Acquisition; Gavruseva, E., Haznedar, B., Eds.; John Benjamins: Amsterdam, The Netherlands, 2008; pp. 301–336. [Google Scholar]
- DeKeyser, R.M. What makes learning second-language grammar difficult? A review of issues. Lang. Learn.
**2005**, 55, 1–25. [Google Scholar] [CrossRef] - Alexandre, Z.; Oleg, S.; Giovanni, P. An information-theoretic perspective on the costs of cognition. Neuropsychologia
**2019**, 123, 5–18. [Google Scholar] - Kuiken, F.; Vedder, I.; Housen, A.; De Clercq, B. Variation in syntactic complexity: Introduction. Int. J. Appl. Linguist.
**2019**, 29, 161–170. [Google Scholar] [CrossRef][Green Version] - Mazgutova, D.; Kormos, J. Syntactic and lexical development in an intensive English for Academic Purposes programme. J. Second Lang. Writ.
**2015**, 29, 3–15. [Google Scholar] [CrossRef][Green Version] - Ansarifar, A.; Shahriari, H.; Pishghadam, R. Phrasal complexity in academic writing: A comparison of abstracts written by graduate students and expert writers in applied linguistics. J. Engl. Acad. Purp.
**2018**, 31, 58–71. [Google Scholar] [CrossRef] - Paquot, M. Phraseological competence: A missing component in university entrance language tests? Insights from a study of EFL learners’ use of statistical collocations. Lang. Assess. Q.
**2018**, 15, 29–43. [Google Scholar] [CrossRef] - Paquot, M. The phraseological dimension in interlanguage complexity research. Second Lang. Res.
**2019**, 35, 121–145. [Google Scholar] [CrossRef] - Shannon, C.E. A mathematical theory of communication. Bell Syst. Tech. J.
**1948**, 27, 379–423. [Google Scholar] [CrossRef][Green Version] - Kullback, S.; Leibler, R.A. On information and sufficiency. Ann. Math. Stat.
**1951**, 22, 79–86. [Google Scholar] [CrossRef] - Sun, K.; Liu, H.; Xiong, W. The evolutionary pattern of language in scientific writings: A case study of Philosophical Transactions of Royal Society (1665–1869). Scientometrics
**2021**, 126, 1695–1724. [Google Scholar] [CrossRef] - Murdock, J.; Allen, C.; DeDeo, S. Exploration and exploitation of Victorian science in Darwin’s reading notebooks. Cognition
**2017**, 159, 117–126. [Google Scholar] [CrossRef][Green Version] - Chen, L.; Wise, S. Neuronal activity in the supplementary eye field during acquisition of conditional oculomotor associations. J. Neurophysiol.
**1995**, 73, 1101–1121. [Google Scholar] [CrossRef][Green Version] - Ehret, K.; Szmrecsanyi, B. Compressing learner language: An information-theoretic measure of complexity in SLA production data. Second Lang. Res.
**2019**, 35, 23–45. [Google Scholar] [CrossRef][Green Version] - Thoiron, P. Diversity index and entropy as measures of lexical richness. Comput. Humanit.
**1986**, 20, 197–202. [Google Scholar] [CrossRef] - Hale, J. Information-theoretical complexity metrics. Lang. Linguist. Compass
**2016**, 10, 397–412. [Google Scholar] [CrossRef] - Huang, Y.; Murakami, A.; Alexopoulou, T.; Korhonen, A. Dependency parsing of learner English. Int. J. Corpus Linguist.
**2018**, 23, 28–54. [Google Scholar] [CrossRef] - Römer, U. A corpus perspective on the development of verb constructions in second language learners. Int. J. Corpus Linguist.
**2019**, 24, 268–290. [Google Scholar] [CrossRef] - Lin, J. Divergence measures based on the Shannon entropy. IEEE Trans. Inf. Theory
**1991**, 37, 145–151. [Google Scholar] [CrossRef][Green Version] - Jurafsky, D.; Martin, J.H. Speech and Language Processing; Prentice Hall: Upper Saddle River, NJ, USA, 2008. [Google Scholar]
- Biber, D.; Johansson, S.; Leech, G.; Conrad, S.; Finegan, E.; Quirk, R. Longman Grammar of Spoken and Written English; Longman: London, UK, 1999; pp. 994–995. [Google Scholar]
- Degaetano-Ortlieb, S.; Teich, E. Towards an optimal code for communication: The case of scientific English. Corpus Linguist. Linguist. Theory
**2019**. [Google Scholar] [CrossRef][Green Version] - Comrie, B. The acquisition of relative clauses in relation to language typology. Stud. Second Lang. Acquis.
**2007**, 29, 301–309. [Google Scholar] [CrossRef] - Wells, J.B.; Christiansen, M.H.; Race, D.S.; Acheson, D.J.; MacDonald, M.C. Experience and sentence processing: Statistical learning and relative clause comprehension. Cogn. Psychol.
**2009**, 58, 250–271. [Google Scholar] [CrossRef][Green Version] - Kyle, K.; Crossley, S.A. Measuring syntactic complexity in L2 writing using fine-grained clausal and phrasal indices. Mod. Lang. J.
**2018**, 102, 333–349. [Google Scholar] [CrossRef] - Norris, J.M.; Ortega, L. Towards an organic approach to investigating CAF in instructed SLA: The case of complexity. Appl. Linguist.
**2009**, 30, 555–578. [Google Scholar] [CrossRef] - Lu, X.; Ai, H. Syntactic complexity in college-level English writing: Differences among writers with diverse L1 backgrounds. J. Second Lang. Writ.
**2015**, 29, 16–27. [Google Scholar] [CrossRef] - Vyatkina, N. The development of second language writing complexity in groups and individuals: A longitudinal learner corpus study. Mod. Lang. J.
**2012**, 96, 576–598. [Google Scholar] [CrossRef][Green Version] - De Bot, K.; Lowie, W.; Verspoor, M. A dynamic systems theory approach to second language acquisition. Biling. Lang. Cogn.
**2007**, 10, 7–21. [Google Scholar] [CrossRef][Green Version] - Kempe, V.; Brooks, P.J. Linking adult second language learning and diachronic change: A cautionary note. Front. Psychol.
**2018**, 9, 480. [Google Scholar] [CrossRef] [PubMed][Green Version] - Hamilton, J.D. Time Series Analysis; Princeton University Press: Princeton, NJ, USA, 2020. [Google Scholar]
- Hyndman, R.J.; Athanasopoulos, G. Forecasting: Principles and Practice. Available online: https://otexts.com/fpp2/ (accessed on 8 August 2021).
- Crossley, S.A.; McNamara, D.S. Does writing development equal writing quality? A computational investigation of syntactic complexity in L2 learners. J. Second Lang. Writ.
**2014**, 26, 66–79. [Google Scholar] [CrossRef] - Grosse, I.; Bernaola-Galván, P.; Carpena, P.; Román-Roldán, R.; Oliver, J.; Stanley, H.E. Analysis of symbolic sequences using the Jensen-Shannon divergence. Phys. Rev. E
**2002**, 65, 041905. [Google Scholar] [CrossRef][Green Version] - Liese, F.; Vajda, I. On divergences and informations in statistics and information theory. IEEE Trans. Inf. Theory
**2006**, 52, 4394–4412. [Google Scholar] [CrossRef][Green Version]

