Диссертация (1136198), страница 23
Текст из файла (страница 23)
We observed attraction with attractors of all threegenders, but only with N and F heads. The gender of the attractordid not even influence the size of the effect. These results suggestthat the gender of the attractor has very little or no influence onits chances to be retrieved (it should only match the gender of theincorrect verb form).Notably, Julie Franck expressed similar ideas in a recenttalk (Franck, 2015). The first part of the talk was dedicatedto summarizing existing data on agreement attraction.Franck adopted the retrieval approach for production andcomprehension and identified the following groups of factors thatcan lead to attraction: semantic factors (primarily related to theconceptual numerosity of the subject NP), stability of the head’sfeatures, accessibility of the attractor (defined by its structuralposition) and similarity between the head and the attractor.Discussing stability of the head’s features Franck examinedasymmetries between feature values, morphophonological andsemantic influences.Franck’s reexamination of attraction phenomena was drivenby the findings on morphophonology (other data she consideredcould be accounted for in the old models).
As we noted inthe introduction, studies on several languages demonstrated thatnumber and gender agreement attraction errors are less frequentwhen heads have regular inflections, but this plays no role forattractors (e.g., Bock and Eberhard, 1993; Vigliocco et al., 1995;Vigliocco and Zilli, 1999; Franck et al., 2008). For attractors,only morphological ambiguity making them more similar to asubject is important (e.g., Hartsuiker et al., 2003; Badecker andKuminiak, 2007)12 . This led Franck to conclude that the featuresof the head are crucial, and she reanalyzed existing data accordingto this idea.
She argued that features that have a semanticcorrelate are more resistant to attraction (for example, Viglioccoand Franck, 1999 observed lower error rates when heads hadconceptual rather than purely grammatical gender) and that thesame is true for marked feature values. The latter conclusionwas based on number agreement attraction findings and onthe results of Badecker and Kuminiak’s and our productionexperiments.Thus, the findings summarized by Franck and the outcomeof our reading experiments point into the same direction,but we still have to explain the difference between ourcomprehension and production results. Of course, to makedefinitive conclusions, it would be great to have data from severallanguages (for example, comprehension data from Slovak), butlet us suggest several hypotheses based on existing findings.Our reading experiments strongly indicate that M heads areresistant to attraction, while N and F heads are not.
Thedata from production experiments on Russian and Slovak areopen to several interpretations because attraction was observedin all head-attractor combinations with mismatched genders.Therefore, we assume that M heads in general are the moststable ones and the least prone to attraction, and production dataneed an independent explanation. This assumption is supportedby independent evidence: several production experiments onnumber agreement attraction in Russian reported by Nicol andWilson (1999) and Yanovich and Fedorova (2006) demonstratedthat the incidence of number errors depends on the gender of thehead noun.
Errors arise most often with N heads and least oftenwith M ones.If our assumption is on the right track, M heads and pluralheads exhibit similar properties in comprehension. But whyshould they do so, given that M features are neither the mostmarked nor the least marked in Russian? Let us come back to theidea expressed in the previous subsection: number is privativelymarked (i.e., singular nouns have no number features), whilegender is not (all nouns have some gender features with plusand minus values). We hypothesize that with privative features,the non-zero value is the most stable, while with non-privativefeatures, where all values are non-zero, other considerationscome into the picture. We are reluctant to appeal to frequency,but maybe it plays a role that M gender vastly outnumbers Fand N in Russian.
In any case, our data indicate that that thereis no straightforward relation between feature markedness andstability. The next subsection considers some differences betweencomprehension and production and how these differences couldexplain our results.6.3. Differences between Production andComprehensionBased on parallel results from number agreement attractionexperiments most authors assume that the same mechanismsunderlie attraction in production and comprehension. Theopposite view has been recently advocated by Tanner et al. (2014).They claim that the mechanisms responsible for attraction incomprehension are a subset of those involved in production.In particular, they argue that attraction in comprehension isdue to retrieval interference, while attraction in production isbest described by the representational account, namely, by theMarking and Morphing model (Eberhard et al., 2005), althoughretrieval interference is also present.As we noted above, the Marking and Morphing model isincompatible with gender agreement attraction.
