Navigation – Plan du site

AccueilNuméros24Spelling issues: what learner cor...

Spelling issues: what learner corpora can reveal about L2 orthography

Questions d’orthographe : ce que les corpus d’apprenant peuvent révéler sur les erreurs d’orthographe en L2
Irina Kor Chahine et Ekaterina Uetova

Résumés

L’article est consacré aux erreurs d’orthographe des apprenants du russe L2 en milieu francophone. Basée sur 1 816 erreurs d’orthographe, l’étude traite les quatre mécanismes (transposition, insertion, omission et substitution) qui entrent en jeu. L’influence des facteurs contextuels et non contextuels (cognitifs, inter- et intralinguistiques, extralinguistiques) est prise en compte pour chaque mécanisme en question. Malgré le caractère multidimensionnel des facteurs entrant en jeu, la récurrence de certaines erreurs conduit à distinguer de nettes tendances dans l’acquisition des faits linguistiques. En parallèle, l’analyse quantitative des données permet une étude des erreurs d’apprenants sur différents niveaux qui met pour la première fois en évidence les étapes dans l’acquisition des différents aspects de l’orthographe en russe L2, du niveau A1 à C1. Effectuée dans un second temps, une analyse qualitative des erreurs d’orthographe présente des données qui peuvent être utilisées pour l’élaboration d’outils pédagogiques.

Haut de page

Texte intégral

Introduction

  • 1 A useful table gathering some learner corpora is available at the website of the Centre for English (...)
  • 2 https://crow.corporaproject.org/

1Databases which collect students’ productions and are commonly called learner corpora, can be used for pedagogical purposes. In this domain, learner corpora1 –whether they are of a native L1 or a foreign language L2– are intended to enhance students’ literacy and grammatical skills (like CROW,2 for instance). In pedagogical uses of learner corpora, spelling skills are normally not subject to particular attention. However, spelling is a part of linguistic command, and studies on native L1 spelling occupy an important place in pedagogical research (see among others Lederlé 2011, Estienne 2014 for L1 French).

2The lack of attention to spelling issues in L2 learner corpora studies has to be viewed from the vantage point of second language acquisition (SLA). Despite the large bibliography devoted to the acquisition of a foreign language, few studies focus on the acquisition of spelling (see among others, Ibrahim 1978, Tesdell 1982, Luelsdorff 1986, 1991, Cook 1997, 2001, Morris 2001, Howard et al. 2012, Khansir 2013, Brosh 2015, Llombart-Huesca 2018). However, beyond its semiotic aspect, spelling itself represents a valuable material in understanding acquisitional processes of foreign languages. And in this respect, learner corpora open new perspectives for spelling studies.

  • 3 The term of heritage learners refers to speakers who use two languages at the same time, with one b (...)

3Based on Russian L2 learner corpora, our study relies on the analysis of spelling error frequencies, among both non-Russian-speaking subjects and heritage learners.3 It seeks to trace trends by conducting a cross-sectional analysis of these errors. It also classifies spelling errors according to mechanisms that come into play.

4The paper is organized as follows. Part 1 outlines Russian L2 spelling studies. Part 2 exposes the methodology and data. In Part 3 we present corpus data and observed trends in spelling error distribution. Part 4 is devoted to the description of spelling errors according to four main mechanisms (transposition, substitution, insertion, and omission). Then we present a cross-sectional overview of spelling errors in Part 5. And a conclusion and further reflections end the paper in Part 6.

1. The issue of spelling in studies on Russian L2

5The issue of spelling has always been in the margin of studies on Russian as a second language. Ever since the publication of a translated version of Russian Orthography by L.A. Češko (ed.) by the Pergamon Press in 1963, no learner textbook has been able to present contemporary Russian spelling in a comprehensive way outside Russia. But even if the language has not developed in a revolutionary way during the last sixty years, there is certainly a need for a new look at spelling rules, not to mention spelling norms that have emerged since its publication (e.g. see Hristova 2011 on Russian internet spelling).

6Nevertheless, sporadic studies exist in the linguistics literature. Thus, in 1974 Mark Elson questioned the morphological aspects of the Russian spelling system and their pedagogical value. Others focus on more narrow spelling issues, like the spelling of н / нн in Russian participles and adjectives (Benson 1973), the possible spelling of phoneme /j/ in some positions that could lead to simplification of Russian orthography (Benson 1993), the spelling and metalinguistic representation of palatalized consonants by L2 learners (Simonchyk & Darcy 2018).

7Some studies adopt very similar perspectives in regard to our study. It is the case with Ogneva 2018 which is based on 294 Russian spelling errors collected from 50 writing samples of Spanish learners aged 30 to 65 at the A2 language level. All errors are divided into six categories: insertion, omission, transposition, and substitution of letters, as well as the use of Latin and capital letters. The paper reviews the most frequent errors of Spanish learners of Russian L2. A majority of errors consists of substitution of letters. Other errors include omission of the soft sign after л /l/ and transposition of u /i/ and у /u/. Some graphemes or phonemes are particularly problematic for learners, namely я /ja/, в /v/, з /z/, and the soft sign. Vowel substitution of ы /i/ by u /i/, and the writing of б /b/ and в /v/ also represent a high rate of errors. Without getting deeply into details, the author finds the main causes of A2 learners’ errors in “phonological differences” of Spanish L1 and Russian L2, as well as in the lack of acquisition of Russian graphemes from Cyrillic alphabet (learners have “not yet become comfortable to using certain letters correctly” (Ogneva 2018: 128)). Besides these specific details on spelling errors, the study raises some questions: although the methodology is similar to the main studies on L2 spelling (Tasdell 1982, Luelsdorff 1986, Cook 1997, Morris 2001), the presented error classification lacks consistency and mixes different levels (formal operations, alphabetical substitution in Latin/Cyrillic, and spelling rules for capitalization). Moreover, there is no deep investigation of causes for each specific case of errors, only some general thoughts based on author’s experience. Finally, this is a very narrow study which focuses only on A2 learners, and the data used for the study is relatively small for identifying general trends. However, Ogneva’s study pointed out some problematic spelling issues in Russian L2 that we could compare and develop on a larger data set by conducting a cross-sectional study across all language levels. Using a quantitative analysis, we tried precisely to determine the most problematic spelling issues for our students and what factors can cause them.

