Effects of Availability, Contingency, and Formulaicity on the Accuracy of English Grammatical Morphemes in Second Language Writing

Classified as probably not a substantial claim.

In L1 acquisition, high token frequency surface forms are produced earlier and more accurately than other forms and words inflected in lower token frequency forms.

“In L1 acquisition, high token frequency surface forms are produced earlier and more accurately in those forms compared to production in other forms and compared to production of other words that are inflected in lower token frequency forms (Aguado-Orea, 2004;Braine et al., 1990;Finley, 2018;Marchman, 1997).”

• cites Aguado-Orea, 2004Not matched

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

• cites Braine et al., 1990 sourcePartly supported

  The Discussion (page not available in our copy) reports that high-frequency suffixes were learned faster than low-frequency suffixes. The Children, Generalization Phase results also report greater accuracy for primary endings, but individual noun–suffix frequency improved ending accuracy only in a limited condition. Because the children learned a miniature artificial language, these results do not establish the claim as a general pattern of L1 acquisition. (LLM, conf=0.88)

  Experimental — accuracy not yet measured
• cites Finley, 2018 sourceNot checked — no full text of the source

May combine two claims.

More formulaic language is processed faster and acquired earlier than less formulaic language in both L1 and L2.

“Studies have demonstrated that more formulaic language is processed faster and acquired earlier than less formulaic language both in L1 (e.g., Bannard & Matthews, 2008;Tremblay, Derwing, Libben, & Westbury, 2011) and in L2 (for review, see Siyanova-Chanturia & Pellicer-Sánchez, 2018).”

• cites Bannard & Matthews, 2008 sourcePartly supported

  The Abstract (page not supplied) reports that three-year-olds repeated frequent word sequences faster than infrequent ones, supporting the L1 processing-speed claim. The paper tested repetition by two- and three-year-old monolingual English-speaking children; it did not establish that frequent sequences were acquired earlier. The L2 portion of the citing sentence is attached to a different citation, not this paper. (LLM, conf=0.88)

  Experimental — accuracy not yet measured
• cites Tremblay, Derwing, Libben, & Westbury, 2011 sourceNot checked — no full text of the source
• cites Siyanova-Chanturia & Pellicer-Sánchez, 2018Not matched

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

May combine two claims.

Reliability influences language change and accuracy and error patterns in L1 and L2 processing and acquisition.

“Reliability has been shown to influence language change (Bybee, 1985) as well as accuracy and error patterns in L1 (N. C. Ellis, O'Donnell, & Römer, 2014;Hay, 2001;Matthews & Theakston, 2006;Tatsumi et al., 2018) and L2 processing and acquisition (N.C. Ellis et al., 2014;Sugaya & Shirai, 2009).”

• cites Bybee, 1985 sourceNot checked — no full text of the source
• cites N. C. Ellis, O'Donnell, & Römer, 2014 sourcePartly supported

  Section 2.2 identifies contingency/reliability as a factor in language learning, and the Discussion reports that contingency affects verb generation by fluent language users. The Introduction also reports evidence concerning L2 acquisition. However, this paper does not establish effects on accuracy or error patterns in L1 or L2 processing and acquisition; its experiments measure verb generation rather than errors (page numbers unavailable in our copy). (LLM, conf=0.85)

  Experimental — accuracy not yet measured
• cites Hay, 2001 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Availability has been widely demonstrated to impact processing, acquisition, and use in both first language (L1) and second language (L2).

“Availability has been widely demonstrated to impact processing, acquisition, and use in both L1 (e.g., Ambridge, Kidd, Rowland, & Theakston, 2015) and L2 (e.g., N. C. Ellis, 2002;N. C. Ellis, Römer, & O'Donnell, 2016).”

• cites Ambridge, Kidd, Rowland, & Theakston, 2015 sourceSupported
• cites N. C. Ellis, 2002 sourceSupported
• cites N. C. Ellis, Römer, & O'Donnell, 2016Not matched

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

Classified as probably not a substantial claim.

R. Ellis (2005) reported factor analyses in which elicited imitation, oral narration, and timed grammaticality judgment tests loaded on Implicit Knowledge, whereas metalinguistic knowledge tests and untimed grammaticality judgment tests loaded on Explicit Knowledge.

“In designing and validating a battery of language tests that separately tap implicit and explicit learning, R. Ellis (2005) reported factor analyses of measures for L2 learners where elicited imitation, oral narration, and timed grammaticality judgment tests loaded on Implicit Knowledge whereas metalinguistic knowledge tests and untimed grammaticality judgment tests loaded on Explicit Knowledge.”

• cites R. Ellis (2005) sourceSupported

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

In Guo and Ellis's (2021) two experiments, there were no effects of lemma frequency.

“3. there were no effects of lemma frequency;”

• cites Guo and Ellis (2021) sourcePartly supported

  Experiment 1 Results report that lemma frequency was not a significant predictor of morpheme-production accuracy (p = 0.97). Experiment 2 Results list the predictors tested but do not report a separate lemma-frequency test, so our copy does not establish that this finding held in both experiments. (LLM, conf=0.84)

  Experimental — accuracy not yet measured

Classified as probably not a substantial claim.

The recognition and preservation of the correct form of target words in EIT is influenced by the forces of availability, contingency, and formulaicity.

“and likewise here too the recognition and preservation of the correct form of target words in EIT is influenced by the forces of availability, contingency, and formulaicity.”