We believe thatthe core mechanism underlying number and gender agreementattraction in production is the same, so we opt for theretrieval approach. Evidently, in case of number, semanticfactors influence agreement, and it is expected that theirinfluence is much more readily detected in production thanin comprehension: in production, we start with the conceptualstructure, while in comprehension, it is our goal. Vigliocco andFranck (1999) demonstrated that gender agreement attractionerrors are less frequent when head nouns have conceptual, ratherthan purely grammatical gender. So semantic factors also play arole here, but, given the relevant distinctions13 between number12 Let us add that Badecker and Kuminiak (2007) demonstrated that ambiguityis important not only for attractors, but also for heads: if the form is ambiguousbetween nominative and accusative, the chances of the head to be retrieved arelower.Frontiers in Psychology | www.frontiersin.org13 In case of number, we can find many words that are formally singular, but denoteplural entities, for example, nouns like crowd or heads of the phrases like the labelon the bottles that have a distributive interpretation.
Gender is usually semantically1755November 2016 | Volume 7 | Article 1651Slioussar and MalkoGender Agreement Attraction in Russianand gender, this role is different: they mainly reduce the size ofthe effect. It would be very interesting to assess their influence ongender agreement attraction in comprehension: we expect that itshould be much smaller, as in case of number agreement. Thus,the differences between production and comprehension noted byTanner et al. (2014) may also be relevant for gender agreement,but the picture revealed by our experiments cannot be explainedby them.In the previous subsection we argued that agreementattraction patterns in comprehension are due to the fact thatheads with plural features and M features are resistant toattraction, i.e., that during the retrieval process, they tend tobe identified correctly, while the retrieval of heads with otherfeatures can be disturbed by attractors.
Findings summarized byFranck (2015) show that the stability of head’s features shouldalso be relevant for agreement attraction in production. This isfurther confirmed by the results from Nicol and Wilson (1999)to Yanovich and Fedorova (2006) indicating that heads withM features are indeed more stable when we look at numberagreement production in Russian. Based on these data, we wouldexpect to see no errors in MF and MN conditions in productionexperiments on gender agreement, but this is not what we found.To address this problem, we should specify in more detail howretrieval may work in comprehension and production.
Wagerset al. (2009) who analyze comprehension show that the retrievalaccount has two versions that may be difficult to tease apartbased on the current experimental data. On the one hand, cuebased retrieval may be initiated every time we deal with anagreeing verb. On the other hand, we may predict the featuresof the upcoming verb relying on the subject NP and initiateretrieval only when our predictions are not met. Both versionsgive roughly the same results if we assume that when the truesubject matches all the cues, it is successfully retrieved in theabsolute majority of cases.
Then in both scenarios, problems areexpected only when we encounter an incorrect verb form and thesentence contains an attractor a non-subject NP that matches theincorrectly specified feature of the verb.We believe that two similar scenarios can also be distinguishedfor production: we can decide which features we need on anagreeing predicate while processing the subject or once we get tothe predicate. Accordingly, retrieval might be initiated every timewe deal with an agreeing predicate or only when a wrong verbform that does not match our predictions is spuriously generated.The models proposed by Solomon and Pearlmutter (2004) orby Badecker and Kuminiak (2007) instantiate the first scenario.For example, Solomon and Pearlmutter argue that attraction inproduction arises because two nouns, the head of the subjectNP and the attractor, are simultaneously active in the syntacticstructure, and a wrong agreement controller may be selected.However, we argue for the second scenario below.To summarize, in comprehension, we construct the set ofretrieval cues based on the verb form that is provided to us.
As wedemonstrated above, different versions of the account share thisbasic observation. If the first scenario is adopted for production(the features of the upcoming verb are predicted, and retrieval isinitiated only when we spuriously generate a wrong verb form),the picture should be quite similar: the set of cues will be basedon this form.However, we do not believe that this scenario is the mostplausible.
![](https://s.studizba.com/z.php?f=/templates/si/i/noavatar.png&w=100&h=100&t=1"){kind=avatar}
