A Linguistic Study of English Double Negation and its Realization in Arabic

The studies on multiple negation have been conducted intensively in linguistics, but very few studies have focused on multiple negation in Modern Standard Arabic (MSA). Although multiple negation appears in informal varieties, researchers find it an important topic for research. First, as linguists, we believe that all kinds of language varieties are worth studying. Second, the complexity of the structure of multiple negation raises queries about the difficulties that Arab students will encounter when they attempt to translate them from English to MSA. Our study focuses on double negation (DN) because, unlike other types of multiple negation, it yields a positive interpretation. This paper begins with a review of studies on multiple negation in English, MSA, and other languages, using a framework of generative grammar and the minimalist program. We then report on our empirical study of 60 randomly selected Arab students of English who were asked to translate 20 sentences containing multiple negation into MSA. To determine whether the intensity of their exposure to English would impact their understanding of these negative English structures, the students studying at levels two and four were selected. The students’ responses were quantitatively analysed. The results showed that MSA exhibits both DN and negative concord constructions. Moreover, the syntactic analyses of DN provided by Watanabe (2004) and Giannakidou (2000) concord with DN in MSA with some considerations. Additionally, all the sampled students had difficulty translating these types of sentences, indicating that intensity of exposure did not cause differences in performance.


INTRODUCTION
Negation is a very interesting structure that has been intensively studied by researchers, possibly due to its numerous uses, applications, and, most importantly, different syntactic representations (see among others, Alhilali, 2020). The structure of negation can be classified as single negation, negative concord (NC), and double negation (DN). NC represents the existence of more than one negative element or multiple negation; however, single negation and NC are semantically similar because both result in a negative reading of sentences. Like NC, DN represents the cooccurrence of more than one negative marker in the same sentence, and yet DN yields a positive interpretation. Zeijlstra (2004) claims that languages are classified in universal grammar into either NC or DN, and that these types do not coexist. However, the literature has shown this to be incorrect. In English, three types of negation are exhibited, namely single negation, NC, and DN. The following illustrations from Blanchette (2015) represent examples of English single negation, NC, and DN, respectively: (1) The student is not sick.
(2) John did not paint the house with no brush.
(3) Speaker A : I didn't know I wasn't supposed to tell him. Speaker B : You wasn't WASN'T supposed to tell him. 'It is not (necessarily) the case that you were not supposed to tell him.' The study of DN has been neglected because it is mostly used in informal situations in English, and only in a limited way in formal situations. The importance of DN also lies in the fact that it deals with a significant area in English grammar which may cause confusion when translating sentences containing DN.
As in English, MSA also exhibits these three types of negation. Alsamara'i (2000) discussed negation in the fourth part of his book entitled Ma'ani Alnahw (the meanings of syntax) under the subtitle Nafi a Nafi (negation of negation). He explains that constructions such as the following represent DN in MSA: Example 4a is an answer to the interrogative sentence 1150 | Studies in English Language and Education, 9(3), 1148-1169 (4b) ‫قائم؟‬ ‫محمد‬ ‫ما‬ ma Mohammed qa?em Neg Mohammed Awake Isn't Mohammed awake?
The following is an example of NC in MSA: (5) ‫بتاتا‬ ‫الطالب‬ ‫يذاكر‬ ‫لم‬ *(Lam) yu-ðaker al-ţalebu batatan Neg.past 3sm-study the-student N-word The student did not study at all Example (5) displays multiple negation in which two negative elements are used: the negative particle Lam, and the N-word batatan.
Regardless of this complexity, no study on DN in Arabic has been conducted. Thus, we suggest that this paper is the first attempt to study the DN construction in MSA, with the intention of checking whether our analysis aligns with the literature on multiple negation in general, and DN in particular. Moreover, our study also focuses on the difficulty that Iraqi Arab students of English may encounter when translating sentences with multiple negation, and the implications of these predicted difficulties.
The framework adopted in this study is Chomsky's generative grammar, with the generative syntax model and the minimalism program being used for the syntactic analysis (Chomsky, 1995). The quantitative method was used to analyse the students' translations of negative constructions from English to MSA. Accordingly, the research questions are: 1. Are there any accounts in the literature on the nature of DN in MSA? 2. Do these accounts on DN align with the nature of DN construction in MSA? 3. Can Iraqi-Arab students of English as a foreign language translate DN sentences correctly from English into MSA? 4. Does the intensity of exposure to English help these Iraqi-Arab students produce accurate translations of DN sentences from English into MSA?