• cites Guo and Ellis (2021) sourcePartly supported

  The Experiment 1 process analysis says recognition and preservation of correct target-word forms in EIT are *likely* influenced by availability, reliability, and formulaicity. The Introduction explains that measuring reliability entails estimating contingency, and the Conclusion reports EIT effects for all three factors. The claim is close in substance but states the proposed effect on recognition and preservation as a fact rather than a likelihood. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

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

In Guo and Ellis's (2021) two experiments, effects of reliability (but not frequency) were greater at lower levels of proficiency.

“4. effects of reliability (but not frequency) were greater at lower levels of proficiency;”

• cites Guo and Ellis (2021) sourceSupported

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

In Guo and Ellis's (2021) two experiments, word-form frequency had smaller but significant effects, particularly on plural -s and third-person present-tense -s.

“2. word-form frequency had smaller but significant effects particularly on plural -s and third-person present-tense -s;”

• cites Guo and Ellis (2021) sourceSupported

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

In Guo and Ellis's (2021) Experiment 2, higher frequency 4-word strings were associated with increased accuracy of production of the morphemes embedded therein (phrase-superiority effects).

“5. Experiment 2 also demonstrated phrase-superiority effects whereby higher frequency 4-word strings were associated with increased accuracy of production of the morphemes embedded therein.”

• cites Guo and Ellis (2021) sourceSupported

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

In Guo and Ellis's (2021) two experiments, reliability (contingency) had highly significant effects on the accuracy of provision, particularly of plural -s, third-person present-tense -s, and progressive -ing.

“1. there were highly significant effects of reliability (i.e., contingency) on the accuracy of provision (particularly of plural -s, third-person present-tense -s, and progressive -ing);”

• cites Guo and Ellis (2021) sourcePartly supported

  In the Experiment 1 and Experiment 2 Results, reliability was a highly significant overall predictor of provision accuracy in both experiments (p = 0.000). The morpheme-specific claim needs qualification: Experiment 1 found a significant effect for plural -s, a marginal effect for third-person -s, and no significant effect for progressive -ing. Experiment 2 found significant effects for all three, although the progressive -ing effect was p = 0.047. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

May combine two claims.

The current study has a larger scope and a more fine-grained picture of the effects of relevant factors than Guo and Ellis (2021).

“The analysis of a large-scale corpus allowed us to target a greater number and range of words and learners, leading to a study with a larger scope and a more fine-grained picture of the effects of relevant factors than those of the Guo and Ellis study.”

• cites Guo and Ellis (2021) sourceSupported

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

The perception and comprehension encoding stages involve three parts: (a) taking word-forms into an auditory/lexical buffer, (b) linking lexical items syntactically, and (c) constructing a meaningful interpretation of the sentence.

“The perception and comprehension encoding stages involve three parts: (a) taking word-forms into an auditory/lexical buffer, (b) linking lexical items syntactically, and (c) constructing a meaningful interpretation of the sentence.”

• cites Guo and Ellis (2021) sourcePartly supported

  The Experiment 1 section, “A Process Analysis of EIT as It Relates to Morpheme Production,” lists all three parts almost verbatim. It specifies that this account concerns the perception and comprehension encoding stages *in EIT*; the citing sentence omits that task-specific qualification. (LLM, conf=0.85)

  Experimental — accuracy not yet measured

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

The random-effects structure was close to the maximal model, except that interaction terms were not included as random slopes.

“This randomeffects structure was close to the maximal model (Barr, Levy, Scheepers, & Tily, 2013), except that we did not include the interaction terms as random slopes.”

• cites Barr, Levy, Scheepers, & Tily, 2013 sourceSupported

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

Appendix S3 presents prior predictive checks.

“Appendix S3 in the online Supporting Information presents the prior predictive checks (Schad, Betancourt, & Vasishth, 2021).”

• cites Schad, Betancourt, & Vasishth, 2021 sourcePartly supported

  The paper’s “Prior predictive checks” section describes and performs prior predictive checks. It does not establish what Appendix S3 of the citing paper’s online Supporting Information contains, so it supports the methodological reference but not the appendix-specific statement. (LLM, conf=0.82)

  Experimental — accuracy not yet measured

May combine two claims.

The finding replicated Murakami (2014), who did not identify a clear effect of L1 on third person -s or progressive -ing but found some L1 influence on plural -s.

“The finding, however, replicated Murakami (2014), who did not identify a clear effect of L1 on third person -s or progressive -ing but found some L1 influence on plural -s in an earlier version of EFCAMDAT.”

• cites Murakami (2014) sourcePartly supported

  The Summary of the regression analysis says plural -s is mildly affected by L1 and third person -s is relatively immune, supporting those parts of the claim. But it places progressive -ing among the morphemes most sensitive to L1 influence, rather than finding no clear effect. The longitudinal clustering analysis reports little apparent L1 effect for progressive -ing, so the claim reflects one analysis but overgeneralizes the paper’s findings. (LLM, conf=0.90)

  Experimental — accuracy not yet measured

• AssumptionThe authors state this as an assumption: The L1 of each nationality group is as specified (e.g., Brazil = Brazilian Portuguese, China and Taiwan = Mandarin Chinese, France = French, Germany = German, Italy = Italian, Japan = Japanese, Mexico = Spanish, Russia = Russian, Saudi Arabia = Arabic, and Turkey = Turkish).

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

Some of the findings of this study replicate those of Guo and Ellis (2021), while some do not.