2. Methodology and data

  • 4 How the language level was determined see below.

8In this article, we propose to focus on spelling errors in a written corpus made up of productions L2 by Russian language learners. We are interested in the handwritten errors produced by foreign learners of Russian. Our corpus consists of written productions by French students learning Russian, collected and annotated in the French subcorpus of the Russian Learner Corpus (http://www.web-corpora.net/​RLC/​). The French subcorpus of texts mainly consists of written student works (42 083 words; from A1 to C1 foreign language levels, as well as from native speakers4) mainly comes from the University Cote d’Azur (Nice), the University Lyon 2 and, in Paris, Sorbonne University (Paris IV) and the École Normale Supérieure (ENS, rue d’Ulm). Details concerning the corpus are reported in Table 1.

Table 1. Annotated corpus (token counts) in the corpus according to French students’ level and group

Language
background

Language level

Ratio

Number of words

Number of texts

Average number of words per text

Standard deviation in number of words per text

Foreign Learners

A1

5.74%

2 416

22

109.82

87.16

A2

27.74%

11 673

103

113.33

77.55

B1

16.24%

6 836

45

151.91

65.19

B2

6.00%

2 527

16

157.94

54.45

C1

4.13%

1 740

9

193.33

87.92

FLs Total

59.86%

25 192

195

129.19

78.53

Heritage Learners

B1

0.44%

187

3

62.33

31.48

B2

2.30%

966

6

161

58.21

C1

12.84%

5 402

34

158.88

78.75

HLs Total

15.58%

6 555

43

152.44

77.81

Native speakers

24.56%

10 336

43

240.37

166.24

TOTAL

100.00%

42 083

281

149.76

104.81

  • 5 Common European Framework of Reference for Languages, see www.coe.int.

9Two groups took part in the study: non-Russian-speaking subjects (foreign learners: FLs) at the A1-C1 level, according to the CEFR5 proficiency level, learning Russian as an L3 and often as a fourth language (L4) or fifth language (L5), and French-Russian learners (heritage learners: HLs) at the B1-C1 level. To these two groups is added a group of Russian students who attend French universities through international exchanges and who formed a control group composed of native speakers (NSs).

10Table 2 presents participants by language level and shows that the corpus is unequally distributed with predominant levels of A2 students (38.22%) and B1 students (17.80%), which represent more than half of the corpus (56.02%) and make up the bulk of students studying Russian in France. In addition, data for certain levels are relatively scarce. We are aware that the number of B1 informants in the heritage language, 1.57% (n=3), is too low for significance testing, and they do not represent a robust sample. However, using a descriptive statistical approach, the data are intended to be purely informative and allow for the facts to be observed and described. Moreover, as the percentage of errors for each level is determined by the number of errors relative to the number of words, the error ratio remains the same regardless of the size of the group.

Table 2. Number of participants according to their level and group

A1

A2

B1

B2

C1

Total students

Foreign

Learners

17

8.90%

73

38.22%

34

17.80%

11

5.76%

7

3.66%

142

74.35%

Heritage

Learners

_

_

3

1.57%

6

3.14%

17

8.90%

26

13.61%

Native Speakers

_

_

_

_

_

23

12.04%

Total

17

8.90%

73

38.22%

37

19.37%

17

8.90%

24

12.57%

191

100%

  • 6 To determine the students’ level, language instructors were guided by existing proficiency online t (...)

11Another aspect of the study concerns language testing. The students’ language proficiency level was determined by their language instructors and in accordance with the participants’ self-assessment.6 This identification method was advantageous, since the knowledge of the learners’ dominant language or L1 would help the annotator who is familiar with them and is able to guess the students’ intentions. However, we are aware that this could also be seen as a flaw in the annotation process, since objectivity is lost.

12Once the data was collected and ordered by language proficiency level, the second step was to annotate it. The text annotations were carried out by at least two annotators. The annotation process was organized as follows: the markup expert was responsible for detecting and marking errors in a raw corpus, the second one (if there was one) completed the annotations, and then the referring annotators (authors of the paper) checked and corrected the annotations if necessary.7

13About other particularities of the corpus and our annotation process see also Kor Chahine & Uetova 2021: 47-48 sqq.

3. Distribution of spelling errors: observed trends

  • 8 A sample of working tables is available at https://russianwheel.univ-cotedazur.fr/research.html.
  • 9 The multi-L1 corpus means that students have various L1s but have spent several years in French env (...)

14Our study is based on the analysis of 1816 spelling errors8 of Russian learners distributed by levels and student groups. The spelling errors come from the multi-lingual L1 corpus.9 The obtained quantitative data clearly show that the rate of spelling errors decreases and, therefore, the difficulties posed by this question gradually disappear and are proportional to the language level.

  • 10 Spelling level ratios on the y-axis are a ratio of the errors to the total number of words within e (...)

15Chart 110 shows that the frequency of spelling errors among foreign students (FLs) at the A1 level is quite moderate at the beginning of learning, and that these errors decrease considerably from the B1 intermediate level, the “starting” level for any HLs proficient in Russian as a heritage language. At the B1 level, the error ratio of these heritage learners (HLs) largely exceeds the number of errors of foreign subjects. This fact may at first seem quite remarkable. However, the very high number of errors made by HLs at the beginning of learning Russian (B1) can be explained by the fact that HLs mainly have a mastery of oral Russian that has been acquired in the family environment and the fact that their writing skills are usually very low.

