Students’ L1 non-standard lexical constructions as an effect of second language formal instruction

learned formally in the classroom to the native one in an environment where the latter is firmly established as the language of the country where learning takes place. In other words, the study is focused on investigating the cross-linguistic influence from English as the second language on creating non-standard lexical constructions in Polish as the mother tongue by native speakers of Polish living in Poland. The paper presents excerpts from written translation assignments of Master’s degree students of English Philology at the University of Szczecin. Despite their advanced linguistic awareness and a few years of translation instruction, students are still prone to commit transfer errors in translation from English to Polish, especially in the field of lexical semantics. Transfer-based errors include omission of a preposition, using a possessive pronoun instead of a reflexive pronoun, adding a preposition, unnatural word order, ignoring phrasemes and collocability, and wrong lexical choice in situations where polysemy occurs in English. This has implications for the methodologies of teaching English as well as Polish.


Introduction
Language transfer is caused by similarities and differences between L1 and L2 and defined as a carryover of previous performance or knowledge to subsequent learning. When the transfer is negative, it is referred to as interference 1 . In this type of transfer, the previous item is incorrectly transferred to an item to be learned. The literature devoted to the phenomenon of cross-linguistic interference in translation is predominantly concentrated on the direction from the mother tongue to the second language. The essential part of analysing language interference in translation concerns error analysis, a branch of applied linguistics. The term was introduced by Corder 2 , who claimed that the main source of errors in the second language is the native one 3 . Multiple studies demonstrated L1 interference in various language areas: lexicon 4 and syntax 5 but also pragmatics 6 and discourse 7 .
Coming down to the topic of interference in student translation, it is amply represented by Dweik 8 , who examined lexical and grammatical interference in written translation assignments from Arabic to English carried out by BA senior students specialised in translation. The results demonstrated that, with regard to written translation, the most frequent errors concern lexicon, syntax and semantics. As for lexical interference, the subjects failed to choose correct English equivalents of Arabic words. Due to insufficient knowledge of the second language, especially multiple meanings of a single word, students opted for the already known English equivalents without considering whether they were suitable in the context given. As for grammatical interference, the main problem were errors in sentence structure, since the subjects tended to translate texts word-by-word rather than adjust them to conform with the principles of English syntax. As a result, the Arabic sentence structure was transferred to English translations. Another sample study on the topic of the influence of the native language on the target L2, conducted by Bloem, Bogaard & La Heij 9 , examined semantic interference present in translation assignments of 26 Dutch-native university students highly proficient in English. The research focused on 32 highly frequent English words which were familiar to the subjects. They demonstrated semantic interference on the lexical level and a clear influence from the native language on the target one.
In an attempt to systematise transfer-based lexical errors in translation, Havlaskova 10 distinguished four subcategories of lexical interference. The first subcategory is false friends, which are the words of similar or identical form in both languages whose meanings are different so that they cannot be treated as equivalents. The next subcategory of lexical interference is the "polysemous character of a word" where "[the] choice from all the 3 Hanna Pułaczewska, "The influence of the acquisition of English as a foreign language on acceptability judgements in L1 Polish among young adults", The Language Learning Journal (2020). 4 E.g. Håkan Ringbom, "The influence of the mother tongue on the translation of lexical items", Interlanguage Studies Bulletin 3 (1978). 5 E.g. Terence Odlin, "Crosslinguistic influence", in: The Handbook of Second Language Acquisition, ed. C.J. Doughty, M.H. Long (Oxford: Blackwell, 2003 possible meanings is inappropriate" 11 . In other words, the word in the source language has multiple meanings and the translator opts for an incorrect equivalent in the target language which is not suitable in the context given. The third subcategory is the product of incapability to adjust the number of words needed to express certain concepts to the principles of the target language. Some phrases or single word expressions in the source language which demand being extended to longer phrases in the target language in order to convey the meaning are effortfully translated to the target language in the same number of words as were used in the source language. The last subcategory of lexical interference is the literal translation of idioms and collocations without taking into account the phraseology of the target language. This categorisation was then applied in an empirical study of students' translation errors from L2 English to L1 Czech. The procedure consisted of three parts: first, students' translations were analysed, then both the students and teachers were supposed to identify interferences in randomly chosen translation assignments and, finally, the students were asked to complete a questionnaire regarding their view of interference 12 . The analysis revealed that lexical and syntactic interferences were the most frequent ones.