“Some of these findings replicate those of Guo and Ellis (2021), who investigated these effects in an EIT; some do not.”

• cites Guo and Ellis (2021) sourceSupported

May combine two claims.

Guo and Ellis (2021) simply replaced pronoun contexts with proper-noun contexts, a blunt instrument that would have caused large reductions in formulaicity however it was measured.

“Guo and Ellis (2021) simply replaced pronoun contexts with proper-noun contexts, a blunt instrument that would have caused large reductions in formulaicity however it was measured.”

• cites Guo and Ellis (2021) sourcePartly supported

  Experiment 2’s Method (page unavailable) says the authors replaced pronouns with person names for the three verbal morphemes, and the resulting four-word strings typically had a corpus frequency of zero. But for noun plurals they inserted adjectives instead. The paper supports a large reduction in its four-word string-frequency measure, not the claim that the replacement was the entire manipulation or would reduce formulaicity however measured. (LLM, conf=0.88)

  Experimental — accuracy not yet measured

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

There is a wide range of ways to measure formulaicity, ranging from string frequency through measures of association including mutual information, mutual expectation, and P.

“there is a wide range of ways to measure formulaicity, ranging from string frequency through measures of association including mutual information, mutual expectation, P (see examples such as Saito, 2020 or Tavakoli & Uchihara, 2020)”

• cites Saito, 2020 sourceSupported
• cites Tavakoli & Uchihara, 2020Not matched

  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).

Only in the last few years has it become possible to bring together cognition, corpora, and computing at scale to triangulate research in usage-based language learning.

“Only in the last few years has it become possible to bring together cognition, corpora, and computing at scale to triangulate research in usagebased language learning (N.C. Ellis, 2017).”

• cites N.C. Ellis, 2017 sourcePartly supported

  The Introduction brings together corpus-linguistic, psychological, and psycholinguistic evidence for usage-based accounts, and Section 2 describes analysis of a 100-million-word corpus. The paper does not say that combining cognition, corpora, and computing at scale became possible only in the last few years; that timing claim is unsupported. (LLM, conf=0.83)

  Experimental — accuracy not yet measured

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

In L1 acquisition, the relative frequency of different forms of the same word predicts usage and error patterns in morpheme acquisition.

“We have already described how, in L1 acquisition, the relative frequency of different forms of the same word have been found to predict the usage and error patterns in morpheme acquisition (Bybee, 1985;Hay, 2001;Matthews & Theakston, 2006).”

• cites Bybee, 1985 sourceNot checked — no full text of the source
• cites Hay, 2001 sourceNot checked — no full text of the source
• cites Matthews & Theakston, 2006 sourceSupported

May combine two claims.

Availability and formulaicity were not predictors of accuracy of morpheme production in L2 English writing.

“Contrary to previous research investigating online processing, we did not identify in our study availability and formulaicity as predictors of accuracy of morpheme production in writing.”

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

For the interaction of writing number and frequency for third person -s, the 95% credible intervals of the posterior distribution of frequency only excluded 0 after 112 writings (for a single learner).

“only excluded 0 after 112 writings (for a single learner) for the interaction of writing number and frequency for third person -s”

May combine two claims.

Only a small number of learners in the data contributed 112/113 writings: four learners for the interaction of writing number and frequency for third person -s and three learners for the interaction of writing number and P for progressive -ing.

“only a small number of learners in our data contributed 112/113 writings (i.e., four learners in the interaction of writing number and frequency for third person -s and three learners in the interaction of writing number and P for progressive -ing).”

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

Reliability and accuracy were positively associated for all the morphemes except for progressive -ing.

“Reliability and accuracy were positively associated for all the morphemes except for progressive -ing.”

May combine two claims.

The 95% credible intervals of the frequency parameter only excluded 0 for progressive -ing, suggesting that the effects of token frequency could not be reliably identified in our data.

“The 95% credible intervals of the frequency parameter only excluded 0 for progressive -ing, suggesting that the effects of token frequency could not be reliably identified in our data.”

• AssumptionThe authors state this as an assumption: The effects of frequency on accuracy are multiplicative.

May combine two claims.

P was positively correlated with accuracy for progressive -ing and third person -s, but the relationship was unclear for the other two morphemes.

“P was positively correlated with accuracy for progressive -ing and third person -s, but the relationship was unclear for the other two morphemes.”

May combine two claims.

P was not strongly associated with accuracy, except for past tense -ed where it was negatively correlated with accuracy.

“As was the case for frequency, P was not strongly associated with accuracy, except for past tense -ed where it was negatively correlated with accuracy.”

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

The distribution of nationality groups in the data was skewed, with Brazilian learners occupying 43% to 48% of observations for each morpheme, followed by Chinese, Mexican, and German learners each contributing 6% to 13% for each morpheme.

“The distribution of the nationality groups was skewed, with Brazilian learners occupying 43% to 48% of observations for each of the morphemes, followed by Chinese, Mexican, and German learners each contributing 6% to 13% for each morpheme.”

May combine two claims.

There was no clear association between frequency and accuracy, except for progressive -ing (across proficiency levels) and for third person -s (intermediate- to high-proficiency levels).

“However, we could not identify a clear association between frequency and accuracy, except for progressive -ing (across proficiency levels) and for third person -s (intermediate-to high-proficiency levels).”

May combine two claims.

Writing number was positively correlated with accuracy, although the 95% credible intervals excluded 0 only for past tense -ed and for progressive -ing.