Chart 1. Spelling Errors by Level and Group: FLs and HLs Compared to Native Speakers

Chart 1. Spelling Errors by Level and Group: FLs and HLs Compared to Native Speakers

16Chart 1 also shows a remarkable progression in spelling mastery, as this sample of HLs notably reduces the gap in the number of errors made at the next language level (B2). As a result, there is almost the same rate of errors at the end of learning (C1 level) for both HLs and FLs, even if the latter are still very competitive (their error rate is lower) in spelling since FLs are more aware of the spelling question from the beginning of learning.

17While most spelling errors involve standard words of Russian, some errors –like errors in hyphenation, errors in word or non-word spacing, and errors in capitalization– represent a deviation from spelling standards as well. Having a very low frequency rate (0.29%), these errors have little impact on the semantic integrity of a word, demonstrating a lack of knowledge with respect to purely orthographic rules (in hyphenation or capitalization, for instance) or a miscomprehension of words’ boundaries (word spacing). We will not detail them in this study.

4. Mechanisms at work: corpus analysis

18Studies on L2 spelling commonly distinguish four main “mechanisms”, which are self-exclusive: these are transposition or inversion, insertion, omission, and substitution (see Tasdell 1982, Luelsdorff 1986, Cook 1997, Morris 2001). The term of “mechanism” is applied here to formal operations according to graphemes’ usage and position in the spelling of words.

4.1 Transposition

19Transposition, or inversion, is a contextual phenomenon that consists of a permutation of graphs. Therefore, transposition is of contextual origin. This phenomenon is well-known in languages where it also bears the name metathesis; it is also well known in the evolution of Slavic languages. The errors produced by this mechanism are quite frequent (n=61) and are observed at all proficiency levels, including native speakers.

20We found that the transposition phenomenon operates in two main situations: on the one hand, it is a sequence of two graphemes (in general, two consonants) and, on the other hand, a sequence of three graphemes.

  • 11 Source <ruscorpora.ru>. While the letters с and т rank 7th and 6th, respectively, in the frequency (...)

21In a sequence of two graphemes, the transposition of the consonant sequence тс used for ст in the target position is particularly noticeable in the corpus. Here, any position in a syllable can be concerned (preceded or followed by a consonant, end of the syllable): учатсвовать A2 FL, NS / участвовать ‘participate’, ретсоран B1 FL / ресторан ‘restaurant’. It is remarkable that the тс sequence, apart from in verbs (одевается ‘dresses up’), is present in the Russian lexicon only in proper nouns (Билл Гейтс ‘Bill Gates’, Массачусетс ‘Massachusetts’), in acronyms (МТС ‘MTS’) or in rare borrowings from English (дартс ‘darts’, тайтсы ‘tights’).11 Thus, this transposition clearly violates phonological laws, just like the following transformations, which are often at the end of the word: ыв / вы – глаыв A2 / главы ‘leaders’, готыв B2 / готовы ‘ready’ and especially йи / ий – Англйиского B2 HL / английского ‘English’, информацйи A2 / информаций ‘information’.

  • 12 Here, we are referring to classical etymological doublets where the pleophonic form молоко ‘milk’ i (...)

22On the other hand, we also observe transpositions that could be called more “regular” since they do not break the phonological norms of Russian. In particular, this is the case with sequences with the liquids л and p. Notably, these graphemes related to corresponding phonemes play a central role in the phonological evolution observed in Slavic languages, e.g., млеко > молоко ‘milk’, древо > дерево ‘tree’, град > город ‘town’.12 Remarkably, in both cases, in diachronic switches and in spelling errors, the sequences in question are always in second position at the attack of the syllable (at the initial) after a consonant, with which they thus form a sequence of three graphemes.

23It is also in a sequence of three graphemes that the transposition between a liquid and another consonant (labial or dental) is carried out, with the liquid often appearing in the second consonant position: встрчеи A2 / встречи ‘meetings’, деверя B1 / деревья ‘trees’, сделудет NS / следует ‘follow’.

24Transpositions in sequences of four graphs are possible but very rare, e.g., a / o in инагдо A2 / иногда ‘sometimes’; премерь B2 / премьер ‘prime’.

4.2 Insertion

25Insertion represents a mechanism that consists of the introduction of an excess item. In the case of spelling, a graphic intruder is inserted into a word. For this type of error, an insertion represents an interference of the context or involve a bundle of associations and the knowledge of the learner. Thus, in Example (1):

1. Я езджу в университете на поезде. (A2, FL) > езжу
‘I commute to university by train’

  • 13 This type of alternation (д / ж) is very common (cf. сидеть ‘to sit’ – сижу ‘I sit’). However, the (...)

26the insertion of the dental consonant д in езджу (> езжу ‘I drive’) has an intralinguistic cause since this dental is preserved in the Russian infinitive ездить ‘to drive’ (the learner summons in his or her memory) but, from there, it disappears in favor of a palatal during the conjugation of the verb in the first person (езжу ‘I drive’). Note, however, that this type of error can also be interpreted as a morphological error: the learner has not acquired the mechanisms operating in conjugation.13 On the other hand, in Example (2):

2. Я овставаюсь в университете весь день. (A2, FL) > остаюсь
‘I stay at the university all day.’

27which represents a case of a double insertion (the consonant в and the syllable ва), we are visibly dealing with a mixed cause, whose complex mechanism is revealed by this error: first, memory restores the infinitive of the verb –оставаться ‘to stay’ (intralinguistic cause)– but according to morphological mechanisms, the suffix ва must disappear in conjugation (which is not the case here, as ва is kept in овставаюсь ‘I stay’). In the second step, at the time of production and under the effect of anticipation (right-hand-side context), the consonant в appears in a close context (the gap between the error and the target is five graphemes), with the suffix ва already being summoned in memory. Thus, these decoding operations of the origins of the error highlight the complex mechanisms that intervene in language production.