Double Negation
It is clear that NC and DN are both present in English and MSA, but most of the studies in the literature highlight NC constructions and their relation to polarity items. Blanchette (2015) classified DN into two types: pragmatic double negation, and longdistance double negation. Pragmatic double negation, or 'metalinguistics' according to Horn (1989), is a structure in which one of the negatives in the statement repudiates a former statement. It has the feature of anaphoric negation. Long distance double negation, on the other hand, contains two negative elements placed so far apart from each other that an NC interpretation is impossible. The following examples from Blanchette (2015, p. 17)  In example (7), the two negatives cannot assume a negative concord relation as they are split by a phrase margin, and the sentence can only mean that John painted a house with windows.
In explaining the phenomenon of NC, Giannakidou (2000) proposed that negative concord items cannot be approved through an indicative clause boundary. This characteristic can help to distinguish between NC and DN interpretations, as proposed by Ladusaw (1992). In DN constructions, the negative elements should be counted because they will cancel each other to produce a positive interpretation. In such cases, it is assumed that instances of sentences with DN include more than one negative clause within the sentence. This proposal aligns with DN construction in MSA as will be shown later under section (2.3). The following study by Haegeman (1995) , however, shows that the distinction proposed by Giannakidou (2000) cannot be generalised. Haegeman (1995, p. 131) proposes that 'NC is achieved [in West Flemish] in specific configurations; if the configurational requirements are not met, a DN reading is obtained'. She claims that, for the negative constituents to yield NC reading with nie in West Flemish, they must scramble, and all move leftward out of VP. We can summarize the configurational constraints as follows: the negation marker en is optionally present; if several negative constituents co-occur, they must occur to the left of nie to enter into an NC relation with nie. If one of these negative constituents occurs to the right of nie, then we will get a DN reading. The following examples from Haegeman (1995, p. 132) illustrate NC and DN readings, respectively: (8) de Valère van niemand nie ketent (en)-was. That Valère of no one not contented en-was 'that Valère was not pleased with anyone' (9) da Valère nie ketent van niemand (en) -was That Valère not contented of no one en -was 'that Valère was not pleased with no one' Example (9) can be interpreted as '…that he was pleased with someone' because, according to Haegeman, the configurational requirements are not met. The negative concord element niemand does not need to move leftward out of the VP. However, Watanabe (2004) argues that negative concord items are inherently negative, i.e., they can never be neutral. Therefore, he argues, they should undergo the process of feature checking. In support of this, Watanabe argues against the other arguments available on multiple negation. In his classification of constructions with multiple negation, Watanabe refers to den Besten (1986), who claims that negative concord covers at least two subcases, i.e., negative doubling and negative spread. Negative doubling refers to cases in which a sentential negative element co-occurs with a negative concord element. Negative spread, on the other hand, refers to sentences that involve two or more negative concords. The following are examples of negative concord and negative spread, respectively, taken from West Flemish (Watanabe, 2004, pp. 559-560): (10a) da Valère niemand nie (en)-Kent that Valère nobody not NEG-know '…that Valère doesn't know anybody' (10b) dat-ter entwine an niemand niets gezeid eet that-there somebody to nobody nothing said has '… that somebody didn't tell anyone anything' The following Example (10c) is an example of double negation. It differs from 10a in word order.
(10c) da Valère nie niemand (en)-kent that Valère not nobody NEG-know '…that Valère does not know nobody' Although Watanabe's (2004) discussion of NC and multiple negation is consistent with Haegeman (1995) and Haegeman and Zanuttini (1996) in many aspects, questions remain which highlight problems in their arguments. The argument here is that if we compare 10a and 10c, niemand in 10c is negative and results in DN, but how to avoid this reading in 10a in which we get an NC reading?
Previous studies (Horn & Kato, 2000;Kato, 1985;Sohn, 1995) show that the important Japanese examples shown below are negative polarity items and not negative concord elements. They are compared with their English counterparts. Based on their assumption, example 11b is held to be ungrammatical because the negative polarity item nani-mo occurs alone in the sentence without a negation element to license its occurrence. According to analysis, this is strong proof that nanimo and similar elements are not negative and are instead polarity items; hence, the final outcome is an NC sentence. This is an attempt to explain why the Japanese never contain DN.
Another analysis is provided by Giannakidou (2000), who calls negative concord elements emphatic when they are stressed and used in elliptical answers, as an example (12): (12) Q: Ti idhes?
