ANALELE UNIVERSITATII BUCURESTI

Anul 2003

Anul 2004

Anul 2005

ANALELE UNIVERSITATII BUCURESTI
  

LIMBI ŞI LITERATURI STRĂINE

2 0 0 5

ON THE ACQUISITION OF GENDER IN L2 GERMAN

HEIDI FLAGNER, LARISA AVRAM

1.  Introduction: gender in L2

The present paper investigates the L2 acquisition of the grammatical feature gender in L2 German by speakers of Romanian as L1.

Previous studies dealing with the acquisition of gender features in L2 seem to agree that during the early stages L2 learners will resort to a default marker of gender. But the explanations which they provide are contradictory. Hawkins (2001), examining production data from adult L2 learners of French whose L1 was English, shows that his subjects (though advanced) resorted to a “default” gender on determiners, which led to overuse of one gender. Hawkins interprets the data as reflecting transfer from L1 (which lacks gender marking) and takes them as an argument that L2 learners fail to set the appropriate values to parameters in L2 (the failed functional features hypothesis). That L2 learners resort to default forms has also been reported in other L2 acquisition studies (Lardière 2000, Prévost & White 2000), but their explanation is different. L2 learners are argued to resort to default forms because they have problems mapping the abstract features and their surface manifestation. According to their study, L2 learners can acquire features which are absent in L1, which proves that gender errors, when present, cannot be a mere L1 effect.

White, Valenzuela, Kozlowska-Macgregor, Leung & Ben Ayed (2001) argue, on the basis of data coming for L2 learners of Spanish whose L1 is French or English,  that “while the representation in early stages may include only L1 features, L2 features are predicted to be acquirable” (p. 795). They notice that L2 learners of Spanish whose L1 has gender cope better with this category in L2 than those whose L1 lacks gender, but both types of learners manage to set the appropriate values in the end.

The examination of the acquisition of gender in L2 German by speakers of Romanian as L1 can provide evidence in favour of/against the claim that the representations in early stages include the values already set in L1. Both Romanian and German have gender. We then expect the category of gender to be present from the early stages of L2 German with these learners. On the view according to which the features associated with the functional categories in L1 transfer, being part of the initial state of the interlanguage development (Schwartz & Sprouse 1994), we expect Romanian learners of German to use only (or mainly) L1 values during the early stages. On the other hand, if there is no transfer in the learning of gender in L2, one should notice certain similarities and the possible differences between L1 and L2 learners in this area should be related to a different cause.

Studies of the acquisition of gender are concerned either with gender assignment or with gender agreement. This paper addresses gender assignment exclusively. Moreover, it only reports results obtained with nonsense nouns.

The paper is organized as follows. In section 2 we sketch the theoretical background concerning the category of gender. In 3 we present the experiment. Section 4 is the analysis of the results. We compare the emergence of gender in (adult) L2 German[1] to the emergence of gender in L1 German, both in monolingual and (balanced) bilingual settings. The conclusions are briefly summarized in section 5.

2.  Theoretical background: the category of gender

2.1. Gender as an inherent property of nouns

Both in German and Romanian, nouns have grammatical gender, which percolates, determiners and articles showing concord with it:

(1)         a. o fată / fata

                        a-_{fem. sg .} girl - _{fem. sg/} girl-- _{fem. sg} the-- _{fem. sg}

            b. un băiat /băiatul

                        a-_masc.sg. boy- _masc.sg/ boy- _masc.sg  the- _masc.sg

(2)     a. die Frau / eine Frau

                        the-_{fem. sg}  woman/ a-_{fem. sg} woman

                  b. der Mann/ ein Mann

                        the- _masc.sg  man / a-_masc.sg  man

Recent analyses on the structure of the functional categories of the noun agree that there is a Number Phrase realized as an independent functional category, which projects between D and N (Delfitto & Schroten 1991, Ritter 1993). However, there is no agreement with respect to gender. Some analyses treat gender as a lexical feature of the noun, part of the information stored in the lexicon entry, which appears at all the levels of syntactic representation (Harris 1991), i.e. nouns are inherently specified for gender. On such accounts, there is a non-trivial difference between gender and number. The latter is the head of a functional projection but there is no independent syntactic projection GenderP. Gender features are parasitic on another head, subject to cross-linguistic variation. In Romance languages, for example, it has been claimed that gender is parasitic on Number (Ritter 1993). Other analyses claim that Gender, like Number, is realized as an independent functional projection, GenP, which occurs in between NP and NumberP (Picallo 1991).