**Figure 1.**Relative entropy among L2 learners at cross-proficiency different levels (EFCAMDAT2). Note that JSD results are also visualized in this figure.

**Figure 2.**The difference of syntactic complexity between different L2 levels. Here, each L2 level (A1, A2, B1, B2, C1) can be treated as time order (date). Here, x-axis is syntactic complexity measures/metrics, and y-axis is the difference of those complexity metrics across proficiency levels (discussed in the section of Methods). The left top plot shows that MLC, MLS, and MLT have a gradual increase, that is, B1_A1 is higher than A2_A1, and B2_B1 is higher than B1_A1, and C1_B2 is higher than CB2_B1. When a metric shows a regular increase, it indicates that this measure can detect patterns of L2 proficiency development. By contrast, in the right top plot, such a regular increase can only be found in 4 of the 12 metrics. In the bottom two plots, none of metrics shows a regular increase. Irregular changes suggest that these metrics cannot capture the patterns of L2 proficiency development.

Word | 1810s | 1820s | 1830s | 1840s |
---|---|---|---|---|

Take | 728 | 556 | 665 | 529 |

Took | 168 | 327 | 351 | 333 |

Taken | 228 | 345 | 344 | 324 |

Taking | 133 | 144 | 164 | 165 |

Takes | 158 | 63 | 76 | 86 |

Total | 1415 | 1435 | 1600 | 1437 |

**Table 2.**Composition of the five sub-corpora of the essays section of the EFCAMDAT2 by language proficiency level.

L2 Learners’ Proficiency Levels | Texts | Learners | Tokens | Lemmas |
---|---|---|---|---|

A1 | 625,985 | 103,742 | 28.8 M | 27,065 |

A2 | 307,996 | 52,734 | 24 M | 32,051 |

B1 | 168,361 | 32,852 | 18.4 M | 26,276 |

B2 | 61,329 | 13,951 | 9.3 M | 21,312 |

C1 | 14,698 | 2839 | 2.8 M | 16,464 |

Cross-Proficiency Levels of L2 | KLD of Grammar | KLD of Lexicon | |||||||
---|---|---|---|---|---|---|---|---|---|