“Writing number was also positively correlated with accuracy, although the 95% credible intervals excluded 0 only for past tense -ed and for progressive -ing.”

May combine two claims.

For L1 type, 95% credible intervals excluded 0 only for plural -s, where the PRESENT group outperformed the ABSENT group.

“Similarly, for L1 type, 95% credible intervals excluded 0 only for plural -s, where the PRESENT group outperformed the ABSENT group.”

• AssumptionThe authors state this as an assumption: The L1 of each nationality group is as specified (e.g., Brazil = Brazilian Portuguese, China and Taiwan = Mandarin Chinese, France = French, Germany = German, Italy = Italian, Japan = Japanese, Mexico = Spanish, Russia = Russian, Saudi Arabia = Arabic, and Turkey = Turkish).

May combine two claims.

The only interaction parameters whose 95% credible intervals did not include 0 were the interaction of mean Englishtown level and frequency for third person -s, the interaction of writing number and frequency for third person -s, and the interaction of writing number and P for progressive -ing.

“the only parameters for which 95% credible intervals did not include 0 were the interaction of the mean Englishtown level and frequency for third person -s, the interaction of writing number and frequency for third person -s, and the interaction of writing number and P for progressive -ing.”

May combine two claims.

Frequency looked to be positively correlated with accuracy for progressive -ing, third person -s, and plural -s but not for past tense -ed.

“Frequency looked to be positively correlated with accuracy for progressiveing, third person -s, and plural -s but not for past tense -ed.”

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

For the interaction of writing number and P for progressive -ing, the 95% credible intervals of the posterior distribution of P only excluded 0 after 113 writings.

“113 writings for the interaction of writing number and P for progressive -ing”

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

There appeared to be positive associations in Figure 1 for progressive -ing, third person -s, and plural -s.

“Although there appeared to be positive associations in Figure 1 for progressive -ing, third person -s, and plural -s”

May combine two claims.

Our study did not identify consistent effects of availability or formulaicity on morpheme accuracy.

“our study did not identify consistent effects of availability or formulaicity on morpheme accuracy.”

• AssumptionRests on an assumption: The large corpora used reflect typical language exposure for L2 learners.
• AssumptionRests on an assumption: The learner corpora are representative of L2 English learners broadly.

Classified as probably not a substantial claim.

Cues that are rare and unreliable are learned late and are relatively weaker, even in adults.

“Cues that are rare and unreliable are learned late and are relatively weaker, even in adults (MacWhinney, 1997).”

• cites MacWhinney, 1997Not matched

  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).

Psychological research has demonstrated profound and omnipresent impacts of contingency in the learning of cue-outcome associations.

“Psychological research into animal and human learning alike has demonstrated profound and omnipresent impacts of contingency in the learning of cue-outcome associations (Shanks, 1995).”

• cites Shanks, 1995 sourceNot checked — no full text of the source

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

Some morphemes are acquired earlier than others.

“Some morphemes are acquired earlier than others as, for example, the classic studies of morpheme acquisition order demonstrated (e.g., Dulay & Burt, 1973, 1974).”

• cites Dulay & Burt, 1973, 1974Not matched

  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).

Many adult L2 learners will never acquire total control of L2 morphology, even after tens of years of English immersion.

“At the extreme, many adult L2 learners will never acquire total control of L2 morphology, even after tens of years of English immersion (e.g., in spoken production, Johnson & Newport, 1989;Schmidt, 1984).”

• cites Johnson & Newport, 1989 sourceNot checked — no full text of the source
• cites Schmidt, 1984Not matched

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

May combine two claims.

The finding that contingency predicts morpheme accuracy supports the usage-based view that language learners are sensitive to distributional properties in their input.

“thereby supporting the usage-based view that language learners are sensitive to distributional properties in their input.”

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

Contingency learning is central to the competition model, a psycholinguistic theory of language acquisition and sentence processing.

“Contingency learning is central to the competition model, a psycholinguistic theory of language acquisition and sentence processing (MacWhinney, 1987;MacWhinney & Bates, 1989), as well as to other psycholinguistic models of construction learning such as the rational learning of form-function contingencies (N.C. Ellis, 2006).”

• cites MacWhinney, 1987 sourceNot checked — no full text of the source
• cites MacWhinney & Bates, 1989Not checked — no full text of the source
• cites N.C. Ellis, 2006 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

For reliability, the analysis of the three verbal morphemes targeted only words with a minimum of five occurrences in each CEFR level A1 through C1.

“In addition, for the sake of reliability, we targeted only words with a minimum of five occurrences in each of the CEFR levels A1 through C1 in the analysis of the three verbal morphemes.”

May combine two claims.

In parsing and interpreting the target morphemes, there are influences of syntactic integrity, for example, auxiliary [be] impacting particularly progressive -ing, and influences of contextual support that could impact the encoding of the past -ed.

“In parsing and interpreting the target morphemes, there are influences of syntactic integrity, for example, auxiliary [be] impacting particularly progressive -ing and influences of contextual support that could impact the encoding of the past -ed.”

Classified as probably not a substantial claim.

The study replaced punctuation, capitalization, and space-related errors with teacher-corrected portions before automated analysis.

“we replaced the original learner writing with the corresponding teacher-corrected portion insofar as these errors were concerned”

• AssumptionRests on an assumption: Space-related errors often lead to erroneous part-of-speech tagging due to inaccurate tokenization.

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

There is immense scope for creativity in choosing where, and at what levels of grain, to follow the norms and where to exploit them.