28The insertion of graphs in our corpus is a moderately common phenomenon (n=166, including n=129 among FLs).

  • 14 The soft sign is also the letter that has a relatively high frequency potential in Russian (18th ou (...)
  • 15 We also note for comparison that the highest rank, 695th, is held by the тш (тш) sequence, the leas (...)

29It is quite remarkable that the “track record” of insertion is held by the soft sign, which, as will be seen, also accounts for the highest number of omissions with the same percentage of occurrences – one-third of cases (n=59, 25%).14 Insertion of the soft sign represents a typical mistake by FLs (Президенть А1 / Президент ‘President’, можеть А2 / может ‘can’, посьле В1 / после ‘after’, дольго В2 / долго ‘long’, помешають С1 / помешают ‘will interfere’). Regarding the erroneous use of the soft sign, we note that this error is not contextual: the context near the left-hand or right-hand side does not contain a soft sign. Therefore, this error is visibly influenced, in the case of insertion, by long-term visual memory. Notably, most errors in inserting the soft sign mainly affect three consonants –т, л and c– and their palatalized counterparts denoted in writing by ть, ль and сь rank 33rd, 41st and 96th, respectively,15 in the frequency of double spellings in Russian (Ljaševskaja & Šarov 2009), having the highest frequency among sequences with the soft sign. As a result, errors of this type were visibly produced as a consequence of the visual memory of the most frequent Russian words, which testifies to the importance of long-term visual memory in the learning of writing.

  • 16 The analytical principle often dominates in the face of the synthetic process in learners’ minds. T (...)
  • 17 The marking in brackets < > refers to the item-source of the contextual error.

30The insertion of и in front of a yodized vowel is also a frequent error (n=17, 10%), where и seems to serve as an analytical tool to mark the wetting of the preceding consonant16: велосипиед А1 / велосипед ‘bicycle’, мения А2 / меня ‘me’, спальние А2 / спальне ‘bedroom’, цария В1 / царя ‘tsar’. It is possible to observe in these errors a transposition of the rules of French spelling onto Russian writing, e.g., niet, pièce, indien or manière, which probably motivated the only case of an error encountered among heritage learners: маниера B2 HL / манера ‘manner’. Therefore, it is visibly a cause of interlinguistic origin influenced by the spelling rules of the original language: if in велос<и>пиед17 one can still see the influence of the context by perseverance (with и being present in the context of the left-hand side), the other examples do not support this hypothesis, as и is absent from the context (cf. спальне, меня, манера).

31Finally, yod (й) insertion errors constitute a separate group. These errors are not very frequent in our corpus (n=9, 5.4%), but they always surprise any native Russian because in most cases, they transgress language norms. The causes of the й insertion can be multiple. However, interlinguistic causes are rare: we found only one case of influence of French at an intermediate level (детайлы В1 / детали ‘details’, from the French détails [detaj]).

32To determine the exact nature of these errors, it is often necessary to go beyond the limits of the word and to use the broader context, that of the sentence. The activation of Russian vocabulary, i.e., the memory of lexicon, is observed in the example of the common confusion between an adjective such as русский ‘Russian’ and the corresponding adverb по-русски ‘in Russian’, which does not have the final yod.

4.3 Omission

33The opposite of insertion, omission is a mechanism that consists of deleting an item. In errors of this type, it is the deletion of a graph(s) in a word. The omission of graphs is a very common error that is observed at all levels. We have listed just over 500 such errors (n=523). Errors of contextual origin are very rare (only n=30, 5.8%: n=16 for FLs and n=14 for HLs): therefore, the context has very little influence on the omission of graphs.

34Most omission errors are of two natures: some errors reflect the peculiarities of the perception of Russian phonemes by foreign speakers, while other errors are related to insufficient mastery of phonological rules.

  • 18 That is, the change in their pronunciation at the end of the word or under the influence of other p (...)

35Thus, the mechanism of omission makes it possible to highlight the Russian phonemes that represent the most difficulty for FLs who took part in our study. The first source of error is the phenomenon of assimilation of not only consonants but also vowels,18 which is a phonetic phenomenon whose effects affect the perception of phonemes and, consequently, the writing process.

36Many of these types of errors relate to the omission of и and of й, often before or after и, or in the final position when it undergoes assimilation at the end of the word: участ(и)я А1 ‘participation’, Росси(й)ски(й) А1 ‘Russian’, услов(и)ях В1 ‘conditions’. However, omission may involve any letter likely to appear in the final position (see also below): кажды(й) А1 ‘each’, Росси(и) А2 ‘Russia’, жизн(и) В1 ‘life’, которы(й) В2 ‘which’, плана(х) NS ‘plans’.

37However, the phenomenon of omission of the final graphs can also be conditioned by cognitive causes, among which we note lack of attention. There is evidence that these kinds of errors are common among highly instructed students, like FLs of intermediate levels (A2-B2), advanced HLs and even among NSs. Additionally, graphs can be omitted not only at the end of the word but also at the beginning, e.g., утро(м) С1 HL ‘in the morning’, это(м) NS ‘this’; (в)стал А2 ‘got up’, (з)ащищать А2 ‘protect’, (с)воей С1 HL ‘mine’, (с)трогого NS ‘strict’. On the other hand, the omission of vowels in the middle of words is often of contextual origin: б<е>р(е)гу А2 ‘shore’, к(о)т<о>рая В1 ‘which’, г<о>р(о)да С1 ‘towns’. Notably, these are mainly the letters o and e, the two most frequent letters (ranking 1st and 2nd, respectively) in the Russian lexicon according to NKRJa data (Ljaševskaja and Šarov 2009).

38In turn, errors related to an insufficient mastery of phonological rules primarily concern the omission of the soft sign, the double consonant and a morpheme.

  • 19 Our recent experimental phonetic study confirms the idea that foreign participants often fail in or (...)