In the present paper we are concerned only with gender as an inherently specified feature of nouns, i.e. as an idiosyncratic property stated in the lexicon. We assume that gender and number represent different types of features, which are processed differently and which are acquired at different stages[2]. In terms of structure, we assume that only Number projects independently, with gender features being parasitic on Number (Ritter 1993, see also Avram & Coene 2003 for further evidence coming from language acquisition).

In this study we concentrate on the use of singular nouns with definite articles.

2.2. Gender cues

Gender assignment has been treated as arbitrary in many studies. In spite of its arbitrariness, various authors acknowledge the existence of associations between the phonological and the morphological shape of nouns and their gender value[3] (see, for example, Tucker et al. 1968, Tucker et al. 1969, and Karmiloff-Smith 1978 for French, Bermann 1981 for Hebrew, Macwhinney 1978, Köpcke 1982, Köpcke & Zubin 1983 for German). Such associations could represent a cue in the acquisition of gender. A brief inventory of morphological cues in German are presented in Table 1:

Table 1

Morphological cues for gender assignment in German

(from Köpcke 1982)

In German, phonological cues are relevant mainly in case assignment to monosyllabic nouns and in identifying masculine nouns. The only exception is the ending-e which signals feminine gender. Almost 90% of a corpus of 15,000 nouns ending in -e are feminine (Mills 1986:33). Phonological cues for gender in German are illustrated in Table 2:

Table 2

Phonological cues for gender assignment in German

(from Köpcke/Zubin 1983)

However, one should mention that there are cases when the gender indicated by a possible cue is not the real gender of the noun. See, for example, the following nouns which end in -e, but which are either masculine, as Junge, der Löwe, Käse, or neuter: Ende, Erbe, Interesse, Auge.

According to Wegener (2000), who analyzed the validity and the frequency of gender cues in German, gender assignment in German is arbitrary for only 1/3 of the nouns, as can be seen in Table 3:

Table 3

Validity and frequency of gender cues in German

(Wegener 2000: 513)

3. Methodology

31 Romanian learners of L2 German, age range 17-23, were tested. All the subjects were first exposed to German as L2 after the age of 16. Their proficiency level, at the time of the study, was intermediate[4]. The subjects were required to fill in (in writing) the gender value for 38 invented nouns (see the list in Table 4). The nonsense nouns were used in the experiment with the aim of testing whether L2 learners have the same intuitions with respect to gender cues as L1 speakers. For the invented nouns, our experiment represents a replica of the experiment reported in Levine (1999), which allowed us to compare our data both to results obtained in the case of native speakers as well as in the case of L2 speakers with a different L1 and a different level of proficiency (Levine’s subjects are advanced speakers of German as L2 with an L1 other than Romanian).

The subjects were asked to identify the gender of the nouns in a table, using the appropriate article: der/die/das.  The 38 nonsense nouns were all taken from Levine (1999); they were created taking into account the morphophonological properties of real nouns in German as well as possible gender cues, such as:

(i)   for masculine gender:

§     (l)er: Klasser,  Plathechsler 

§      ist: Zwigonist

§     en: Spegen, Kloden 

§     [d/t+r___ #]: Dras, Treld

§     [#___+s] : Zuß, Nals

§     [#___NASAL + k]: Zwand, Gand

(ii) for feminine gender:

§     ung:  Zählkerung

§     ität: Bellität

§     e: Alafose, Sische, Deuge

§     [#___ (k) +f/x/ç+t]: Suft

4. Results and analysis

The results for nonsense nouns are summarized in Table 4 below:

Table 4

Results for gender assignment to nonsense nouns

The quantitative analysis of the data clearly shows that the Romanian learners of German as L2 are sensitive to phonological and morphological cues for gender. In the case of Klasser, for example, 26 subjects (i.e. 83%) correctly identified -er as indicating masculine gender. In the case of Zwigonist, 25 subjects (80, 6%) assigned masculine gender to the noun, and in the case of Zahlkerung 30 subjects (96, 8%) took -ung as a signal of feminine gender. -e was correctly identified as a feminine gender signal for Alafose by 74% of the subjects, Sische by 93,5% and Deuge by 87,1%. The 21 monosyllabic nouns were preferentially identified as masculine (16 nouns). Only 5 were preferentially assigned neuter gender and no monosyllabic noun was assigned feminine gender.

