Диссертация (1136198), страница 22
Текст из файла (страница 22)
The conditions with M attractors are blue, with F attractors - red, with N attractors - green. Dark colors indicategrammatical conditions, light colors - ungrammatical conditions. (A) Feminine heads, (B) Neuter heads.TABLE 10 | Results of the analysis for Experiment 3.ConditionsFF vs. FNRegion56NN vs. NF56FactordfMSeffectF1pdfMSeffectF2p<0.01Gram1.3542202.8513.54<0.011.1727667.3614.05GenMatch1.3524964.008.730.011.1714489.693.890.07GenMatch * Gram1.3545411.6112.98<0.011.1713689.613.500.08<0.01Gram1.35187272.5617.49<0.011.1789662.0112.62GenMatch1.3567487.3820.50<0.011.1724857.072.330.15GenMatch * Gram1.3565732.4111.47<0.011.1740945.687.260.020.02Gram1.3585176.4223.48<0.011.1741103.227.33GenMatch1,3520168.735.430.031.176258.941.050.32GenMatch * Gram1.3525434.935.580.021.177525.602.480.13<0.01Gram1.35284248.9232.88<0.011.17136808.0941.17GenMatch1.3579120.3112.49<0.011.1725233.813.980.06GenMatch * Gram1.3576130.017.650.011.1729658.304.780.04Analyses with p ≤ 0.05 are shown in bold.5.5.
Discussion6. GENERAL DISCUSSIONLet us summarize the results of Experiments 2a, 2b, and 3. Genderagreement attraction was observed with F heads and M or Nattractors and with N heads and M or F attractors, but not withM heads and F or N attractors. This leads us to the conclusionthat attraction depends primarily on the features of the headrather than on the features of the attractor. If the features ofthe attractor played an additional role, ungrammatical sentenceswith M attractors would be read faster than ungrammaticalsentences with other attractors. However, when we comparedsentences with F heads and N or M attractors and sentenceswith N heads and F or M attractors, the Attractor Gender orthe interaction between this factor and Grammaticality neverreached significance, and average RTs even showed the oppositepattern: they were longer in the ungrammatical sentences with Mattractors. This goes against the assumptions entertained in theabsolute majority of previous agreement attraction studies, so adetailed analysis of this result will be presented in the GeneralDiscussion Section.In this paper we reported four experiments on gender agreementattraction in Russian.
We observed attraction effects both inproduction and in comprehension. Badecker and Kuminiak(2007) is the only previous production study where genderagreement attraction was examined in a language with threegenders (Lorimor et al., 2008 elicited very few gender errorsin their experiments on Russian). In this paper, we replicatedone of Badecker and Kuminiak’s experiments and conductedthe first comprehension experiments analyzing attraction withnon-binary features.Two outcomes of our experiments can be identified as themost important. Firstly, our results suggest that gender attractionworks differently in production and comprehension. This doesnot agree with previous studies of number agreement attraction,in which production and comprehension results were largelyparallel: only the combination of a singular head and a pluralattractor triggered attraction.
Secondly, our reading experimentsFrontiers in Psychology | www.frontiersin.org1553November 2016 | Volume 7 | Article 1651Slioussar and MalkoGender Agreement Attraction in Russiansuggest that the features of the head, rather than the features ofthe attractor are crucial to determine the pattern of agreementattraction, while the absolute majority of previous agreementattraction studies rely on the opposite assumption.preambles, while in Spanish and French the situation is theopposite.
Badecker and Kuminiak (2007) do not comment on thisdiscrepancy, and we cannot offer any explanation for it so far. Wecan only note that the pattern observed in Slovak and Russian issimilar to what we see with number: more errors when the headis less marked than the attractor.Now let us turn to comprehension experiments. Attractionwas observed in NMM, NFF, FMM, and FNN conditions, but notin MFF and MNN conditions. As we already noted, this indicatesthat features of the heads rather than features of the attractorsplay a crucial role for attraction. Before discussing this finding inthe next section, we want to make two important observations.Firstly, the M gender exhibits a different pattern from the F andN genders. This can hardly be attributed to feature markedness:N is the grammatical default in Russian, and the psycholinguisticrelevance of this fact is confirmed by the production datadiscussed above.