What saw-2SG 'What did you see? A: TI POTA 1 Nothing She argues that it is unnecessary to state that emphatics are inherently negative because the ignored part contains the negation marker that licenses and this argument is supported by Déprez (1999Déprez ( , p. 408, 2000. Giannakidou (2000) gives the following illustration: Nothing NEG saw-1SG 1 Capitalization indicates that the word is stressed to show that it is used for emphatic. Watanabe (2004), however, suggests that the idea that NC words are inherently negative provides a straightforward explanation for their use in elliptical answers to convey negation. Watanabe explains that nani-mo is inherently negative, unlike polarity items which are licensed by a negative marker. The neg-features of polarity items are uninterpretable and need to be raised to a position where these features are checked. Watanabe (2004) thus applies to Japanese the four diagnostic tests proposed by Vallduví (1994) to distinguish between concord and polarity to show that nani-mo is a negative concord and not a polarity item. These tests include: (1) negative concord items must appear in negative contexts, (2) they can appear above negation in the preverbal subject position, (3) they can co-occur with expressions like almost, and (4) they can be used as an elliptical answer. To these four, Giannakidou (2000) added a fifth test; that cannot be licensed across an indicative clause. Test four is most related to our study: NC elements can be used as an elliptical answer in MSA. He also shows that Giannakidou's analysis is weak because she argues that the affirmative open proposition idhes, saw-2SG, is the antecedent of the negative open proposition dhen idha, not saw-1SG.
However, assuming that Giannakidou's argument is weak and that negative concord words are inherently negative, this analysis still has to account for instances in Japanese and similar languages in which NC elements co-occur with negative words and yield negative concord interpretations and not double negation (cf. Hazem & Kanaan, 2020). Watanabe (2004) argues that all NC elements should be inherently negative and that NC constructions should be explained and understood through the theory of feature checking. Therefore, he adjusted Haegeman and Zanuttini's (1991) Neg-Criterion in terms of feature checking 2 . Haegeman and Zanuttini (1996) and Watanabe (2004) all agree that concord elements are inherently negative. However, unlike Watanabe, they have the following proposal: in West Flemish, Neg-head alone cannot support sentential negation. It cannot occur alone, and also the NegP can be null, or it can be realized by the clitic en. In their opinion, this indicates that Neg-head is not inherently negative. It has uninterpretable features which need checking with concord elements and hence yield negative interpretation. The following is an illustration of an ill-formed sentence followed by the grammatical one: (14a) *… da Valère dienen boek en-eet that Valère that book NEG-has '…that Valère doesn't have that book' (14b) … da Valère dienen boek nie en-eet that Valère that book not NEG-has '…that Valère doesn't have that book' However, Watanabe (2004) argues that this goes against Chomsky's (2000) agree and move approach, in which the goal must be active for Agree to occur. He postulates that this problem can be accounted for if we assume that Neg-head in Japanese and similar languages have an EPP and thus need the Spec to be filled; in this context, it is filled with a negative concord element. Watanabe argues that both negative concord elements and the head of NegP both carry interpretable features, and this accounts for elliptical answers and simple sentential negative sentences. In addition, this solves the problem of needing both the concord and head to be interpretable in languages that prohibit DN. Watanabe (2004) claims that to block double negation in languages like Japanese, concord elements are expected to be indicated by the morphophonological realization of the features that drive Agree. He argues that some uninterpretable features must be thrown into force Agree between the Neg head and the concord item. He claims that if the final interpretation of the sentence with multiple negation is DN then this indicates that negative concord elements appear without an uninterpretable focus feature. He adds that if this proposed uninterpretable focus feature has an overt realization then DN reading is not possible. However, if the language does not display focus morphology, the presence of an uninterpretable focus feature becomes optional and hence we get either reading; DN or NC. For example, in Japanese, as illustrated in example (11a) above, the negative concord word nani-mo, what-Mo, displays a focus morphology which is Mo in this example. This results in blocking any occurrence of DN constructions in Japanese. In contrast, West Flemish does not display any overt focus morphology, and thus it is assumed by Watanabe that DN reading occurs when no focus morphology is attached to negative concord elements. On the other hand, if an uninterpretable focus feature is attached to concord items, then this results in a final NC construction in which the focus element moves for feature checking and yields a negative reading. This proposal is compatible with the structure of DN in MSA with some modifications as illustrated later in this article.

