Gender mismatch in Russian: quantitative study Текст научной статьи по специальности «Науки о Земле и смежные экологические науки»

А.А. Герасимова

Московский государственный университет им. МБ. Ломоносова, 119991 г. Москва, Российская Федерация; Московский педагогический государственный университет, 119991 г. Москва, Российская Федерация

Рассогласование по роду в русской именной группе: квантитативное исследование

Статья посвящена явлению рассогласования по роду в русском языке, которое рассматривается с синхронной и диахронической точек зрения. С помощью экспериментального исследования устанавливается частотное распределение моделей атрибутивного и предикативного согласования для именных групп, допускающих рассогласование по роду. Затем проводится сравнение данных современного русского языка с результатами более ранних статистических исследований. Результаты сравнения говорят в пользу анализа, согласно которому смена родового признака происходит в зоне референциальной семантики именной группы.

Ключевые слова: рассогласование по роду, референциальный род, именная группа, группа определителя, русский язык.

A^. Gerasimova

Lomonosov Moscow State University, Moscow, 119991, Russian Federation; Moscow State University of Education, Moscow, 119991, Russian Federation

Gender mismatch in Russian: quantitative study1

This paper addresses the issue of gender mismatch in Russian. I discuss mixed agreement patterns in Russian nominal phrases and evaluate different approaches

1 The study has been supported by Russian Scientific Foundation (RSF), project #16-18-02003 "Structure of meaning and its mapping into lexical and functional categories of Russian" at MSPU.

to their analysis. Using experimental data I estimate the current distribution of gender mismatch patterns for both attributive and predicate agreement. I compare recent data with statistical studies from mid-twentieth century and then discuss possible interpretations of the change in distribution of the patterns. In particular, I show that the phenomenon of gender mismatch in Russian correlates with the referential domain in the syntactic structure. New empirical evidence presented in the paper provides both synchronic and diachronic estimation of gender mismatch frequency in Russian.

Key words: gender mismatch, mixed agreement, noun phrase, DP, referential domain, Russian.

1. Gender agreement patterns

In Russian the standard agreement pattern implies the same gender values for adnominals and verbs inflected for past tense choosing between masculine, feminine and neuter agreement. However, there is a set of nouns that refer to humans denoting their position, profession, degree etc. and trigger masculine grammatical agreement, but that are also used for denoting female humans as far as they do not have a feminine parallel. When referring to women, in nominative case they may trigger both masculine and feminine agreement and, as a consequence, gender mismatch may occur: constituents may demonstrate different values of the same feature.

The phenomenon of mixed agreement appeared not long ago. [Muchnik, 1971] and [Crockett, 1976] date the first examples back to 1920s. Due to historic reasons it became essential to differentiate nouns that denote social status or profession by gender. For a group of nouns, it was impossible to derive feminine parallel for several reasons: derivational affixes might add expressive meaning (vrach - vrachiha 'doctor') or lead to phonetically difficult alternations (hirurg - hirurzhka 'surgeon') (examples from [Muchnik, 1971]).

Thus, constituents inside noun phrases with such heads may demonstrate different values of gender feature: not only masculine grammatical value but also feminine referential value. According to grammars, the preferred agreement pattern is when adnominals agree with the formal features of the noun and predicate shows referential agreement. It is important to note that adnominals may bear feminine agreement only if the predicate demonstrates the same value. Moreover, adnominals cannot trigger referential feminine gender if adnominals which have higher position in the syntactic structure or the predicate are masculine. In (1) the paradigm adopted from [Lyutikova, 2015] with added demonstratives is exemplified.