“There is immense scope for creativity in choosing where, and at what levels of grain, to follow the norms and where to exploit them (Hanks, 2013).”

• cites Hanks, 2013 sourceNot checked — no full text of the source

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

Writing involves much more conscious analysis and attention to complex aspects of writing during decision making, problem-solving behaviors involving heuristic searches, and well-differentiated control strategies.

“it involves much more conscious analysis and attention “to complex aspects of writing during decision making; problem-solving behaviors involving heuristic searches; and well-differentiated control strategies” (Cumming, 1989, p. 81).”

• cites Cumming, 1989 sourceNot checked — no full text of the source

May combine two claims.

Elicited imitation requires fluent online perception involving implicit or automatized processing, unlike written composition.

“Unlike written composition (i.e., with content generated by the learner), it requires fluent online perception involving implicit or automatized processing.”

• AssumptionRests on an assumption: Frequency effects are more often evidenced in automatic online language processing than in more considered production tasks where there is more time for conscious creation and editing.
• AssumptionRests on an assumption: Implicit and automatized processing takes place automatically, ballistically, and unconsciously.
• AssumptionRests on an assumption: Implicit processing runs on the system of linguistic representations that have been tuned by prior usage.

Classified as probably not a substantial claim.

The more the demand for immediacy and automaticity of production, the greater the likely impact of these factors.

“The more the demand for immediacy and automaticity of production, the greater the likely impact of these factors.”

Classified as probably not a substantial claim.

Shatz (2020) estimated the correct topics of writings in EFCAMDAT.

“Shatz (2020) estimated the correct topics of writings in EFCAMDAT”

• cites Shatz (2020) sourceNot checked — no full text of the source

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

Writing is considerably slower than speech.

“On the whole, writing is considerably slower than speech”

• cites Cumming, 1989 sourceNot checked — no full text of the source

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

Cue-weights differ between languages because different languages use different cues to signal meanings.

“Cue-weights differ between languages because different languages use different cues to signal meanings.”

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

Models of L1 auditory sentence processing such as chunk-and-pass processing suggest that the language system rapidly integrates all available incoming information, interactively satisfying multiple probabilistic constraints as quickly as possible, to update the current interpretation of what has been said so far.

“Models of L1 auditory sentence processing such as chunk-and-pass processing (Christiansen & Chater, 2016) suggest that the language system rapidly integrates all available incoming information, interactively satisfying multiple probabilistic constraints as quickly as possible, to update the current interpretation of what has been said so far.”

• cites Christiansen & Chater, 2016Not matched

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

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Incremental identification of incoming units is influenced by the sequential probabilities of what has been processed to date: the next word in a well-entrenched word sequence is more easily identified, as is an incoming morpheme that is highly predicted in its context.

“Incremental identification of incoming units is influenced by the sequential probabilities of what has been processed to date: The next word in a well-entrenched word sequence is more easily identified as is an incoming morpheme that is highly predicted in its context.”

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

Conscious strategies in writing override the distributional norms offered up from the automatized/implicit system.

“These conscious strategies override the distributional norms that are offered up from the automatized/implicit system.”

May combine two claims.

The encoding of third person present -s and plural -s on subjects is also under the influence of syntactic integrity, although in English, agreement processing is generally less obligatory than processing for tense and aspect.

“The encoding of third person present -s and plural -s on subjects is also under the influence of syntactic integrity, although in English, agreement processing is generally less obligatory than processing for tense and aspect (MacWhinney, 1997(MacWhinney, , 2008)).”

• cites MacWhinney, 1997Not matched

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

• cites MacWhinney, 2008 sourceNot checked — no full text of the source

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

The highly controlled contexts of morpheme occurrences and the use of elicited imitation as a language processing task afforded experimental validity through controlled decontextualized language perception and repetition.

“Also, the highly controlled contexts of the morpheme occurrences and the use of elicited imitation as a language processing task afforded experimental validity through controlled decontextualized language perception and repetition at the expense of ecological validity in situations of rich, meaningful, and morenaturalistic communication.”

Classified as probably not a substantial claim.

Examining multiple morphemes increases confidence that distributional factors such as frequency are general principles underlying accurate morpheme use if the factors are positively correlated with accuracy across the morphemes.

“We examined multiple morphemes because we could be more confident that distributional factors such as frequency are the general principles underlying accurate morpheme use if the factors were positively correlated with accuracy across the morphemes.”

May combine two claims.

The final stage of EIT, production (whether oral or written), could also be sensitive to frequency effects and sequential probabilities at word, morpheme, and, particularly, phrasal levels: a well-entrenched formulaic phrase can support provision of its component morphemes whether they are analyzed or not.

“The final stage of EIT, production (whether oral or written), could also be sensitive to frequency effects and sequential probabilities at word, morpheme, and, particularly, phrasal levels: A well-entrenched formulaic phrase can support provision of its component morphemes whether they are analyzed or not (Arnon & Clark, 2011).”

• cites Arnon & Clark, 2011 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

L2 learners must learn which cues are important in which languages.

“Thus, L2 learners must learn which cues are important in which languages.”

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

Formulaic language is more common in speech than in writing.

“Formulaic language is more common in speech than in writing (Erman & Warren, 2000)”

• cites Erman & Warren, 2000 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Explicit knowledge correlates more strongly with written skills (writing and reading) than implicit knowledge does.

“explicit knowledge correlated more strongly with the written skills (writing and reading).”