39The omission of the soft sign represents one-third of the errors of omission and mainly involves FLs. This high rate of errors of omission and insertion involving the soft sign clearly indicates that, having no phonological and, therefore, phonetic reality in itself, the writing of the soft sign is difficult to conceive when a language-source does not have the same codes. However, in the omission of the soft sign, we still observe certain trends: thus, the soft sign is often omitted in consonant sequences when the consonant л (most often) appears in the attack position (стабил(ь)ности ‘stability’, фил(ь)м ‘film’), or in regard to a “yodized consonant-vowel” sequence (я, е, ю, и, ё: сем(ь)я ‘family’, воскресен(ь)е ‘Sunday’, п(ь)ю ‘I drink’, сем(ь)и ‘families’, сем(ь)ёй ‘family’). Additionally, French learners phonetically make little distinction between the two variants (with and without the soft sign)19: the phoneme /l/ has no hard “equivalent” in French, and /j/ remains poorly audible in the analyzed syllables.

40For all learners, FLs and HLs, there are errors in the non-writing of a double consonant where it is required: дол(л)аров ‘dollars’, Рос(с)ии ‘Russia’, ак(к)уратно ‘carefully’. Having little impact on the semantic integrity of the word, as is the case as well for hyphens or capital letters, these errors remain purely typographic and are of little interest to our study.

41Finally, we also note the errors that are indicative of weaknesses in morphological acquisition. However, these errors occur only occasionally. These are often cases of omission of a morpheme that essentially characterize the writings of FLs and that simply indicate the lack of knowledge of derivational suffixes, such as the ск suffix of adjectives (Араб(ск)ые A2/ Арабские ‘Arabic’). Some examples must also be considered cases of indecision with regard to word endings (suffix and/or flexion), and therefore, it is quite common to determine the truncated words, amputated from final morphemes: свобод(ой) А1 ‘freedom’, студентк(ами) А2 ‘students’, обшежит(ие) А2 ‘dormitory’. Omissions of morphemes have not to be viewed as spelling errors.

4.4 Substitution

42The last mechanism that we discuss here and that operates in the analyzed errors is substitution. Substitution consists of replacing one item with another item. In our case, it is a question of replacing one graph with another. The causes of this error can be both contextual and non-contextual.

43When a Latin graph is substituted for a Cyrillic graph, the origin of the error is interlinguistic in nature. The most common substitution errors concern the three “pairs” u / у, b / б, e / э: обедаю с дризями A2 ‘I’m having lunch with friends’, Вилет стоит B1 ‘ticket costs’, в етом кино B1 ‘in this movie’, об екологие A2 ‘about ecology’.

  • 20 The Russian letter э is among the rarest letters in the NKRJa data: 30th out of 33, according to Lj (...)

44Substitution errors testify to learners’ cognitive abilities to adapt to a new writing system, and as our data show, this process usually takes between one and three levels (A1-B1 levels). During this period, when a learner uses Cyrillic, he or she is likely to reproduce the graphs of his or her linguistic environment (in this case, French) under the effect of what could be called “writing memory”, that is, the visual memory of written words. It is well observed in examples of written words that are close to French (cf. Вилет B1 “billet”, екологие A2 “écologie”, економическогго A2 “économique”). This mark left by “writing memory” is well illustrated by the writing of the Russian graph э /e/, which is mainly found in loan words.20

  • 21 These kinds of errors mainly concern the graph м (transcribed by the French m), the soft sign ь (of (...)

45Apart from the cases of lack of application in graph writing,21 our data show that only a few Latin letters are more particularly affected by this phenomenon. These are the letters u, b, e, c, y, p, r, and occasionally i and z (Привет Iван! / Иван А1 ‘Hello Ivan!’, Zатем / Затем A2 ‘then’).

  • 22 Some (n=49, 5.1%) of the errors in the substitution of letters are related to the association betwe (...)

46In addition to the substitution of letters involving both alphabets, the substitution of Cyrillic letters certainly appears as the most common error among both FLs and HLs (n=947, including 773 cases among FLs).22

  • 23 The remainder being represented by substitutions between consonants or between vowels and consonant (...)

47More than two-thirds of substitution errors concern vowel substitution (69.9%, n=662)23, and they are often influenced by the phenomenon of assimilation during pronunciation. Three types of substitution are very far ahead of this subgroup: the vowels a and o (a / o, o / a), then the vowels и and ы (и / ы, ы / и), and finally и and e (и / е, е / и). These five vowels alone account for 46.9% of the errors (n=445) in the substitution subgroup.

48Cases of substitution between a and o are by far the most frequent. The writing of a instead of o (a / o) occurs most often (n=103, 10.8%). This error of intralinguistic origin appears under the effect of the perception of these phonemes in oral language. According to phonetic standards, the phoneme /o/ in an unstressed position does not have a rounded pronunciation, and due to the relaxing of the lips, o tends towards [a]. The spelling error of substituting o with a is characteristic of learners of all proficiency levels, whether foreign or heritage. Interestingly, this type of substitution occurs most often in the stressed pre-position: this is systematic in HLs and occasional in NSs writing (HLs: кавото В2 / кого́-то ‘someone’, откравенным С1 / открове́нным ‘frank’, Ривалюции NS / Револю́ции ‘revolutions’) and very common (2/3) in the writing of FLs (FLs: говарю А1 / говорю́ ‘I say’, отнашения А2 / отноше́ния ‘relationships’, подаждить B1 / подожда́ть ‘to wait’). This error is often made at the beginner level, probably because the learner does not yet have sufficient control of the tonic accent or he or she minimizes its importance. However, insufficient application in the writing of certain letters can also explain some errors.

  • 24 Incidentally, popular Russian knows similar cases of replacing a vowel a with an o in a stressed po (...)