The few studies listed above are only a tiny fracture of literature on interference occurring in translating from the mother tongue to the second language. With respect to interference from the second language to the native one, some of them demonstrated a positive impact of L2 on L1 development 13 while other focused on transferred-based errors. There is no lack of papers dealing with lexical and semantic errors in the first language that can be traced back to the second one; examples are Balcom 14 , Jarvis 15 , Laufer 16 , and Pavlenko 17 .
The fact that L1 may influence L2 simultaneously with L2 influencing L1 was the focus of the study conducted by Pavlenko and Jarvis 18 . Russian second language users of English who learned English post-puberty and had lived in the USA 3-8 years before the 11 Ibidem, 44. 12 Ibidem, 39. 13 Gregory W. Yelland, Jacinta Pollard, Anthony Mercuri, "The metalinguistic benefits of limited contact with a second language", Applied Psycholinguistics 14 (1993) experiment were asked to produce two oral narratives after watching four non-dialogue films, one in Russian and one in English. The subjects demonstrated bidirectional transfer in semantic extension, lexical borrowing and loan translation. This means that when talking in L1 they exhibited L2 influence, whereas when speaking L2, they showed L1 influence. The above-mentioned studies lead to the conclusion that language transfer is bidirectional. However, the subjects under study lived in the L2 environment which makes it difficult to pinpoint the source of interference errors. The errors can be viewed as symptomatic of L1 attrition rather than L2 influence 19 .
Against the backdrop of abundant literature on the above-mentioned types of interlingual transfer, studies focused specifically on bilinguals living in their L1 environment are but a few. For instance, Flege and Eefting 20 studied the influence of L2 English acquired through formal education on the pronunciation in L1 Dutch speakers. Liu and Ni 21 demonstrated L2 to L1 influence at the semantic level by studying the Chinese university students' interpretation of positive and negative question tags in L1 Chinese. They had L2 English as a major, and their responses revealed L2 English transfer on their judgements of L1 Chinese question tags. Also, Dong, Gui and MacWhinney 22 studied L2 to L1 influence in the area of semantics by using association tests on L1 Chinese L2 English bilinguals. The researchers wanted to investigate the shared and separate conceptual relations of translation equivalents. The study showed that in highly proficient L2 students L1 concepts assigned to lexical names are more comparable to the closest L2 equivalents than in monolinguals.
Studying the impact of L2 acquired by formal learning, Zabawa 23 aimed to investigate whether Polish advanced learners of English ("bilinguals") and Polish "monolinguals" in the same age group differed in accepting incorrect expressions. As for semantic borrowings, the "monolingual" group managed to detect errors significantly more frequently than bilinguals. Concerning "borrowings at the level of phrases" 24 , monolinguals performed better than bilinguals in noticing erroneous expressions, yet the difference was not statistically significant. Wrong syntactic constructions, which constituted the last type of errors, were detected with similar, low frequency in both groups. Another similar study was conducted by Sunde and Kristoffersen 25 who investigated acceptability judgements of native Norwegians on calques from English (L2). The subjects were bilingual young adults (i.e. learners of English) who differed in levels of English proficiency. There were also two control groups consisting of adults (mean age 55) and seniors (mean age 75) -one bilingual and one slightly bilingual. The hypothesis predicted that calques would be accepted more frequently by young adults rather than seniors and more frequently by proficient bilinguals than less proficient ones. As for the first part of the hypothesis, adults and seniors accepted calques less frequently. Yet, with respect to the second assumption, the results revealed the opposite tendency. The less proficient subject had a significantly higher acceptance rate than intermediate and proficient bilinguals. Somewhat different results were obtained by Pułaczewska 26 . 99 young adults and native speakers of Polish were asked to proofread and introduce needed corrections into a text translated from L2 English to L1 Polish. In addition, their level of English proficiency was tested through a standardised written placement test. The biggest dissimilarity among the three different proficiency groups (low, intermediate, and high) proved to be in the area of lexical semantics. The subjects with intermediate proficiency in English managed to detect erroneous expressions significantly less frequently than both the low and high proficiency groups. Another study concerning grammaticality judgements was conducted by Ewert 27 , which aimed to verify whether L2 users have poorer or better knowledge of L1 syntax than L2 learners. Another aim was to demonstrate that L2 users' knowledge of L1 differs from that of monolinguals. For this purpose, two groups (Polish-French bilinguals and Polish monolinguals), both learning English, were asked to judge the grammaticality of 14 items. Each item contained four versions of the same sentence, in which only 2 of them were correct. The results cannot lead to a conclusion that one group performed better since the differences were not statistically significant. Yet, the syntactic preferences were different between the groups. Therefore, the knowledge of L1 is different between bilinguals and monolinguals.
The aim of the paper is to provide evidence of L2 to L1 interference, develop a typology of errors and measure the frequency of lexical and syntactic errors in English to Polish translation.