(1) a. masculine grammatical agreement

Этот новый

etot novyj

this-M.NOM.SG neW-M.NOM.SG

пришел

prishel

amved-M.sG

b. feminine predicate agreement

этот новый

etot novyj

this-M.NOM.SG neW-M.NOM.SG

зубной zubnoj

dental-M.NOM.sG

■ эта eta

this-F.NOM.SG

пришла

prishla

arrived-F.sG

e. ill-formed

*эта eta

this-F.NOM.SG

f. ill-formed

*этот *etot

this-M.NOM.SG

новая novaja

neW-F.NOM.SG

зубная

zubnaja

dental-F.NOM.sG

новая novaja

neW-F.NOM.SG

новый novyj

new-M.NOM.SG

врач vrach

doctor.M.NOM.SG

зубной zubnoj

dental-M.NOM.sG

врач vrach

doctor. M.NOM.SG

пришла

prishla

arrived-F.SG

c. feminine attributive and predicate agreement

эта новая зубной

eta novaja zubnoj

this-F.NOM.sG new-F.NOM.sG dental-M.NOM.sG

пришла

prishla

arrived-F.SG

d. feminine attributive and predicate agreement

врач vrach

doctor.M.NOM.SG

врач vrach

doctor.M.NOM.SG

врач vrach

doctor.M.NOM.SG

зубная

zubnaja

dental-F.NOM.sG

пришел

prishel

arrived-M.SG

врач vrach

doctor.M.NOM.SG

2. Gender mismatch analysis

2.1. Inherent semantic feature [female]

Former analyses suggest that the observed variance results from the "feminization" at some stage of derivation, that henceforth determines the agreement pattern ofthe nominal. For instance, [Matushansky, 2013] assumes that gender switch occurs when a gender semantic feature [female] is introduced into syntactic structure. In the process of derivation grammatical features of merging nodes must either match already existing inherent gender features of the NP or be interpreted semantically and therefore introduce an inherent semantic feature [female]. The ungrammaticality of (1e) and (1f) is therefore explained by both syntactic and semantic feature clash: syntactic features do not match and the semantic feature [male] cannot override the [female] feature.

2.2. The Feminizing Head W

[Pesetsky, 2013] proposes that feminization occurs through a phonolog-ically null morpheme M ([ze]). Once the morpheme merges, the nominal triggers feminine agreement. The feminizing head demonstrates structural restrictions: it cannot be merged lower than a certain structural threshold that is above the lowest level of the nominal phrase at which adjectives with nonintersective, idiomatic or argumental interpretation are introduced. Once the morpheme merges, the nominal triggers feminine agreement.

[Lyutikova, 2015] develops the idea of the feminizing head and more precisely describes its possible positions inside the syntactic structure of Russian DPs. According to the analysis, M should be introduced not only above the level of lexical noun, but also above subsective adjectives and above cardinal and collective numerals.

2.3. DP as a referential semantics domain

Note that Pesetsky in his analysis suggests the structure from (2a) with two high adjectives only one of which demonstrates referential gender value to be marginally possible. The ill-formedness of (2b) is explained by the fact that the lower constituent already demonstrates female agreement which means that the feminizing head has already merged.

(2) a. "'очень ocen' very

b. *очень ocen' very

интересная

interesn-aja

interesting-F.NOM.sG

интересный

interesn-yj

interesting-M.NOM.sG

новый nov-yj

neW-M.NOM.SG

врач vrach

dOCtOr-NOM.SG

новая врач

nov-aja vrach

neW-F.NOM.SG dOCtOr-NOM.SG

[Pesetsky, 2013, (38)]

[Pereltsvaig, 2015] considers (2a) to be ill-formed. She notes that Pesetsky's analysis is too permissive. The analysis with the feminizing head does not account for the fact that constituents of the same nature cannot have different gender value. It appears that for some reason the feminizing head cannot merge between projections of the same quality. [Lyutikova, 2015] develops a hypothesis that referential agreement correlates with the DP projection. Pereltsvaig supposes that ill-formedness of (2a) can be explained by gender switch occurring in the referential semantics domain, which corresponds the functional D-layer [Pereltsvaig, 2001, 2015]. If this is the case, we may expect that the pattern with a feminine demonstrative and a masculine high adjective would be as unacceptable as the pattern from (2a). To estimate which analysis is more appropriate, it is essential to find out the range of the variance and its quantitative parameters.