• cites Elder and Ellis (2009) sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

The study annotated original and corrected versions with part-of-speech tags via TreeTagger with the English model to solve ambiguities.

“We then annotated the original (i.e., learner-written) and corrected (i.e., teacher-corrected) versions of each writing with part-of-speech tags via TreeTagger (Schmid, 1994) with the provided English model in order to solve ambiguities”

• cites Schmid, 1994Not matched

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

• AssumptionRests on an assumption: TreeTagger with the provided English model solves part-of-speech ambiguities such as plural -s vs. third person -s.

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

Most writing is far less mundane than texting, and at the other extreme, writing can look more like this current article, the result of numerous edits, reworkings, and conscious revisions.

“However, most writing is far less mundane than texting, and at the other extreme, writing can look more like this current article, the result of numerous edits, reworkings, and conscious revisions.”

Classified as probably not a substantial claim.

Consciousness affords novelty.

“consciousness affords novelty (N.C. Ellis, 2015).”

• cites N.C. Ellis, 2015 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Automaticity relies upon memorized structures and habits.

“Automaticity relies upon memorized structures and habits”

• cites N.C. Ellis, 2015 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Implicit knowledge correlates more strongly with oral skills (speaking and listening) than explicit knowledge does.

“implicit knowledge correlated more strongly with the oral skills (speaking and listening)”

• cites Elder and Ellis (2009) sourceNot checked — no full text of the source

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

Relevant cues include sentence-internal information about lexical and structural biases as well as extra-sentential cues from the referential and pragmatic context.

“Relevant cues include sentence-internal information about lexical and structural biases as well as extra-sentential cues from the referential and pragmatic context”

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

As the incoming auditory information is chunked, it is rapidly integrated with prior information to recognize words and morphemes that are in turn chunked into larger multiword units.

“As the incoming auditory information is chunked, it is rapidly integrated with prior information to recognize words and morphemes that are in turn chunked into larger multiword units.”

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

Psycholinguistic research has demonstrated a variety of frequency effects in the perception and processing of words, morphemes, multiword chunks, and syntactic constructions.

“Psycholinguistic research has demonstrated a variety of frequency effects in the perception and processing of words, morphemes, multiword chunks, and syntactic constructions”

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

L2 learners begin with cues that are more available in the input and/or those they are most attuned to from their L1, after which they come to rely upon cues that are more reliable in their interpretations.

“To do this, they begin with cues that are more available in the input and/or those that they are most attuned to from their L1 (MacWhinney, 2008), after which they come to rely upon cues that are more reliable in their interpretations.”

• cites MacWhinney, 2008 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

The study retained spelling errors uncorrected.

“we retained these spelling errors uncorrected”

• AssumptionRests on an assumption: The correction of grammatical morphemes is occasionally coded as spelling errors in EF-CAMDAT.

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

Appendix S5 presents the numerical summary of each parameter in each model as well as histograms showing the posterior distribution of each fixed-effects parameter.

“Appendix S5 in the online Supporting Information presents the numerical summary of each parameter in each model as well as the histograms showing the posterior distribution of each fixed-effects parameter.”

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

Cognitive linguistic theories of construction grammar focus on lexical, morphological, and syntactic forms as form-function pairings.

“Cognitive linguistic theories of construction grammar focus on lexical, morphological, and syntactic forms as form-function pairings (Goldberg, 1995(Goldberg, , 2006;;Trousdale & Hoffmann, 2013).”

• cites Goldberg, 1995Not matched

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

• cites Goldberg, 2006Not checked — no full text of the source
• cites Trousdale & Hoffmann, 2013Not matched

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

Classified as probably not a substantial claim.

Contingency of association is key in cognitive-linguistic, corpus-based, and statistical models of language structure such as collostructional analysis.

“Contingency of association is similarly key in cognitive-linguistic, corpus-based, and statistical models of language structure like collostructional analysis (Gries & Stefanowitsch, 2004;Stefanowitsch & Gries, 2003).”

• cites Gries & Stefanowitsch, 2004 sourceNot checked — no full text of the source
• cites Stefanowitsch & Gries, 2003 sourceNot checked — no full text of the source

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

The analysis of the interaction effects was restricted to the range of the mean Englishtown level and writing number observed in the study, avoiding extrapolation.

“without extrapolation (i.e., within the range of the mean Englishtown level and writing number observed in this study)”

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

The priors regularized the intercept, slopes, and random-effects standard deviations toward 0, penalizing extreme values and potentially allowing more reliable parameter estimation.

“The priors, thus, regularized the intercept, slopes, and random-effects standard deviations toward 0, penalizing extreme values and potentially allowing more reliable parameter estimation despite the complexity of the model.”

May combine two claims.

The results were partly consistent with Murakami and Alexopoulou (2016), who found that L1 hardly influenced the accuracy of third person -s although it exerted a stronger effect on plural -s.

“The results were also partly consistent with Murakami and Alexopoulou (2016), who found that L1 hardly influenced the accuracy of third person -s although it exerted a stronger effect on plural -s in the Cambridge Learner Corpus (Nicholls, 2003).”

• cites Murakami and Alexopoulou (2016)Not matched

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

• cites Nicholls, 2003Not matched

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

• AssumptionThe authors state this as an assumption: The L1 of each nationality group is as specified (e.g., Brazil = Brazilian Portuguese, China and Taiwan = Mandarin Chinese, France = French, Germany = German, Italy = Italian, Japan = Japanese, Mexico = Spanish, Russia = Russian, Saudi Arabia = Arabic, and Turkey = Turkish).

