Диссертация (1136198), страница 19
Текст из файла (страница 19)
The number of errors in each category is given in Table 2.In case of self-corrections, only the first variant was counted,both when participants changed an answer with an error to acorrect one and when they did the opposite (this happened inthree cases).At the following stage of analysis, we collapsed all agreementerrors together. The distribution of errors by experimentalconditions is given in Table 3. In total, there were 77 agreementerrors (5.4% from all responses). Only 13 out of them werenot due to attraction (they are discussed in more detail below).The difference between the number of agreement errors withand without attraction is statistically significant according to the2chi-square test9 [χ(1,N = 77) = 18.97, p < 0.01], so our resultsshow that gender agreement in Russian is subject to attraction.As Table 3 shows, agreement errors were more frequent inpredicate mismatch conditions, but were not limited to them.Out of 13 errors without attraction, in eight cases, a mismatchedpredicate was not changed, but there were also five cases whereparticipants produced a neuter predicate with an MF subject,a masculine predicate with an NN subject etc., although theywere provided with other forms, matched or mismatched withthe subject.
Out of 64 attraction errors, 11 errors occurred inpredicate match conditions, i.e., participants changed the correctgender of the predicate they were provided with to an incorrectone due to attraction.Conditions with matched and mismatched predicates arecollapsed in Table 4 showing that the number of agreementattraction errors differs depending on the combination of gendersof the head and attractor nouns.
To test whether these differencesare statistically significant, we modeled the data with a mixedeffects logistic regression in the statistical software program R(R Core Team, 2014) using the glmer function from the lme4package (Bates et al., 2015).Firstly, we compared MF and FM conditions. The logisticregression evaluated the likelihood of an agreement attractionAgr.
errorAgr. errorOther(attraction)(no attraction)errorsCondition 1 (MM + M)a690016Condition 2 (MM + F)690016Condition 3 (MF + M)663120Condition 4 (MF + F)531911725Condition 5 (FF + F)6500Condition 6 (FF + M)570330Condition 7 (FM + F)661023Condition 8 (FM + M)501012916Condition 9 (MM + M)7400Condition 10 (MM + N)640323Condition 11 (MN + M)691020Condition 12 (MN + N)591102025Condition 13 (NN + N)6401Condition 14 (NN + M)680220Condition 15 (NM + N)626121Condition 16 (NM + M)6313014a Dueto our mistake, there are 85 responses in conditions 1 and 2 rather than 90.TABLE 4 | The Number of gender agreement attraction errors by conditionin Experiment 1.Head/attractorCorrectAttractionOtherresponseserrorserrorsMF1192239FM1161153MN1361242NM1251936gendererror (coded as 1) vs.
a correct response (coded as 0). Thecombination of genders was treated as a fixed effect. For thepredictors we used contrast coding: MF was coded as 0.5,FM was coded as −0.5. Random intercepts by participant andby item were also included in the model. The results of theanalysis are reported in Table 5. The coefficient for the interceptwas significant, reflecting that most responses were correct.There was also a significant main effect of Gender Combinationindicating that F attractors trigger significantly more errors thanM attractors.Secondly, we compared MN and NM conditions in thesame way. MN was coded as 0.5, NM was coded as−0.5. The coefficient for the intercept was again significantbecause most responses were correct. But the main effectof gender combination did not reach significance.
We alsocompared MF and MN conditions and FM and NM conditions,as well as the number of non-agreement (“other”) errorsin different conditions, but did not find any significantdifferences.9 In half of the conditions, where the genders of the head and the attractorcoincided, no agreement errors with attraction were possible, while in the otherhalf of the conditions, these errors prevailed, but there were also agreement errorswithout attraction, as Table 3 shows. This is why we chose the chi-square test.Frontiers in Psychology | www.frontiersin.orgCorrectresponse846November 2016 | Volume 7 | Article 1651Slioussar and MalkoGender Agreement Attraction in Russian(10)TABLE 5 | Results of the analysis for Experiment 1.ConditionsPredictorMF vs.
FM(Intercept)GenComb(Intercept)GenCombMN vs. NMCoefficientStd. errorWald Zp−3.040.43−7.01<0.01−0.950.48−1.960.05−2.680.30−8.82<0.010.620.391.590.112.5. DiscussionThe results of Experiment 1 are similar to the results of B&K’sfirst experiment, which can be explained by the fact that thetwo languages have similar gender systems, as we demonstratedin the introduction. In both studies, F attractors triggered moreerrors than M attractors.