POS-Trigram | Sub-Conj. | Token | Lemma | ||||||

A1→A2 | A1→(A2, B1, B2, C1) | 0.37 | (coef = 0.21, p < 0.001) | 0.56 | (coef = 0.07, p = 0.31) | 1.41 | (coef = 0.03, p = 0.18) | 0.52 | (coef = 0.15, p = 0.2) |

A1→B1 | 0.58 | 0.81 | 1.41 | 0.84 | |||||

A1→B2 | 0.78 | 0.68 | 1.42 | 1.09 | |||||

A1→C1 | 1.0 | 0.82 | 1.52 | 0.93 | |||||

A2→B1 | A2→(B1, B2, C1) | 0.2 | (coef = 0.16, p = 0.13) | 0.24 | (coef = 0.12, p = 0.016) | 1.29 | (coef = −0.25, p = 0.13) | 0.71 | (coef = 0.025, p = 0.66) |

A2→B2 | 0.3 | 0.35 | 1.13 | 0.81 | |||||

A2→C1 | 0.51 | 0.47 | 0.78 | 0.76 | |||||

B1→B2 | B1→(B2, C1) | 0.13 | 0.085 | 1.1 | 0.62 | ||||

B1→C1 | 0.21 | 0.11 | 0.87 | 0.54 | |||||

B2→C1 | B2→(C1) | 0.16 | 0.052 | 0.52 | 0.51 |

Cross-Proficiency Levels of L2 | KLD of Grammar | KLD of Lexicon | |||||||
---|---|---|---|---|---|---|---|---|---|

POS-Trigram | Sub-Conj. | Token | Lemma | ||||||

A1→C1 | (A1, A2, B1, B2)→C1 | 1.0 | (coef = −0.28, p = 0.056) | 0.82 | (coef = −0.27, p = 0.034) | 1.52 | (coef = −0.29, p = 0.12) | 0.93 | (coef = −0.15, p = 0.034) |

A2→C1 | 0.51 | 0.47 | 0.78 | 0.76 | |||||

B1→C1 | 0.21 | 0.11 | 0.87 | 0.54 | |||||

B2→C1 | 0.16 | 0.05 | 0.52 | 0.51 | |||||

A1→B2 | (A1, A2, B1)→B2 | 0.78 | (coef = −0.33, p = 0.171) | 0.68 | (coef = −0.3, p = 0.04) | 1.42 | (coef = −0.16, p = 0.28) | 1.09 | (coef = −0.24, p = 0.07) |

A2→B2 | 0.3 | 0.35 | 1.13 | 0.81 | |||||

B1→B2 | 0.13 | 0.085 | 1.1 | 0.62 | |||||

A1→B1 | (A1, A2)→B1 | 0.58 | 0.81 | 1.41 | 0.84 | ||||

A2→B1 | 0.2 | 0.24 | 1.28 | 0.71 | |||||

A1→A2 | (A1)→A2 | 0.37 | 0.56 | 1.41 | 0.52 |

Cross-Proficiency Levels of L2 | KLD of Grammar | KLD of Lexicon | |||||||
---|---|---|---|---|---|---|---|---|---|

POS-Trigram | Sub-Conj. | Token | Lemma | ||||||

A1→A2 | adjacent levels | 0.37 | (coef = −0.07, p = 0.16) | 0.56 | (coef = −0.16, p = 0.086) | 1.41 | (coef = −0.28, p = 0.063) | 0.52 | |

A2→B1 | 0.2 | 0.24 | 1.28 | 0.71 | |||||

B1→B2 | 0.13 | 0.085 | 1.1 | 0.62 | |||||

B2→C1 | 0.16 | 0.081 | 0.52 | 0.51 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Sun, K.; Wang, R. Using the Relative Entropy of Linguistic Complexity to Assess L2 Language Proficiency Development. *Entropy* **2021**, *23*, 1080.
https://doi.org/10.3390/e23081080

**AMA Style**

Sun K, Wang R. Using the Relative Entropy of Linguistic Complexity to Assess L2 Language Proficiency Development. *Entropy*. 2021; 23(8):1080.
https://doi.org/10.3390/e23081080

**Chicago/Turabian Style**

Sun, Kun, and Rong Wang. 2021. "Using the Relative Entropy of Linguistic Complexity to Assess L2 Language Proficiency Development" *Entropy* 23, no. 8: 1080.
https://doi.org/10.3390/e23081080