The comparison of our results to the ones reported in Levine (1999) for native and near-native speakers of German led to the conclusion that both L1 and L2 learners of German seem to be sensitive to gender cues, i.e. gender assignment is not radically different in L1 and L2 and, in the case of gender assignment in L2 the properties of L1 do not seem to play an important part (at least not when L1 has gender). Table 5 briefly summarizes the results for some of the nonsense nouns for the three categories of learners of German:

Table 5

The data indicate that L1 and L2 learners of German with different L1 are sensitive to the same gender signals. There are, however, certain small differences. As can be seen in Table 5, -en seems to be a clearer gender signal with native and near-native speakers of German. But the Romanian learners of German as L2 are more sensitive than the others with respect to other clues. In the case of monosyllabic nouns, Levine’s natives preferentially assigned masculine gender to 13 nouns out of 21, and the near-natives to only 11, whereas the Romanian learners assigned masculine gender to 16 nouns.

One more significant difference is related to the uniformity of gender assignment within the three groups. The native speakers provide more “unanimous” answers, with the intermediate learners at the other end of the scale. The native speakers assigned the same gender in 80% of the cases to 11 nouns (i.e. 27.5%), the near-natives agreed 80% only in the case of 9 nouns (22.2%) and the intermediate L2 learners agreed 80% only with respect to 18% of the nouns (i.e. 7 nouns out of 38). The analysis of these 7 nouns (Alafose, Sische, Deuge, Zahlkerung, Plathechsler, Klasser, Zwigonist) shows that they are all associated with the signals: -e, -er (-ler), -ist and -ung. The phonological cue -e is also important in the acquisition of German as L1, being the first gender signal which children seem to identify (Mills 1986: 62, Müller 1994: 72, Wegener 2000: 534-5), and also in the early acquisition of German as L2 (Flagner & Avram 2004: 611). Certain signals, such as -e, -(l)er, -ist, -ung  are detected during the early stages of acquisition; they can be easily identified by L1 learners, as well as by L2 beginners and intermediates. The other signals seem to be less clear. Suft, for example, is monosyllabic and, since the majority of monosyllabic nouns are masculine, the intermediate learners assigned masculine gender to it, being insensitive to the signal [#___ f+t] for feminine gender. But both near-natives and natives are sensitive to this signal, assigning feminine gender to Suft in 76% of the cases. 

Analyzing the 23 nouns for which over 50% of our subjects had the same gender judgment, we noticed that 15 out of 23 nouns were assigned masculine gender (i.e. 65%), which suggests that the masculine is the default gender for these learners. We then compared our results to the ones in Levine’s (1999) study. Over 50 % of his native subjects assign the same gender to 36 nouns (out of 40), i.e. 90, which is the masculine in the case of 24 nouns (66,7%). Over 50% of his near-natives assign the same gender to 33 nouns out of 40 (82,5%), which is the masculine in the case of 18 nouns (54 %). The data indicate that all the learners (L1, near-natives L2 and intermediate L2 learners) choose the same default gender, the masculine. Choosing one gender as default has also been attested in the L1 acquisition of German, in both monolingual and bilingual settings (Müller 1990) as well as in the L2 acquisition of German by children (Flagner & Avram 2004). The default is, in the case of early L2 acquisition, either the masculine or the neuter. 

Summarizing our findings, we can say that both L1 and L2 learners of German have the same intuitions with respect to gender signals. They are not rote learners, but are able to detect the available gender signals in the target language. Both categories of language learners resort to a default gender. So one can safely assert that the acquisition of gender is not radically different in L1 and L2. At the same time one cannot discard the existence of certain differences. The intuitions with respect to gender assignment are more uniform in the case of L1 speakers of German. Also Romanian L2 learners seem to be more sensitive to monosyllabicity as a masculine gender signal. Such small differences also need explaining.

One possibility would be to relate these findings to the use of the masculine as simply the default gender; the fact that the default is also assigned to many monosyllabic nouns could be a mere “accident”. But this solution would also lead to treating early and adult L2 acquisition of gender as different processes since, as already mentioned, children acquiring German as L2 resort to either the masculine or the neuter as default. Even if one adopted this account, one would still have to find an explanation for why learners resort to a default gender in the first place.