Methodology
The study consisted in analysing students' written translation assignments. The task of the assignment was to translate an article written by Geoffrey K. Pullum titled The Great Eskimo Vocabulary Hoax, which consisted of 2263 words 28 . The students were asked to selfcorrect their translations on a following day before the final result (i.e., after self-correction) was submitted to the researcher.
The participants were 20 second-year students, Polish native speakers enrolled in master's degree of English Philology at the Faculty of Humanities at the University of Szczecin. They all specialised in translation, meaning that since the second year of their bachelor's degree studies they had attended courses devoted to oral and written translation of various types of texts.
The aforementioned written assignments were analysed later on in order to find any non-standard constructions that derived from L2 inference, that is, could be easily traced back to words and phrases in the English original. The word "non-standard" is used to refer to any construction that consists of word(s) existing in the Polish language but which collocate or are ordered differently and, as a result, are considered as an error. To detect any nonstandard constructions, two experts were separately asked to analyse the subjects' written assignments. The first expert was a professor of English philology specialised in linguistics, whereas the second one was a master of Polish philology, a fourth-year PhD candidate, teacher of Polish and history with 10 years' experience. Included in the following analysis are errors on which the experts' opinions coincided.

Analysis
The examination of all the written translation assignments resulted in detecting 52 nonstandard Polish constructions. The errors may be divided into five categories: lexical -polysemy and phraseology, and syntactic -omission of a preposition, addition of a preposition, substitution of a reflexive possessive pronoun with a possessive pronoun, and a wrong word order.

Lexical errors deriving from polysemy in English
As for the first category, lexical -polysemy, this group of errors derives from polysemy of an English word and a wrong choice of the Polish equivalent in a particular context. For instance, a word discretion has various meanings, including "keeping a secret" (Pol. dyskrecja) and "an ability to behave without causing embarrassing or attracting too much attention" (Pol. powściągliwość). This group constitutes 31% of all mistakes, i.e., 16 out of 52 non-standard constructions in total, as can be seen in Figure 1.

Lexical errors deriving from phraseology
The next group is related to phraseology. The subjects translated English fixed phrases or collocations word by word into Polish ignoring corresponding language-specific expressions in their native language. To illustrate, an expression fire prevention inspector should be translated as inspektor ochrony przeciwpożarowej, since it is the equivalent fixed Polish name of this occupation. The English expression is a noun phrase consisting of two premodifiers ( fire and prevention) and the head of the phrase -inspector. The Polish equivalent begins with the head -inspektorwhich is followed by post-nominal ochrona przeciwpożarowa (Eng. fire prevention) inflected in genitive. The Polish expression, the word pożarowy is an adjective deriving from a noun pożar meaning "unwanted destructive fire" and requires a prefix anti-(Pol. przeciw-). Here, there is a significant distinction between the two Polish lexemes. The word fire can be simply translated to ogień (fire1), but if it is dangerous and unwanted, the word pożar (fire2) is used. In some assignments, the English expression fire prevention inspector was translated incorrectly as *inspektor zabezpieczenia przed ogniem (inspector of prevention against fire1). The subjects were aware of the fact that in Polish the head of the phrase had to appear at the beginning, yet the postmodifiers remained nouns as in the original English expression with an addition of a preposition przed (Eng. against). This type of errors, i.e., phraseological ones, proved to be the most frequent accounting for 22 out of 52 errors, which constitutes 42% of all non-standard constructions.

Omission of a preposition or wrong preposition
As for the third category, wrong/no preposition represents all the English expressions occurring in the text which either do not contain prepositions or consist of prepositional phrases and Polish translations of these constructions either include a wrong preposition or there is no preposition even though it is required. For instance, self-regenerating myth of Eskimo snow terminology should be translated as powielający się mit o nazewnictwie śniegu u Eskimosów. In some assignments, preposition "u" (Eng. at) was omitted. This procedure would change the meaning by adding a possessive relation between Eskimos and snow, since nazewnictwo śniegu Eskimosów creates a noun phrase meaning terminology of Eskimo's snow. To avoid this, preposition "u" is needed to separate the noun phrase snow terminology and to introduce Eskimo as defining the word terminology, not snow. There were four distinct errors of this type, which constitutes ca. 8% of all types.