3. The Survey

The variability within Russian agreement patterns was last measured in [Graudina et al., 1976], where statistics for predicate and attributive agreement are provided separately. However, gender agreement is found not only with high adjectives and verbs in past tense, but also with low adjectives, demonstratives and relative pronouns. Therefore, the data is incomplete. Moreover, the preferable position for K may depend on the number and the type of adnominals. In order to determine the frequency of masculine and feminine agreement I have collected quantitative data from a survey of over 100 native speakers. The aim of the survey was to find out how native speakers would inflect various combinations of adnominals (including low and high adjectives and determiners) and a past tense verb with a noun, knowing that the noun was referring to female human. All combinations of stimuli used are listed in (3).

(3)

1. det high adj. low adj. 2. det high adj. our hard-working executive supervisor organized our hard-working supervisor organized

3. det low adj. our executive supervisor organized

4. det our supervisor organized

5. high adj. low adj. hard-working executive supervisor organized

6. high adj. 7. low adj. 8. (no adnominals) hard-working supervisor organized executive supervisor organized supervisor organized

54

det = determiner (possessive/demonstrative pronoun)

The study was conducted via Google Forms and consisted of a fill-in-the-blanks task. Respondents were asked to read a compound sentence: the first clause provided the context that explicitly indicated the gender of the human denoted by the subject in the second coordinate clause. The second clause contained the noun phrase and the verb in past tense with gaps instead of endings. Native speakers were asked to write the adnominals and the verb with the endings in the textbox so that the sentence was complete. An example of a stimulus is presented in (4); the two most frequent variants produced by respondents are presented in (5).

(4) Всю ночь Тане не удалось сомкнуть глаз: наш_ ответствен_ проект_ менеджер готовил_ презентацию рекламной кампании для радиохолдинга.

Vsju noch' Tane ne udalos' somknut'glaz: nash_ otvetstvenn_ proektn_ menedzher gotovil_ prezentaciju reklamnoj kampanii dlja radioholdinga.

'All night long Tanya didn't have a chance to get a wink of sleep: our responsible project manager was preparing a presentation of promotional campaign for the radio corporation.'

(5) a. наш ответственный проектный готовил

nash otvetstvennyj our-M responsible-M

proektnyj gotovil project-M was preparing-M

b. наш nash our-M

ответственный

otvetstvennyj

responsible-M

проектный

proektnyj

project-M

готовила

gotovila

was preparing-F

The results show that on the average in 68% of answers speakers used the adnominals in masculine while the verb was feminized (6a). Agreement with the formal features of the noun was found in 25.21% of answers (6b). Purely feminine agreement is possible only for types 1, 2, 3 and 5 of combinations (3). At least one modifier was feminine in 4.5% of answers (6c).

(6) a. наш nash our-M

готовила gotovil-a

ответственный

otvetstvenn-yj

responsible-M

презентацию prezentaciju

проектный

proektn-yj

project-M

менеджер menedzher manager.M

was preparing-F

presentation

55

(6) b. наш nash our-M прогноз prognoz forecast

финансовый finansov-yj financial-M цен на cen na prices for

аналитик analitik analyst.M нефть neft' oil

представил

predstavil

presented-M

c. ваша ученый секретарь организовала конференцию vash-a uchen-yj sekretar' organizoval-a konferenciju your-F academic-M secretary-M organized-F conference

Another remarkable result was that determiners demonstrated feminine agreement significantly more often when the high adjective was also feminine (7). This means that native speakers preferred the pattern where there was no gender mismatch between the determiner and the high adjective. The number of answers with feminine lower adjectives was non-significant.

(7) a. feminine determiner + masculine high adj.: 9 answers b. feminine determiner + feminine high adj.: 20 answers

4. The comparison

In this section I will compare the results of the conducted survey with the data from previous statistical studies. I will first consider results of the questionnaire-based research conducted by [Muchnik, 1971]. The aim of his survey was to investigate how wide gender mismatch patterns are used across native speakers of different age and social background. The questionnaire was distributed in 1963 across the country (3780 participants) and included several lexical variants of "high adjective + noun" and "noun + verb" combinations. The questions were as following: How would you say referring to a woman: "nice-M doctor" or "nice-F doctor"? As far as in this survey participants had to choose between feminine and masculine agreement in written form knowing the context, it is appropriate to compare the results with the data from the current survey with fill-in-the-gaps task.