N attractors triggered fewer errors thanM attractors, but this difference was statistically significant onlyin B&K’s study. As we mentioned in the introduction, otherauthors studying gender attraction in French and Spanish (whichhave two genders and where M is grammatical default), observeda different pattern: there were more errors with M attractorsthan with F attractors. We postpone further discussion until thegeneral discussion section.c. Receptna porošokbylarecipeM.NOM.SG for powderM.ACC.SG wasF.SGpom’atojiz-za sil’nogocrumpledF.SG due.to strongGEN.SGvolnenijapacienta.nervousnessGEN.SG patientGEN.SGd. Receptna maz’bylarecipeM.NOM.SG for ointmentF.ACC.SG wasF.SGpom’atojiz-za sil’nogocrumpledF.SG due.to strongGEN.SGvolnenijapacienta.nervousnessxGEN.SG patientGEN.SG3. EXPERIMENT 2AExperiment 2a was designed to find out whether genderagreement attraction can also be detected in comprehension. Forthe sake of comparison with Experiment 1, we used the samecombinations of head and attractor noun genders.Additionally, we constructed 120 fillers, which had roughly thesame structure as experimental sentences.
Subject NPs in fillersconsisted of a single noun modified by an adjective, or of acomplex NP, where the embedded noun was not in accusative.All fillers were grammatical. Thus, we had 24 ungrammaticaland 144 grammatical sentences, making the grammatical-toungrammatical ratio 6:1. Experimental sentences and fillers weredistributed in four counterbalanced experimental lists. Every liststarted with ten fillers; then stimuli and fillers were presented inpseudo-random order with the constraint that a maximum of twostimuli could occur consecutively.3.1.
ParticipantsForty-eight native Russian speakers (19 female and 29 male) tookpart in the experiment. Ages ranged from 19 to 26 (mean age 20.9,SD 1.9).3.2. MaterialsThe materials consisted of target and filler sentences. All targetsentences were 9–10 words long and followed the schema: NP1 –preposition–NP2 –copula (byt’) - adjective/participle - four-fivewords modifying the predicate.
We had the same 16 conditionsas in Experiment 1 (see Table 1 above). Almost all subject NPsand predicates were based on the materials from Experiment 1and followed the same constraints. In half of the conditions, thepredicate did not agree with the subject.
Given existing findingson number agreement attraction, we expected parallel results inproduction and comprehension. In particular, we expected tofind grammaticality illusions in conditions MFF, FMM, MNN,and NMM (this would mean that they would be read significantlyfaster than the other four ungrammatical conditions: MMF, FFM,MMN, NNM).As in Experiment 1, conditions were grouped in sets, eachset containing four conditions with the same head nouns. Anexample of a stimuli set is given in (10)10 . For each condition setwe constructed 12 sentences, 48 target sentences in total.10 The translation3.3. ProcedureThe sentences were presented on a PC using Presentationsoftware (http://www.neurobs.com). We used the word-by-wordself-paced reading methodology (Just et al., 1982).
Each trialbegan with a sentence in which all words were masked withdashes while spaces and punctuation marks remained intact.Participants were pressing the space bar to reveal a word andre-mask the previous one. One third of the sentences wasfollowed by forced choice comprehension questions to ensurethat the participants were reading properly. Two answer variantswere presented on the left and on the right of the screen.Participants pressed “f ” to choose the answer on the left, and “j”to choose the answer on the right. Participants were instructedto read at a natural pace and answer questions as accuratelyas possible. They were not informed in advance that sentenceswould contain errors. An experimental session lasted around14 min.for all sentences is identical, so we only give it for the first one.Frontiers in Psychology | www.frontiersin.orga.
Receptna porošokbylrecipeM.NOM.SG for powderM.ACC.SG wasM.SGpom’atymiz-za sil’nogocrumpledM.SG due.to strongGEN.SGvolnenijapacienta.nervousnessGEN.SG patientGEN.SG“The recipe for the powder was crumpled due to thepatient’s extreme nervousness.”b. Receptna maz’bylrecipeM.NOM.SG for ointmentF.ACC.SG wasM.SGpom’atymiz-za sil’nogocrumpledM.SG due.to strongGEN.SGvolnenijapacienta.nervousnessGEN.SG patientGEN.SG947November 2016 | Volume 7 | Article 1651Slioussar and MalkoGender Agreement Attraction in RussianFIGURE 1 | Plots of mean RTs (in ms) by conditions in Experiment 2a.
Error bars represent standard errors of the means. Regions: NP1 (1) - preposition (2) NP2 (3) - copula byt’ (4) - Adj/Part (5) - spillover (6–9). Ungrammatical conditions are red, grammatical ones are blue. Conditions where the gender of the attractor andthe predicate coincide (for example, FFF and FMM) have dark colors, conditions where they do not (for example, FMF and FFM) have light colors. (A) Feminine head,masculine and feminine attractors, (B) Masculine head, feminine and masculine attractors, (C) Neuter head, masculine and neuter attractors, and (D) Masculine head,neuter and masculine attractors.3.4.
Resultsmodifying the predicate. The results of the tests for the relevantregions are given in Table 6.We analyzed participants’ question-answering accuracy andreading times. Two participants answered more than 20%questions incorrectly, so their data were discarded. Otherwiseno participant made more than two mistakes when answeringquestions to target sentences (i.e., 10% at most). Readingtimes that exceeded a threshold of 2.5 standard deviations, byregion and condition, were excluded (Ratcliff, 1993).
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