49Similarly, writing o instead of a is also a common mistake (n=74, 7.8%). Even if the stressed pre-position still predominates, the number of errors for the stressed position is higher (1 out of 4 cases). This is also a typical error of FLs (we did not find any case of this type of error in heritage subjects): подает А1 / па́дает ‘falling’, искоть А2 / иска́ть ‘to search’. This finding may partly be explained by cases of “hypercorrection”, in which a student self-corrects by voluntarily writing o24 since he or she knows that this letter is often “hidden” behind [a] and its identification is not always obvious without sufficient knowledge of Russian.

50However, this error can also be influenced by context (n=31, 41.9%), left-hand-side or right-hand-side: помоготь / п<о>м<о>гать ‘to help’ А2, волкоми / в<о>лками ‘wolves’ В1. More than a half of errors (58.1%) of this type (n=43) are caused by cognitive factors (hypercorrection, phonetic influence, long-term memory associations, etc.).

5. Cross-sectional overview of spelling errors

51In addition to the quantitative differences in the rate of errors made by foreign students and heritage students presented at the beginning of the article (Chart 1), the two groups of learners behave differently in relation to the different mechanisms identified here.

52In Charts 2 and 3, the analyzed data are detailed according to the CEFR proficiency level of the students (the curves), and the rate of errors per word (y-axis) are detailed according to the four analyzed spelling error categories (x-axis). Chart 2 is dedicated to FLs, and Chart 3 presents HLs data.

53Thus, HLs (Chart 3) make more errors in letter substitution but these errors tend to show the greatest regression with writing proficiency (from B1 to C1). The insertion of letters, present at the initial stage (B1), is not representative in the other levels. Additionally, the omission of letters is noteworthy from the B2 level, when students’ vocabulary is enriched with new words, and errors tend to regress from B2 to C1. Transposition errors remain unimportant for this learner group regardless of level. In their turn, foreign students (Chart 2) make the most mistakes by substituting and omitting graphs. The substitution and omission of graphs will remain the most frequent spelling errors produced by FLs regardless of level. Transposition and insertion errors are much less common but still decrease over the years of learning.

Chart 2. Distribution of the Spelling Errors of FLs by CEFR Proficiency Level

Chart 2. Distribution of the Spelling Errors of FLs by CEFR Proficiency Level

Chart 3. Distribution of the Spelling Errors of HLs by CEFR Proficiency Level and of Native speakers

Chart 3. Distribution of the Spelling Errors of HLs by CEFR Proficiency Level and of Native speakers

6. Conclusion

  • 25 However, the identification of Fly! in Ну! could also be conditioned by the extralinguistic context (...)

54Through the four mechanisms that operate in the errors of learners of Russian in a French-speaking environment (transposition or inversion, insertion, omission and substitution), it appears that the identified errors were motivated either by the close linguistic context or by various noncontextual factors. Among these factors it is worth mentioning the importance of long-term memory (in particular, word memory and visual memory of writing at both the interlinguistic and intralinguistic levels) and the interference of other languages, both the L1 mother tongue and languages already learned or being acquired. The following fact recorded in our corpus testifies to the complexity of interlinguistic interference mechanisms: a student of B1 level in Russian (L4), with Italian as her mother tongue (L1) and having perfectly mastered French as her L2 (C1 level), spontaneously recognized the English word Fly! in a sequence using cursive handwriting Ну! (Ну!, see Figure 1 opposite). The Cyrillic characters written in italics were decoded here by activation of the memory of the writing system of her L3 English (B2 level), and not of her Russian L4 (B1 level).25 This fact confirms the observations already made in other studies on language interference about a predominant influence of the last acquired language (cf. De Angelis 2007, Rothman & Cabrelli 2009).

Figure 1. Picture from the Soviet cartoons “Nu, pogodi!”, episode 8 (dir. Vyacheslav Kotenochkin, 1974, USSR)

Figure 1. Picture from the Soviet cartoons “Nu, pogodi!”, episode 8 (dir. Vyacheslav Kotenochkin, 1974, USSR)

55Error decoding is a multifactorial process; thus, each error is committed under the influence of a parameter bundle. However, the recurrence of some errors leads us to think about their systemic nature. Although this study focused on francophone learners, recent investigations involving other corpora (Spanish, English, German, Finnish, Kazakh) have shown that most of the found errors were characteristic of all learners of Russian, regardless of their origin (Ogneva 2018, Uetova 2020). Nevertheless, further research is needed to confirm the universality of the spelling errors addressed in this study.

56The results of our quantitative and qualitative analysis of spelling errors of FLs and HLs represent interesting findings for research not only in acquisition and linguistics, but also in instructional design methods. The results of this study could be used for the design of specific activities.

Acknowledgements

57The authors would like to express their gratitude to the Center of Second Language Acquisition and Teaching (SLAT) at the University of Arizona (Tucson, AZ, USA), and in particular to Dr. Christine Tardy, Dr. Liudmila Klimanova, Dr. Beatrice Dupuy and Dr. Suzanne Panferov-Reese (dir.), for providing a productive environment for this research. We would also like to thank the two anonymous reviewers for their valuable comments and suggestions both on the subject as on style of the paper.

Haut de page

Bibliographie

de Angelis G. (2007). Third or Additional Language Acquisition, series “Second Language Acquisition” 24. Clevedon, Buffalo, Toronto: Multilingual Matters Ldt, 152 p.

Alonso Alonso R. (ed.). (2016). Crosslinguistic influence in Second Language Acquisition. Bristol: Multilingual Matters, 230 p.

Benson M. (1973). “The spelling of past passive participles in Russian”, The Slavic and East European Journal 17(4): 433-436.

Benson M. (1993). “A note on Russian orthography”, The Slavic and East European Journal 37(4): 530-532.

Brosh H. (2015). “Arabic Spelling: Errors, Perceptions, and Strategies”, Foreign Language Annals 48(4): 584-603.