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

The R code and data used for modeling are available at IRIS and OSF.

“For the sake of transparency, we have made available the R code used for the preprocessing of corpora, data retrieval, and data analysis, together with the data used for modeling (Murakami & Ellis, 2022) at IRIS (www.iris-database.org)and OSF (https://osf.io/ba8mf/).”

• cites Murakami & Ellis, 2022Not matched

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

May combine two claims.

Bayesian models were suited for our analysis because the number of target words was not particularly large and weakly informative priors allow more reliable parameter estimation.

“Bayesian models were suited for our analysis because the number of target words was not particularly large and Bayesian analyses with weakly informative priors would allow us to more reliably estimate the parameters of interest (see Gudmestad, House, & Geeslin, 2013, for similar arguments in L2 research).”

• cites Gudmestad, House, & Geeslin, 2013 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Bayesian methods yield the full posterior distribution that is more informative than point estimates.

“(b) the full posterior distribution that is more informative than point estimates”

• cites Kruschke, 2014Not checked — no full text of the source
• cites Lambert, 2018 sourceNot checked — no full text of the source
• cites McElreath, 2020 sourceNot checked — no full text of the source

Classified as probably not a substantial claim.

Bayesian methods yield more intuitive uncertainty metrics.

“(c) more intuitive uncertainty metrics”

• cites Kruschke, 2014Not checked — no full text of the source
• cites Lambert, 2018 sourceNot checked — no full text of the source
• cites McElreath, 2020 sourceNot checked — no full text of the source

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

We did not perform variable selection due to the many problems associated with this procedure.

“We did not perform variable selection due to the many problems associated with this procedure (e.g., Harrell, 2015).”

• cites Harrell, 2015 sourceNot checked — no full text of the source

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

The first three levels of Englishtown correspond to A1 in CEFR, the second three levels to A2, and so forth.

“According to EF Education First, the first three levels of Englishtown correspond to A1 in CEFR, the second three levels to A2, and so forth (Geertzen et al., 2014).”

• cites Geertzen et al., 2014Not matched

  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).

All the models converged reasonably well (all R values < 1.05).

“All the models converged reasonably well (all R values < 1.05).”

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

Appendix S4 presents trace plots, posterior predictive checks, and a sensitivity analysis.

“Appendix S4 in the online Supporting Information presents trace plots, posterior predictive checks, and a sensitivity analysis.”

Classified as probably not a substantial claim.

COCA (version released March 2020) can be used as a first approximation of L2 learners' input.

“We computed availability, contingency, and formulaicity based on the COCA version released in March 2020 (Davies, 2008-) that has been used in previous research as a first approximation of L2 learners' input (e.g., Monteiro, Crossley, & Kyle, 2020;Wolter & Yamashita, 2018).”

• cites Davies, 2008-Not matched

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

• cites Monteiro, Crossley, & Kyle, 2020Not matched

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

• cites Wolter & Yamashita, 2018Not matched

  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).

Contingency of cue-outcome mapping is a driving force of all associative learning, human and animal alike.

“Contingency of cue-outcome mapping is a driving force of all associative learning, human and animal alike, to the degree that the field in which it is studied has become known as contingency learning (Shanks, 1995).”

• cites Shanks, 1995 sourceNot checked — no full text of the source

May combine two claims.

In a case study of a native Russian speaker learning Japanese over 10 months, verbs that are more consistently conjugated in a certain common form compared to other possible forms were produced exclusively in the common form early in the learning trajectory, whereas this preferential bias was not observed for verbs that do not have a common form.

“They found that verbs that are more consistently conjugated in a certain common form compared to other possible forms, such as siru”

• cites Sugaya and Shirai (2009) sourceNot checked — no full text of the source

Probably not a substantial claim: may not pass S2 (checkable). May combine two claims.

Form-form and form-function associations interact in various complex and adaptive ways in usage, and speakers' language systems reflect the history of their processing these associations.

“Form-form and form-function associations interact in various complex and adaptive ways in usage, and speakers' language systems reflect the history of their processing these associations.”

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

Collostructional analysis focuses on form-form reliability in measuring the degree of attraction or repulsion that words exhibit to constructions.

“Collostructional analysis focuses on form-form reliability in measuring the degree of attraction or repulsion that words exhibit to constructions.”

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

We held firm to our choice of formulaicity measure rather than fishing with different measures.

“We held firm to this choice rather than fishing with different measures.”

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

The aspect hypothesis has investigated these patterns in the initial acquisition of tense-aspect for past tense -ed, progressive -ing, and third person -s.

“The aspect hypothesis has investigated these patterns in the initial acquisition of tense-aspect for past tense -ed, progressive -ing, and third person -s”

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

Functional theories of SLA emphasize the interplay of form and meaning in acquisition.

“Functional theories of SLA emphasize the interplay of form and meaning in acquisition.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based). May combine two claims.

Morphemes and lemmas are linguistic constructions with particular functions and meanings; they are symbolic.

“morphemes and lemmas go beyond being mere forms, they are linguistic constructions with particular functions and meanings; they are symbolic.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

When learners are processing usage, they are tallying the associations among forms, among interpretations, and between forms and their interpretations.

“When learners are processing usage, they are tallying the associations among forms, among interpretations, and between forms and their interpretations.”

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

Word n-gram does not capture nonadjacent structural dependency of words or association between abstract categories.