Negation and DN in MSA
MSA consists of five main negative particles. The five negative markers in Standard Arabic (laa, lam, lan, laysa, and maa) can be inflected for tense and agreement. While lam and lan are inflected for tense, laysa inflects for agreement and is only compatible with the present tense, and laa inflects for neither tense nor agreement. Maa, like laa, inflects for neither tense nor agreement, but as well as negating verbs, it can also negate NP (see among others, Kamil & Haem, 2019).
DN construction in MSA is rarely introduced or discussed in grammar books. Alsamara'i (2000), among others, discussed this construction in the fourth part of his book entitled Maʕaani Alnaħw (The Meanings of Syntax), under the subtitle Nafi a Nafi (Negation of Negation). He explained that constructions such as the following represent DN in MSA: (15a) ‫قائم.‬ ‫محمد‬ ‫ما‬ ‫ما‬ maa maa Muħammad-un qaaʔim-un Neg. Neg. Muhammad-NOM awake-NOM 'Mohammed is not not awake' The above example is an answer to the following question as shown earlier: (15b) maa Muħammad-un qaaʔim-un Neg. Muhammad-NOM a wake-NOM 'Isn't Mohammed awake' The following is another illustration of DN in MSA: Thus, in example 17, the two negatives, the semantically negative verb, manʕk with the negative particle la, negate each other, resulting in a positive interpretation. To conclude, it is clear from the above discussion that DN is a universal structure and does occur in MSA.