Cook V. J. (1997). “L2 Users and English Spelling”, Journal of Multilingual and Multicultural Development 18(6): 474-488.

Cook V. J. (2001). “Knowledge of writing”, International Review of Applied Linguistics in Language Teaching 39(1): 1-18.

Češko L. A. (ed.). (1956/1963). Russian Orthography. Moscow: Učpedgiz, 1956; translated by T. J. Binyon, edited by C. V. James, 1963, Oxford/London/New York/Paris: Pergamon Press.

Elson M. J. (1975). “Morphological aspects of Russian Spelling”, The Slavic and East European Journal 19(1): 85-90.

Estienne Fr. (2014). Dysorthographie et dysgraphie: 300 exercices: comprendre, évaluer, remédier, s'entraîner, Collection “Orthophonie”. Issy-les-Moulineaux: Elsevier Masson, 182 p.

Hamers J. F. & Blanc M. H. A. (1989). Bilinguality and Bilingualism. Cambridge: Cambridge University Press.

Howard E. R. (2012). “Can yu rid guat ay rot? A Developmental Investigation of Cross-Linguistic Spelling Errors Among Spanish-English Bilingual Students”, Bilingual Research Journal 35: 164-178.

Hristova D.S. (2011). “Velikij i mogučij olbanskij jazyk: The Russian Internet and the Russian Language”, Russian Language Journal 61: 143-162.

Ibrahim M. H. (1978). “Patterns in Spelling Errors”, English language teaching journal 32(2): 207-202.

Khansir A. A. (2013). “Error Analysis and Second Language Writing”, Theory and Practice in Language Studies 3(2): 263-270.

Kor Chahine I. & Uetova E. (2021). “From Error Annotation to Quantitative Analysis: Patterns in Russian Language Learning”, Russian Language Journal 71(3): 39-70. hal-03376956.

Lederlé E. (ed.) (2011). Les troubles du langage écrit: regards croisés. Ortho Edition, 430 p.

Ljaševskaja O. & Šarov S. (2009). Častotnyj slovar’ sovremennogo russkogo jazyka (na materialax Nacional’nogo russkogo jazyka). Dictionnary of frequency of Modern Russian (based on National Corpus of Russian). Moskva: Azbukovnik. Open access: http://dict.ruslang.ru/freq.php?

Llombart-Huesca A. (2018). “Understanding the Spelling Errors of Spanish Heritage Language Learners”, Hispania 101(2): 211-223.

Luelsdorff P. (1986). Constraints on error variables in grammar bilingual misspelling orthographies. Amsterdam: J.Benjamins Pub. Co.

Luelsdorff P. (1991). Developmental orthography. Amsterdam/Philadelphia: J.Benjamins Pub. Co.

Morris L. (2001). “Going Through a Bad Spell: What the Spelling Errors of Young ESL Learners Reveal about Their Grammatical Knowledge”, The Canadian Modern Language Review 58(2): 274-286.

Ogneva A. (2018). “Spelling errors in L2 Russian: evidence from Spanish-speaking students”, Estudios interlingüísticos 6: 116-131.

Ortega L. (2020). “The study of heritage language development from a bilingualism and social justice perspective”, Language Learning 70(S1): 15-53.

Robinson P. & Ellis N. C. (eds.) (2008). Handbook of cognitive linguistics and second language acquisition. New-York & London: Routledge, 566 p.

Rothman J. & Cabrelli J. (2009). “What variables condition syntactic transfer? A look at the L3 initial state”, Second Language Research 25(4): 1-30.

Simonchyk A. & Darcy I. (2018). “The effect of orthography on the lexical encoding of palatalized consonants in L2 Russian”, Language and Speech 61(4): 522-546.

Tesdell L.S. (1982). ESL spelling errors: a taxonomy. Iowa State University. https://lib.dr.iastate.edu/rtd/7903.

Uetova E. (2020). Comparison of errors in agreement made by students of Russian as a foreign language: study based on Russian Learner Corpus. Master dissertation, HSE, Moscow.

Haut de page

Annexe

Questionnaire distributed to participants

Haut de page

Notes

1 A useful table gathering some learner corpora is available at the website of the Centre for English Corpus Linguistics: Learner Corpora around the World. Louvain-la-Neuve: Université catholique de Louvain. https://uclouvain.be/en/research-institutes/ilc/cecl/learner-corpora-around-the-world.html (last accessed 01.08.2022).

2 https://crow.corporaproject.org/

3 The term of heritage learners refers to speakers who use two languages at the same time, with one being reserved for the family environment and the other being used in a linguistic environment outside the family (work, study, social life). About heritage speakers see also Ortega 2020.

4 How the language level was determined see below.

5 Common European Framework of Reference for Languages, see www.coe.int.

6 To determine the students’ level, language instructors were guided by existing proficiency online tests (https://russianwheel.univ-cotedazur.fr/test_level.html and https://textometr.ru/) as well as their own experience of students’ achievements. The participants’ self-assessment was based on a questionnaire (see an annex).

7 A sample of annotation tags can be found on https://russianwheel.univ-cotedazur.fr/research.html# > Pilot Multi-lingual L1 French corpus > Orthographic errors.

8 A sample of working tables is available at https://russianwheel.univ-cotedazur.fr/research.html.

9 The multi-L1 corpus means that students have various L1s but have spent several years in French environment. For more detail on students’ profile see Kor Chahine & Uetova 2021.

10 Spelling level ratios on the y-axis are a ratio of the errors to the total number of words within each proficiency level. For example, there are 2 416 words and 187 errors on A1 level of foreign learners subcorpus: 187 / 2416 = 0,08.