Addition of a preposition
With respect to adding a preposition, the situation is reverse. In other words, there are certain English expressions that collocate with a preposition, hence they were translated into Polish word by word even though they should not have been since Polish equivalent phrases do not include prepositions. To exemplify, the English phrase throw X to be eaten by polar bears should be translated into Polish using declension rzucić X na pożarcie niedźwiedziom polarnym or rzucić X niedźwiedziom polarnym na pożarcie rather than *rzucić X na pożarcie przez niedźwiedzie polarne. This example also illustrates the interplay and blurred boundaries between syntax and phraseology, in an original formulation as well as a translation task, because the phrase rzucić X-accusative Y-dative na pożarcie is a standard lexicosyntactic construction in Polish. The English expression consists of the passive voice (to be eaten) and the agent in the prepositional phrase (by polar bears) whose action causes a certain state of the object (X). In the Polish translation, the passive voice should be translated as prepositional phrase (na pożarcie) consisting of a preposition "na" and a noun form of the verb pożreć (Eng. devour). The correct translation of the remaining part of the construction should be an indirect object of the verb rzucić (Eng. throw) inflected in dative niedźwiedziom polarnym (Eng. polar bears). Since Polish is an inflecting language and its syntax provides a certain freedom in word order, the indirect object, marked as such by the case ending, may appear before or after the prepositional phrase, as shown above. While translating this construction, the subjects managed to substitute the passive form to be eaten in the English expression with the prepositional phrase na pożarcie, yet it was frequently followed by a literal translation of by polar bears (Pol. przez niedźwiedzie polarne), semantically and syntactically related to the verb pożreć (Eng. devour) like in the English original where they are related to "eat" rather than "throw". There were 3 (6%) distinct errors of this kind, that is, adding a preposition where a case ending would be used instead in Polish.

Substituting a possessive pronoun for a reflexive possessive pronoun
As for substituting a possessive pronoun for a reflexive possessive pronoun, in the Polish language, when the sentence subject and the possessor in a noun phrase referring to a direct or prepositional object is the same person, a reflexive possessive pronoun should be used, whereas in English a possessive pronoun is used no matter what connection there is between the subject and the possessor in the object phrase. For instance, the sentence She presented her research should be translated as Zaprezentowała swoje (reflexive possessive pronoun) badania not *Zaprezentowała jej (fem. sing. possessive pronoun) badania. There were 4 phrases (8%) where pronouns were translated incorrectly.

Wrong word order
The last category of syntactic errors pertains to a wrong word order which derived again from word by word translation occurred in 8 sentences. This means that a wrong word order represented 6% of all error types. To exemplify, in the original phrase It is quite obvious that in the culture of the Eskimos...snow is of great enough importance to split up the conceptual sphere that corresponds to one word and one thought in English into several distinct classes there is an interjection that corresponds to one word and one thought in English, which if placed similarly in the Polish translation of this sentence hinders its understanding. To avoid confusion, the word order should be altered to Jest dość oczywiste, że w kulturze Eskimosów śnieg ma na tyle duże znaczenie, by podzielić sferę pojęciową na kilka różnych klas, które odpowiadają jednemu słowu i jednej myśli w języku angielskim (Eng. It is quite obvious that in the culture of the Eskimos… snow is of great importance to split up the conceptual sphere into several distinct classes that/which correspond to one word and one thought in English).

Identical errors committed by different subjects -tokens
The results listed in section 3.1-3.6 concern error tokens which occurred in the whole sample. The next step in the analysis of the translated texts was to examine the frequency of particular non-standard constructions, and demonstrate patterns of exactly the same errors committed by different subjects. The total number of tokens of erroneous translation deriving from the influence from English as L2 on L1 Polish is 269, as can be seen in Table 1. Each error occurred on average in five subjects while this proportion was slightly higher for polysemy and phraseology than for the remaining types.
As for incorrect lexical choice deriving from polysemy in English, 16 distinct errors appeared repeatedly resulting in the total number of 89 tokens. The number of repetitions varied, for instance, the expression [the] section of American Anthropological Association's journal was translated incorrectly as *rozdział/sekcja czasopisma Amerykańskiego Stowarzyszenia Antropologów seven times, whereas the phrase English speakers was translated as the non-standard *mówcy (back transl. orators) języka angielskiego only twice. Another example can be word discretion in a phrase discretion for once getting the upper hand over valor, I just held my face in my hands for a minute. The word discretion can either mean the ability to keep a secret (Pol. dyskrecja) or to behave without causing embarrassment or attracting too much attention (Pol. powściągliwość). Even though the phrase corresponds with the second meaning and should be translated as powściągliwość, the word was translated as dyskrecja 6 times. With respect to the next category, incorrect lexical choice deriving from differences in phraseology, 22 non-standard expressions occurred repeatedly in translations, summing up to as many as 132 tokens. For instance, a phrase negated auxiliary verbs was translated by 8 subjects as negatywne czasowniki posiłkowe (back transl. negative), whereas it should be translated as either negujące, negacyjne or przeczące. Also, a phrase popular eagerness to embrace exotic facts [...] without seeing the evidence was translated 14 times as popularny zapał by przyjąć egzotyczne fakty […] bez zapoznania się z dowodami. Here, popularny zapał can be back translated as popular enthusiasm. The more suitable option would be tendencja (Eng. tendency) or trend. Four distinct errors concerning wrong/no preposition were committed 12 times by different subjects. As for substituting a reflexive possessive pronoun with a possessive pronoun, 4 distinct non-standard constructions appeared 15 times in the assignments of different students. Three distinct errors involving an addition of a redundant preposition were committed 10 times by different subjects. As for 3 distinct syntactic errors, different students made them in 11 tokens. In percentages, phraseological errors proved to be the most numerous, since 49% of all non-standard constructions were caused by incorrect translation of English phrases, as shown in Figure 2. Wrong lexical choice resulting from polysemy in English constituted 33% of all the errors. Four per cent of identical errors committed by different subjects consisted in omitting a preposition or using a wrong one. The substitution of a possessive pronoun with a possessive reflexive pronoun constitutes 6% of all the errors. The lowest percentage of errors is related to adding a preposition and a non-standard word-order, since it is only four per cent respectively. In sum, 82% errors may be classified as lexico-semantic and phraseological, whereas the remaining 18% are recognised as syntactic ones.