Multiple Negation in English and MSA
Comparing DN in both English and MSA, we noticed the following. First, pragmatic negation exists in both languages, the following are examples from English and MSA respectively: (19) I didn't know I wasn't supposed to take the money.
You wasn't wasn't supposed to take the money.
(from Blanchette, 2015) Second, long distance negation as defined by Blanchette (2015) exists in both English and MSA. However, we noticed that English illustrates sentential double negation in two forms. In the first form, the sentence is complex, and the two clauses of the sentence include one negative marker each and the final interpretation of the sentence is positive. The second form is a simple negative sentence which includes two negative markers yielding a final positive interpretation, too. Interestingly, all the sentential double negative sentences presented by the Arab grammarians exemplify long distance DN of the first form. The researchers did not find a single illustration for sentential double negation in MSA in which the two negative markers co-occur in the same clause. The following are illustrations of long-distance negation from MSA and English respectively: (21) La ʔu-ridu-ha ʔn La ta-ʔti Neg 1S-want-her that Neg 3rds.f-comes I don't want that she does not come./ I don't want her not to come.
(22) John didn't paint [the house with no windows].
The following, however, is an example from English illustrating the cooccurrence of negative elements in the same clause and yielding a positive reading.
In Example (21), we have two clauses and both of them include the negative marker La. The second clause includes the complementizer ʔn, followed by the negative particle La.
In example (22), the negative elements are separated by boundaries and as claimed by Giannakidou (2000) make it impossible for them to enter into concord relation; the final interpretation is positive. Sentence number (23) though simple includes two negative elements and yields a positive interpretation.
A point to highlight in DN in MSA is that the negative marker La is always used in the second clause. However, in the first clause, negation is indicated either by one of the negative markers used in MSA or by a semantically negative verb. The following is an illustration. However, substituting the negative particle La used in the second clause with one of the other types of negative markers in MSA yielded ill-formed sentences. The following is an illustration: (28) ‫تعتذر‬ ‫لم/لن/ليس/ما‬ ‫ان‬ ‫اريدها‬ ‫ال‬ La ʔu-ridu-ha an ( * Lam/Lan/Laysa/Ma) ta-ʕtaðir Neg 1sts-want-her that (*Neg ) 3rdf-apologize As native speakers of MSA, we attempted to create a DN sentence in which a negative concord is used to check if the result would be an incorrect sentence. The following is an illustration: (29) ‫للحفلة‬ ‫شيء‬ ‫ال)‬ ‫و‬ ‫ر(‬ ِ ‫ُحض‬ ‫ي‬ ‫ال‬ ‫ان‬ ‫يوافق‬ ‫لم‬ Lam yu-wafiq ʔn La yu-hder wa-la ʃaiʔ l-il-hafla Neg 3rdsm-agree to Neg 3rdsm-bring and-Neg thing to-the-party He did not agree not to bring nothing/something to the party.
It is interesting to note that the sentence is correct and that it yields a positive reading.
Following Watanabe (2004), therefore, we have two assumptions. We assume that the negative elements participating in sentences with DN in MSA; negative particles, and negative concord words, are all inherently negative. We also assume that there is an uninterpretable focus feature attached optionally to the negative concord markers which explain the positive final reading of DN constructions in MSA. It is worth mentioning that the occurrence of negative concord elements in sentences with DN in MSA is optional as shown in (29). This distinguishes MSA from languages like West Flemish and Japanese.
However, NC constructions show a different characteristic from the description presented in the literature. Negative concord elements are never used in the construction of NC; NC structure always includes a negative particle and an N-word, as illustrated earlier in (5). Besides, Example (5) shows that the omission of the negative marker lam will render the sentence ungrammatical in sentences in which Nwords are used. In this case, Watanabe's (2004) assumption that NC structures are distinguished from DN construction by the presence or absence of an optional uninterpretable focus feature needs to be reconsidered in MSA. The assumption is applicable to other languages, and it helps in explaining how the positive reading is arrived at in constructions with multiple negation in MSA, but it cannot be used to distinguish NC constructions from DN constructions because, in NC constructions, negative particles do not co-occur with negative concord elements as in DN constructions.