Another source for comparison will be the frequency-based stylistic dictionary of variants [Graudina et al., 1976]. This dictionary provides results of a statistical research of the data from newspaper corpus from 1960s-70s. The results include the frequency distribution of NPs denoting female humans with masculine and feminine agreement. The frequency distribution of agreement patterns was also provided separately for attributive agreement with high adjectives and for predicate agreement.

Table 1

The comparison of the data from the three statistical studies

[Muchnik, 1971] 1960s [Graudina et al., 1976] 1960s-70s Current survey

a. новый врач novyj vrach neW-M.NOM.SG doctor.M.NüM.sG 69.9% 69.05% 92.14%

b. новая врач novaja vrach neW-F.NOM.SG doCtOr.M.NOM.SG 25.0% 30.96% 7.86%

c. врач пришел vrach prishel doctor.M.NOM.sG arrived-M.sG 38.6% 4.57% 25.21%

d. врач пришла vrach prishla doctor.M.NOM.sG arrived-F.sG 51.7% 95.43% 74.79%

e. мой врач moj vrach my-M.NOM.sG doctor.M.NOM.sG - - 91.16%

f. моя врач moja vrach my-F.NOM.sG doctor.M.NOM.sG - - 8.84%

In Table 1 we can see that data from [Muchnik, 1971] and [Graudina et al., 1976] concerning attributive agreement do not differ significantly. Statistics for predicate agreement are, on the contrary, significantly different. There may be two reasons for that: on the one hand, [Muchnik, 1971] provides data from spoken language while [Graudina et al., 1976] analyzed written discourse. On the other hand, the frequency of feminine agreement could have risen. The analysis of correlations with age from [Muchnik, 1971] provides evidence for the latter reason (Table 2). The younger respondent was the more likely it was that he/she would stick to the feminine agreement pattern. The usage of mixed agreement patterns was expanding with younger generations leaning towards speech innovations.

The comparison of data from previous and present studies shows that in 50 years the distribution of masculine and feminine agreement within verbs and high adjectives has significantly changed. For both attributives and predicates the percent of agreement with formal features of the noun has increased.

Table 2

The distribution of the agreement patterns for native speakers of different age

Born врач пришел vrach prishel doctor.M.NOM.SG arrived-M.SG врач пришла vrach prishla doctor.M.NOM.SG arrived-F.SG hesitant

before 1910 49.8% 42.2% 8.0%

1910-1919 44.1% 45.2% 10.7%

1920-1929 38.1% 51.0% 10.9%

1930-1939 36.7% 53.7% 9.6%

1940-1949 37.3% 53.1% 9.6%

First, let's try to explain this change using the idea of the feminizing head. Note that the frequency of feminine attributive agreement has significantly dropped, as demonstrated in Table 1 (a-b). We may assume that the structural threshold below which K cannot merge has moved upwards. Consequently, feminine agreement became unacceptable for high adjectives. Thus, feminine agreement is still possible for higher constituents inside the noun phrase, namely, determiners. However, our study shows that the frequency of agreement patterns for demonstratives and possessive pronouns does not differ significantly from the distribution for high adjectives (Table 1 (e-f)). Therefore, we may assume that the feminizing head is moving even higher, to a position above determiners, in other words above DP. In this case the analysis involving K becomes unjustified because there remains only one position where it may appear. However, I reject the analysis based on the structural threshold change because for high adjectives feminine agreement is not prohibited.

Let's assume now that the location of the feminizing head does not change. Then the frequency decline of feminine adjectival agreement may correlate with the general decline in choosing gender mismatch strategy. The comparison with data from [Graudina et al., 1976] concerning predicate agreement shows that the frequency of feminization is decreasing with masculine agreement becoming more frequent and increasing from 4.57% up to 25.21%.