Another explanation could be that a default gender is assigned because the association “signal-gender” is not well stored in memory yet. This explanation builds on the assumption that learners form associations by analogy, which they store in their associative memory (Pinker 1999/2003: 292). When the association is not well stored, the learner cannot retrieve the correct gender from the lexicon and default gender will be mistakenly assigned. Default gender assignment can be accounted for as the result of a sort of retrieval failure. The advantage of such an account is that it links gender assignment to memory. Memory storage is probabilistic and it depends on frequency of exposure to each particular idiosyncratic form or to members of a certain “association”. This will predict that gender assignment applies in a similar way with children and adults and that, at earlier developmental stages, L2 learners can assign both the correct and the default gender to one and the same noun[5] and that the default gender will be assigned more often during the earlier stages. In the case of beginners or intermediate learners the associations are not all well stored in the associative memory, which could explain the differences between the three groups of speakers compared in the present study. The immediate prediction of such a view would be that one should come across more inconsistencies (resulting in random choice) in the case of less advanced speakers. This prediction is borne out by the data. The results indicate a higher degree of uniformity in the case of native speakers and more variation in the case of intermediate learners.

A third explanation adopts the second one but also assumes the possibility of an overlapping of L1 and L2 gender signals, resulting in the reinforcements of certain signals. The monosyllabic nouns to which the Romanian subjects preferentially assigned masculine gender are Dras, Treld, Plas and Zwand. Table 6 summarizes the answers of the three groups of learners in the case of monosyllabic nouns:

Table 6

Gender assignment to monosyllabic nouns across the three groups

The masculine gender signals [CC___ C] and   [#___+C] in German may be reinforced by similar gender signals in Romanian, where [CC___ C] şi [#___+C] indicate the masculine or the neuter (i.e. masculine in the singular). In Romanian, monosyllabic nouns with the phonological structure [#____C] (tub, tun, dud, duş) and [CC___ C] (clan, drum, tren, truc) are masculine or neuter (i.e. masculine in the singular).

The strategy of gender assignment to the German monosyllabic nouns in the case of the Romanian L2 learners can be the result of the interference between the available gender signals in L1 and those in L2, which might indicate that associations which are already available for L1 can influence the creation of associations in the acquisition of L2, at least during the early stages.

5. Conclusions

The results of our study show that both L1 and L2 learners are sensitive to phonological and morphological gender signals available in the target language. When the associations are not well stored in their associative memory, they resort to a ‘default’ gender on determiners. We believe that this cannot be taken as a reflex of transfer from L1 ( as argued in Hawkins 2001, for example), because, on the one hand, both L1 and L2 learners resort to a default gender and, on the other hand, L2 learners resort to default gender irrespective of their L1. Our results confirm the conclusions of Lardière 2000, Prévost & White 2000, where it is argued that L2 learners resort to default forms because they have problems mapping the abstract features and their surface manifestation. The difference is that we argue that L1 and L2 learners encounter the same type of problem, which we take to be a failure to retrieve the correct gender from the lexicon in the absence of well stored gender information in the lexicon (either as a property of each item or as existing associations between phonological/morphological signals and gender). We assume that there is no significant difference between the acquisition process in the domain of gender between L1 and L2 acquisition. Both L1 and L2 learners resort to associative memory in order to create associations between possible gender cues and gender in the target language. The only difference is that L2 acquisition might be affected by an overlapping of gender signals resulting in the speaker’s treating one signal as stronger in L2 due to the availability of a similar signal in L1.

CATEGORIA GENULUI ÎN LIMBA GERMANĂ CA L2 

Rezumat

Studiul de faţă  reprezintă o analiză a achiziţiei genului în limba germană, ca limbă străină (L2), de către vorbitori adulţi cu limba română, ca limbă maternă (L1). Pornind de la premiza că în limba germană atribuirea genului este în strânsă legătură cu proprietăţile morfologice şi fonologice ale substantivelor, încercăm să arătăm că în procesul de învăţare a genului atât vorbitorii nativi  cât şi cei nenativi detectează aceste semnale de gen. Confirmarea acestei ipoteze pune sub semnul întrebării existenţa transferului din L1 în L2 în privinţa categoriei genului, demonstrând că în procesul de achiziţie a acestei categorii nu există diferenţe semnificative între vorbitorii nativi şi cei nenativi.

REFERENCES 

AVRAM, L., COENE, M., (2003) “From feature selection to feature percolation: the view from early clitics and determiners”, paper presented at GLOW 2003, workshop II: “Generative Approaches to Language Development Workshop”, Lund, Sweden, April 12, 2003.

BERMANN, R. A., (1981) “Regularity vs. Anomaly: The Aquisition of Hebrew Inflectional Morphology”, Journal of Child Language, 8: 265-282.

CORBETT, G., (1991) Gender, Cambridge: Cambridge University Press.