11 Source <ruscorpora.ru>. While the letters с and т rank 7th and 6th, respectively, in the frequency table of Russian letters (Ljaševskaja & Šarov 2009), the sequences тс and rank 116th and 1st, respectively (with nearly 7.2 million absolute frequency occurrences for ст), in NKRJa. These data allow us to observe that the ст sequence typically associated with the Russian language generates the most errors among French learners. Moreover, this is a beautiful illustration of a law of ascending sonority, observed in many languages, since the fricative /s/ is less sonorous than the occlusive /t/.

12 Here, we are referring to classical etymological doublets where the pleophonic form молоко ‘milk’ is of old-Russian (OR) origin, while млеко comes from the old Slavic (OS). Historically, the difference is explained by two different ways of modifying the common Slavic *melko by following the law of the open syllable: metathesis in OS, which gives млѣко (with jat’), the addition of a vowel with apophony (e>o) in OR, and, hence, молоко. The same holds for the common Slavonic *dervo > OS древо (with jat’) + OR (without apophony) дерево ‘tree’. Cf. also common Slavic *gord > OS (with apophony) град + OR город ‘city’.

13 This type of alternation (д / ж) is very common (cf. сидеть ‘to sit’ – сижу ‘I sit’). However, the д / жд alternation is rarer: characterizing the forms of OS origin, it is currently found only in the aspectual derivation (cf. победить > побеждать) and never in the conjugation paradigm (cf. победить ‘to win’ – *побежду ‘I win’). The error in Example (1) could also come from the memory of forms of this type (with, however, the inversion of the graphs).

14 The soft sign is also the letter that has a relatively high frequency potential in Russian (18th out of 33, according to Ljaševskaja & Šarov 2009).

15 We also note for comparison that the highest rank, 695th, is held by the тш (тш) sequence, the least common occurrence in the Russian lexicon (4 342 in absolute frequency) according to NKRJa data (Ljaševskaja & Šarov 2009, http://dict.ruslang.ru/freq.php?act=show&dic=freq_2letters).

16 The analytical principle often dominates in the face of the synthetic process in learners’ minds. This has long been observed especially when there is a choice between a construction with a preposition and a casual form. The variant with a preposition always has the advantage of being more explicit (cf. дать пищу для голодных vs. дать пищу голодным ‘to give food to the hungry’).

17 The marking in brackets < > refers to the item-source of the contextual error.

18 That is, the change in their pronunciation at the end of the word or under the influence of other phonemes.

19 Our recent experimental phonetic study confirms the idea that foreign participants often fail in oral recognition of soft consonants (Del Mar Cordero Rull, Kor Chahine & Perova-Nouvelot, forthcoming).

20 The Russian letter э is among the rarest letters in the NKRJa data: 30th out of 33, according to Ljaševskaja & Šarov 2009.

21 These kinds of errors mainly concern the graph м (transcribed by the French m), the soft sign ь (often stretched to be confused with b) and the letter ш, which often lacks a final bar, graphically bringing it closer to w.

22 Some (n=49, 5.1%) of the errors in the substitution of letters are related to the association between phonemes and Russian graphs. These include cases of non-differentiation of hissing consonants (ж, ч, ш, щ) and ц, observed in the writings of both HLs and FLs: овоши А1 / овощи ‘vegetables’, женшине C1 HL / женщине ‘woman’. We also deal with substitution of ц and of щ and vice versa: заящ А1 / заяц ‘hare’, возврацаюсь А2 / возвращаюсь ‘I’m coming back’, экранизащие В1 / экранизации ‘film adaptations’. These errors point out cognitive dysfunctions in the connection between phonemes and graphs.

23 The remainder being represented by substitutions between consonants or between vowels and consonants.

24 Incidentally, popular Russian knows similar cases of replacing a vowel a with an o in a stressed position, e.g., a phrase in a famous film Тебя посо́дют, а ты не воруй! instead of the normative поса́дят. (thanks to S. Sakhno). Moreover, NKRJa lists 7 examples of use for this form (from 1899 to 2016): ― Ты с ума сошла! Меня же посодют! ― Никто тебя не посадит. Он на три дня уехал. [Андрей Клепаков. Опекун // «Волга», 2016]

25 However, the identification of Fly! in Ну! could also be conditioned by the extralinguistic context since the students had the “mission” of spotting the maximum of motion verbs in the watched video.

Haut de page

Table des illustrations

Titre Chart 1. Spelling Errors by Level and Group: FLs and HLs Compared to Native Speakers
URL http://journals.openedition.org/corpus/docannexe/image/8226/img-1.png
Fichier image/png, 13k
Titre Chart 2. Distribution of the Spelling Errors of FLs by CEFR Proficiency Level
URL http://journals.openedition.org/corpus/docannexe/image/8226/img-2.png
Fichier image/png, 19k
Titre Chart 3. Distribution of the Spelling Errors of HLs by CEFR Proficiency Level and of Native speakers
URL http://journals.openedition.org/corpus/docannexe/image/8226/img-3.png
Fichier image/png, 19k
Titre Figure 1. Picture from the Soviet cartoons “Nu, pogodi!”, episode 8 (dir. Vyacheslav Kotenochkin, 1974, USSR)
URL http://journals.openedition.org/corpus/docannexe/image/8226/img-4.png
Fichier image/png, 371k
URL http://journals.openedition.org/corpus/docannexe/image/8226/img-5.png
Fichier image/png, 349k
Haut de page

Pour citer cet article

Référence électronique

Irina Kor Chahine et Ekaterina Uetova, « Spelling issues: what learner corpora can reveal about L2 orthography  »Corpus [En ligne], 24 | 2023, mis en ligne le 15 janvier 2023, consulté le 21 juin 2024. URL : http://journals.openedition.org/corpus/8226 ; DOI : https://doi.org/10.4000/corpus.8226

Haut de page

Auteurs

Irina Kor Chahine

University Cote d’Azur, CNRS, BCL, France

Ekaterina Uetova

Technological University Dublin, Ireland

Haut de page

Droits d’auteur

Le texte et les autres éléments (illustrations, fichiers annexes importés), sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search