We will explore alternative explanations below.Secondly, no ungrammaticality illusions (differences betweengrammatical conditions depending on whether the head and theattractor have matched or mismatched gender features) weredetected in our experiments, which lends further support to theretrieval approach to agreement attraction.6.1. Overview of Experimental FindingsIn our comprehension experiments, attraction was observed insome combinations of head and attractor genders, but not inthe others, while in the production experiment, all combinationsexhibited attraction, only to a different extent. We will firstconsider production results, and then comprehension findings.The outcome of the production study was similar to the resultsof the first experiment conducted by Badecker and Kuminiak(2007): there were more errors with MF subjects than with FMsubjects and with NM subjects than with MN subjects.
Bothdifferences were statistically significant in the Slovak study, whilein our experiment, only the first one was.Badecker and Kuminiak ran an additional experimentcomparing NF and NM preambles and found that the errorrates in these conditions were roughly the same. They claimthat this pattern can be explained only in an optimality-theoreticframework where markedness effects are by definition relational.We believe that this is not the case. Given the impressive body ofliterature on number and gender features, we do not think thatwe can select a particular approach based on experimental datawithout a detailed consideration of other arguments. So we chosetwo models that have been applied to Russian to demonstrate thatthey are also compatible with the pattern described by Badeckerand Kuminiak and may be better suited to explain other findingswe reported.In Kramer (2015), F is encoded as [+FEM], M is [−FEM] andN corresponds to no gender features.
When zero and non-zerofeature values are compared, the latter are marked, and it canbe argued that for this comparison, it is not important whethernon-zero values are plus or minus. Therefore, the same errorrates are observed with NF and NM preambles. When non-zerovalues are compared, plus values are more marked. In Nevins(2011), F is [+FEM], [−MASC], M is [−FEM], [+MASC] and Nis [−FEM], [−MASC]. N is less marked than M and F because itcontains only minus values, while M and F both contain one plusvalue. But when we compare F and M directly, it can be arguedthat feature hierarchy becomes important.
[FEM] is standardlyassumed to be lower than [MASC], so F is more marked than M.Now let us focus on another property of production findingsfrom Slovak and Russian that is not discussed by Badecker andKuminiak (2007), but seems crucial to us. In case of genderagreement, attraction errors are produced with all preambles inwhich the genders of the head and the attractor are mismatched,while in case of number agreement, errors are virtually absentwith plural heads and singular attractors.
One way to capturethis would be to assume that all genders are marked by somefeature combinations, as Nevins (2011) suggests, while singularcorresponds to no number features.Another important problem is the difference betweenexperimental findings from Slovak and Russian on the one handand Romance languages on the other. In Russian and Slovak,more errors are produced with MF preambles than with FMFrontiers in Psychology | www.frontiersin.org6.2. The Role of Head and AttractorFeatures in AttractionIn the literature on agreement attraction, the presence or absenceof the effect is traditionally associated with the features ofthe attractor.
There are at least two reasons for this. Firstly,experimental findings suggest that some properties of attractorsdo influence attraction effects [e.g., as we discussed in theintroduction, Hartsuiker et al. (2003) showed that the incidenceof agreement errors was much higher when attractors wereformally similar to nominative plural forms]. The second reasonis tradition: the first proposed account of agreement attractionrelied on feature percolation, which means focusing exclusivelyon the attractor whose features can erroneously spread upwards.The assumption that the features of the attractor are crucialhas been maintained in the more recent retrieval account.However, it is important to realize that in this account theproperties of the head can influence the agreement process aswell.
For example, to explain the plural markedness effect, it istraditionally assumed that singular nouns are not marked fornumber, and “the system is biased to return explicitly numbermarked constituents” (Wagers et al., 2009, p. 233), thereforeplural attractors can easily be retrieved, while singular onesalmost never are.
But another interpretation is possible: the pluralfeature makes the heads easier to retrieve and thus more stable,less prone to attraction errors. This is why attraction in the pluralsingular configurations is virtually non-existent. On the otherhand, the retrieval of singular heads is prone to error, hence theabundance of errors in singular-plural configurations11 .While we look at binary features or at the cases whereattraction is observed in all feature combinations (as inproduction experiments on Slovak and Russian), we can only11 Letus note that under this scenario an attractor can also be retrieved in asingular-singular configuration, but this will not provoke any agreement errors.1654November 2016 | Volume 7 | Article 1651Slioussar and MalkoGender Agreement Attraction in Russianuse indirect evidence to estimate the contribution of headand attractor features to the agreement process. Our readingexperiments allow for the first direct comparison and show that atleast in comprehension, the features of heads, not attractors playthe crucial role.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