“It, for example, does not capture nonadjacent structural dependency of words (e.g., subject and verb) or association between abstract categories (e.g., part of speech).”

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

Availability entails that higher-proficiency learners will have at their disposal a wider range of items than will have low-proficiency learners.

“Availability entails that higher-proficiency learners will have at their disposal a wider range of items than will have low-proficiency learners.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

In creative written production, in response to the demands of the prompt and the current state of exposition, various ideas come to the minds of learners, and they respond with whatever available and relevant language forms that are associated with these ideas.

“In contrast, in creative written production, in response to the demands of the prompt and the current state of exposition, various ideas come to the minds of learners, and they respond with whatever available and relevant language forms that are associated with these ideas.”

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

Associative learning theory demonstrates that contingency of association trumps token frequency.

“Associative learning theory demonstrates that contingency of association trumps token frequency (as described in the section Contingency).”

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

Associative learning theory explains that people learn contingencies that are relevant to their predictive processing.

“Associative learning theory explains that people learn contingencies that are relevant to their predictive processing.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

Cognitive linguistics explains why certain form-meaning relations are more contingent in usage.

“Cognitive linguistics explains why certain form-meaning relations are more contingent in usage.”

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

Verbs and morphemes can be more or less reliably associated (form-form contingency).

“Verbs and morphemes can be more or less reliably associated (form-form contingency).”

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

EITs prioritize online implicit and automatized processing running on the system of linguistic representations that have been tuned by prior usage, whereas writing involves more explicit, conscious processing.

“EITs prioritize online implicit and automatized processing running on the system of linguistic representations that have been tuned by prior usage, whereas writing, in contrast, involves more explicit, conscious processing.”

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

The aspect hypothesis emphasizes these effects in the initial acquisition of tense-aspect.

“The aspect hypothesis emphasizes these effects in the initial acquisition of tense-aspect.”

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

A major limitation of our study is the lack of transparency of process while learners write.

“A major limitation of our study is the lack of transparency of process while learners write.”

Probably not a substantial claim: may not pass S4 (based). May combine two claims.

Cognitive linguistic theories of construction grammar view lexical, morphological, and syntactic forms as symbolic form-function pairings and hold that people learn language from usage.

“Cognitive linguistic theories of construction grammar view lexical, morphological, and syntactic forms as symbolic form-function pairings and hold that people learn language from usage.”

Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing), S4 (based).

It is quite likely that learners are more sensitive to certain types of structural and semantic association than to others.

“It is quite likely that learners are more sensitive to certain types of structural and semantic association than to others.”

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

Our work is open and transparent as a result of our making available the R code and data.

“our work is open and transparent as a result of our making available the R code and data.”

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

The range of learners is wider than that of typical experimental studies.

“the range of learners is wider than that of typical experimental studies”

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

A possible reason for the absence of noticeable token frequency effects is that we only examined reasonably frequent forms in which frequency effects had presumably approached ceiling.

“Another possible reason is that we only examined the reasonably frequent forms (i.e., inflected forms) in which frequency effects had presumably approached ceiling.”

• AssumptionThe authors state this as an assumption: Frequency effects had approached ceiling in the reasonably frequent forms examined.

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

Our research is reasonably large in scale.

“it is reasonably large in scale”

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

Bardovi-Harlig and Comajoan-Colomé concluded from their review of about 30 studies that the aspect hypothesis accurately predicts the adult L2 acquisition of past morphology in a number of languages.

“Bardovi-Harlig and Comajoan-Colomé concluded from their review of about 30 studies that the aspect hypothesis accurately predicts the adult L2 acquisition of past morphology in a number of languages.”

• cites Bardovi-Harlig and Comajoan-ColoméNot matched

  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), S4 (based).

The present investigations would not have been possible without large corpora of language that reflect typical language exposure as well as large learner corpora covering tens of thousands of participants ranging widely in demographics.

“The present investigations would not have been possible without large corpora of language that reflect typical language exposure as well as large learner corpora covering tens of thousands of participants ranging widely in demographics.”

Classified as probably not a substantial claim.

One answer for the absence of noticeable token frequency effects lies in the differences between EITs and free written production.

“One answer lies in the differences between EITs and free written production.”

Classified as probably not a substantial claim.

The study manually calculated the precision and recall of the identification of errors and accurate uses.

“Also, and more importantly, we manually calculated the precision and recall of the identification of errors and accurate uses (see Appendix S1 in the online Supporting Information).”

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

The difference between the present study's null effects of raw frequency and formulaicity and the previous experimental study's positive effects is possibly due to methodological differences between the experimental task and the written compositions in EFCAMDAT.

“This difference is possibly due to methodological differences between the experimental task (elicited imitation, demanding online processing of predetermined stimuli) and the more conscious, written compositions in EF-CAMDAT analyzed for the current study, in which learners decide which language they need, and so the language may be less prone to effects of formulaicity or the frequency of inflect”

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

Murakami and Alexopoulou (2016) found a strong influence of L1 on progressive -ing, a finding inconsistent with the present study.

“Murakami and Alexopoulou (2016) also found a strong influence of L1 on progressive -ing, a finding inconsistent with the present study.”

• cites Murakami and Alexopoulou (2016)Not matched

  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).

The accuracy of POS tagging per se is not relevant to the study's conclusions because the accuracy was examined at a lower end of the data retrieval process.

“Because we examined the accuracy at a lower end of the data retrieval process, the accuracy of POS tagging per se is not relevant.”