Within Pesetsky's analysis it remains unclear why determiners and high adjectives demonstrate feminine agreement to the same extent. We have

already noticed that determiners and high adjectives rarely disagree (7). This probably means that demonstrative and possessive pronouns and high adjectives have the same effect as two high adjectives regarding referential agreement (2а). That is why the merge of Ж between two adnominal constituents (excluding low adjectives) is troublesome. Consequently, the number of possible positions for the feminizing head is decreasing to two: right above low adjectives and above DP. In other words, the gender switch occurs at the level of DP. Therefore, the idea of a feminizing head becomes superfluous. Referential agreement corresponds to the DP projection which functions as a referential semantics domain, where gender switch occurs. This idea is implied in approaches suggested by [Lyutikova, 2015] and [Pereltsvaig, 2015].

Pesetsky argues that the advantage of the feminizing head approach leans upon the idea of optionality of the head that introduces semantic features. This means that the account is purely syntactic and is not connected to the semantics of feminine gender or the morphology. The current study has shown that the phenomenon of gender mismatch in Russian correlates with the referential domain. Therefore, semantic component cannot be detached from the analysis of such phenomenon.

To sum up, we have discussed the phenomenon of gender mismatch in Russian. In this paper I provided both synchronic and diachronic estimation of gender mismatch frequency in Russian. I have also tested the approaches against experimental data.

The survey has shown that native speakers continue choosing gender mismatch strategy for adnominals in speech production. It would be interesting to find out how native speakers evaluate different agreement patterns; however, this is the matter of future research.

References

Crockett, 1976 - Crockett D. Agreement in contemporary standard Russian. Cambridge (Mass.), 1976.

Graudina et al., 1976 - Граудина Л., Ицкович В., Катлинская Л. Грамматическая правильность русской речи (опыт частотно-стилистического словаря вариантов). М., 1976. [Graudina L., Ickovich V., Katlinskaya L. Grammaticheska-ya pravil'nost' russkoy rechi (Opyt chastotno-stilisticheskogo slovarya variantov) [Grammatical correctness of Russian speech. An attempt to compiling frequency-based stylistic dictionary of variants]. Moscow, 1976.]

Lyutikova, 2015 - Lyutikova E. Features, agreement, and structure of the Russian noun phrase. Russkii yazyk v nauchnom osveshchenii. 2015. No. 30. Pp. 44-74.

Matushansky, 2013 - Matushansky O. Gender confusion. Diagnosing syntax. Cheng L., Corver N. (eds.). Oxford, 2013. Pp. 271-294.

Muchnik, 1971 - Мучник И. Грамматические категории глагола и имени в современном русском литературном языке. М., 1971. [Muchnik I. Grammati-cheskie kategorij glagola i imeni v sovremennom russkom literaturnom jazyke [Verbal and nominal categories in modern standard Russian]. Moscow, 1971.]

Pereltsvaig, 2001 - Pereltsvaig A. On the Nature of Intra-Clausal Relations: A study of copular sentences in Russian and Italian. Ph.D. dis. McGill University, 2001.

Pereltsvaig, 2015 - Pereltsvaig A. Nominalizations in Russian: argument structure, case, and the functional architecture of the noun phrase. Paper presented at the 6th Workshop on Nominalizations, 30-Jun-2015 - 01-Jul-2015, Verona, Italy.

Pesetsky, 2013 - Pesetsky D. Russian case morphology and the syntactic categories. Cambridge, 2013.

Статья поступила в редакцию 30.08.2017 The article was received on 30.08.2017

Герасимова Анастасия Алексеевна - магистрант кафедры теоретической и прикладной лингвистики филологического факультета, Московский государственный университет им. М.В. Ломоносова; лаборант-исследователь лаборатории общей семантики Института современных лингвистических исследований, Московский педагогический государственный университет

Gerasimova Anastasia А. - MA student at the Department of Theoretical and Applied Linguistics of Faculty of Philology, Lomonosov Moscow State University; Research assistant of Institute for Modern Linguistic Studies, Moscow State University of Education

E-mail: anastasiagerasimova432@gmail.com