DELFITTO, D., SCHROTEN, J., (1991) “Bare plurals and the number affix in DP”, Probus, 30: 155-185.

EISENBERG, G., (1989) Grundriss der deutschen Grammatik, Stuttgart: Metzler.

FLAGNER, H., AVRAM, L., (2004) „Die Kategorie Genus im Erwerb des Deutschen als früher Zweitsprache“, Zeitschrift der Germanisten Rumäniens 2003, 608-614, Paideia, Bucureşti.

HARRIS, J., (1991) “The exponence of gender in Spanish”, Linguistic Inquiry, 22: 27-62.

HAWKINS, R., (2001) Second Language Syntax: A Generative Introduction, Oxford: Blackwell.

KARMILOFF-SMITH, A., (1978) “The Interplay between Syntax, Semantics and Phonology in Language Acquisition Processes”, in R. N. Campbell, P. T. Smith (eds.),  Recent Advances in the Psychology of Language, New York: Plenum Press, 1-23.

KÖPCKE, K.-M., (1982) Untersuchungen zum Genussystem der deutschen Gegenwartssprache, Tübingen: Narr.

KÖPCKE, K.-M. / ZUBIN, D., (1983) “Die kognitive Organisation der Genuszuweisung zu den einsilbigen Nomen der deutsche Sprache”, in Zeitschrift für germanistische Linguistik, 11, 166-182.

LARDIÈRE, D., (2000) “Mapping features to forms in second language acquisition”, in J. Archibald (ed.), Second Language Acquisition and Linguistic Theory, 102-129, Oxford: Blackwell.

LEVINE, G. S., (1999) “Gender assignment to German nonsense nouns: what does the native speaker know that the non-native speaker doesn’t?”, in A. Greenhill, H. Littlefiled, C. Iano (eds.), Proceedings of the 23^rd Annual Boston University Conference on Language Development, vol. 2, 397-406, Sommerville, Mass.: Cascadilla Press.

MACWHINNEY, B., (1978) “The acquisition of morphophonology”, in Monographs of the Society for Research in Child Development, 174, 1-123.

MILLS, A., (1986) The Acquisition of Gender: A Study of English and German, Berlin: Springer.

MÜLLER, N., (1990) “Developing two gender assignment systems simultaneously”, in J. M. Meisel (ed.), Two First Languages. Early Grammatical Development in Bilingual Children, 194-234, Dordrecht: Foris.

MÜLLER, N., (1994) “Gender and Number Agreement within DP”, in J. M. Meisel (ed.), Bilingual First Language Acquisition: French and German Grammatical Development, Amsterdam/Philadelphia: Benjamins, 53-87.

PICALLO, C., (1991) “Nominals and nominalization in Catalan”, Probus 3: 279-316.

PINKER, S., (1999/2003) Words and Rules, London: Phoenix.

PRÉVOST, P., WHITE, L., (2000) “Missing surface inflection or impairment in second language acquisition? Evidence from tense and agreement”, Second Language Research, 16: 103-133.

RITTER, E., (1993) “Where’s gender?”, Linguistic Inquiry, 24: 795-803.

SCHWARTZ, B. D., SPROUSE, R., (1994) “Word order and nominative case in nonnative l language acquisition”, in T. Hoekstra, B. Schwartz (eds.), Language Acquisition Studies in Generative Grammar, 317-368, Amsterdam: John Benjamins.

TUCKER, G. R., LAMBERT, W. E., RIGAULT, A. & SEGALOWITZ, N., (1968) “A psychological invetsigation of french speakers’ skill with grammatical gender”, Journal of Verbal Learning and Verbal Behaviour, 7, 312-316.

TUCKER, G. R., LAMBERT, W. E., RIGAULT, A., (1969) “Students’ aquisition of French Gender Distinction: A Pilot Investigation”, IRAL, vol. VII/I, 51-55.

WEGENER, H., (2000) “German gender in children’s second language acquisition”, in B. Unterbeck, M. Rissannen (eds.), Gender in Grammar and Cognition, 511-544, Berlin,/New York: De Gruyter.

WHITE, L., VALENZUELA, E., KOZLOWSKA-MACGREGOR, M., LEUNG, I., & BEN AYED, H., (2001) “The status of abstract features in interlanguage: gender and number in L2 Spanish”, in A. H.-J. Do, L. Dominguez, A. Johansen (eds.), Proceedinggs of the 25^th Annual Boston University Conference on Language Development, vol. 2: 792-802, Sommerville, Mass.: Cascadilla Press.