Conclusion
As the results show, the errors that appeared in the translation from English to Polish clearly demonstrate an impact of a second language on the native one. This study proves that with respect to the influence from L2 on L1, lexical and syntactic errors are acute problems. The results coincide with the findings of the studies conducted by, among others 29 . The results of this paper were obtained in the subjects L1 environment; therefore, there are no grounds for assigning the observed L2 influence as a sign of L1 attrition. The studies devoted to investigating errors in translation from L1 to L2 demonstrate high frequency of lexical and syntactic mistranslations, hence the cross-linguistic interference occurs bidirectionally and it concerns the same language areas.
With regards to lexical choice, two issues proved to be problematic. The first one is choosing a wrong Polish equivalent where the English word was polysemic, and the second is retaining English phraseology and collocations while translating into Polish. In both cases, rendering one word at a time and/or not considering multiplicity of meanings of words while translating were the source of errors and may be treated as the main cause of nonstandard lexical constructions, since subjects (students) seemed to not take into account the idiosyncrasy of their own mother tongue. This may be caused by insufficient comparative language instruction during several stages of formal education in L2. Since the current foreign language teaching approach focuses mainly on eliminating the native language in the classroom and seeking explanations of foreign words rather than their L1 equivalents, cross-linguistic interference from L2 to L1 is likely to occur. Another reason, suggested by the subjects themselves, might be the lack of formal teaching of Polish at school, where it is only taught implicitly, i.e. through absorption of contents in this language. The only explicit teaching of Polish grammar and linguistic correctness is provided in the form of the teacher's feedback from students' essays (a few per year), where language errors are marked.
Students' L1 non-standard lexical constructions as an effect of second language formal instruction S u m m a r y While a lot of research has been carried out in order to investigate cross-linguistic transfer in the direction from the native language (L1) towards the second language (L2) as well as L2 impact on L1 in childhood bilinguals (Yelland, Pollard & Mercury 1993;Cunningham & Graham 2000) and immigrants (Jarvis 2003;Pavlenko & Jarvis 2002;Laufer 2003;Pavlenko 2003), studies on the impact of foreign language learning in the classroom on L1 in L1-dominant environment are scarce. Regarding the L1 to L2 influence, studies such as Bloem, Bogaard & La Heij (2004), Havlaskova (2010), SattiHamad & Yassin (2015) and Dweik (2017) demonstrated that most frequent errors occurring in translation are lexical and/or syntactic ones. The aim of the paper is to demonstrate that the same sort of interference occurs in the direction from the second language learned formally in the classroom to the native one in an environment where the latter is firmly established as the language of the country where learning takes place. In other words, the study is focused on investigating the cross-linguistic influence from English as the second language on creating non-standard lexical constructions in Polish as the mother tongue by native speakers of Polish living in Poland. The paper presents excerpts from written translation assignments of Master's degree students of English Philology at the University of Szczecin. Despite their advanced linguistic awareness and a few years of translation instruction, students are still prone to commit transfer errors in translation from English to Polish, especially in the field of lexical semantics. Transfer-based errors include omission of a preposition, using a possessive pronoun instead of a reflexive pronoun, adding a preposition, unnatural word order, ignoring phrasemes and collocability, and wrong lexical choice in situations where polysemy occurs in English. This has implications for the methodologies of teaching English as well as Polish.