METHODS
This study tested students' awareness of the structure and meaning of DN in the English language through their ability to translate accurately into their first language.

Participants
The researchers selected sixty Iraqi-Arab students in the English Department, College of Education, University of Al-Hamdaniya. Half of the students were selected randomly from study level two, and the other half were selected randomly from level four.

Instrument
The students were given an evaluation test which consisted of 20 English sentences randomly collected from several grammar books and websites. All the students were asked to translate the sentences into MSA. The sentences covered three types of negation, with fifteen representing double negation, and the remaining five divided between two sentences for negative concord and three for single negation (Appendix A). The sentences were arranged as follows. Sentences three, seven, and thirteen with single negation were distributed among the sentences on double negation to check the students' ability to translate sentences with single negation. Instances of simple sentential negation were also important for bringing the two structures, simple negation, and multiple negation, into contrast to help the students notice the differences between them. Sentences nineteen and twenty were samples of negative concord. Although this type of negation is rarely used in English, the aim of including this was to check if the use of more than two negative elements in a sentence would make the students produce a different translation and interpretation if they did not know the rules on negative concord and double negation.
Various types of negative elements were used in the sentences representing double negation. In some, the negative element 'not' occurs first attached to an auxiliary verb and the second negative element is 'no', which occurs independently before nouns or adjectives, such as: (30a) She ain't got no class.
In the second type, the first negative element is also 'not' attached to an auxiliary verb and the second negative element is a negative concord: (30b) I don't have nothing.
In the third type, the first negative element is a negative concord and the second is 'not' attached to an auxiliary verb: (30c) Nobody can't cheat me.
In the fourth type, the first negative element is a negative concord and the second another negative concord: (30d) No one will love nothing when it gets cold.
In the fifth type, the first negative element is the negative polarity item 'never' and the second a negative indefinite pronoun: (30e) Never none shall be mistress of it.
The last type contains the first negative element 'not' attached to an auxiliary verb, with the second a negative prefix, i.e., in/un: (30f) The price of the car isn't insignificant.

Data Collection
The sample was collected randomly from study levels two and level four. We assumed that level two students should have had less exposure to the grammar of English compared to level four students, which would be reflected in the performance of the students (cf. . The participants were informed that we needed them to translate the sentences into Arabic because we were doing research work. The test was demonstrated and collected on the same day for both levels. The students were expected to take the test seriously because the researcher who distributed the test was a staff member.

Data Analysis
The quantitative method was used to analyse the data. The number of correct and incorrect translations was counted first, and this is illustrated in Appendices B & C. Then the incorrect translations were classified into their types with the intention to check which of the structures included in the test were difficult to understand and why. The respondents of this study were coded as S1 for student number one, S2 for student number two, S3 for student number three, and so forth.

Results for Level Two Students
We analysed DN in MSA to see how far MSA concords with the analyses provided in the literature. It is worth repeating that very few references on DN in MSA were found, and those which include multiple negation present a plain descriptive discussion without reference to any framework. Table 1 includes a summary of the number and percentage of incorrect translations produced by the level two and level four students. The results are presented in the same table to facilitate their comparison. Generally, the level two students translated most of the DN sentences incorrectly, 354 of 450 were incorrect (79%). However, the students correctly translated 88 of 90 of the single negation sentences (98%), and there were 36 correct attempts of 60 for the negative concord sentences (60%). Significantly, not all the correct translations of sentences with double negation contained translations of the meaning. Thirty-one (66%) out of forty-seven correct translations were literal, the following sentence for example: (31) No one will love nothing when it gets cold. was translated by S24 as: la aħad sawfa yarɣab bi-la ʃaiʔ ʕindama yabrud Neg one will like for-Neg thing when 3 rd p.s.mas. gets cold 'No one will like nothing when it gets cold' The following is another illustration of the translation: (32) Nobody can't cheat me. was translated by S1 as: la aħad la yastˤa tˤi:ʕ xida:ʕi Neg body Neg 3 rd p.s.mas-can cheat-1 st P.s 'Nobody can't cheat me' We noticed that in all of the incorrectly translated sentences containing multiple negation the students applied one of the following processes: dropping the negative element 'nothing'; replacing the negative element 'nothing' with ʃaiʔ, (a) thing/something, or ʔei, any; replacing the second negative with ʔei, any, dropping the negative element 'no' or 'not'; or, dropping both the negative elements 'nothing' and 'no' or 'not' for words with a semantically negative meaning, such as 'without'. Table 2 illustrates the above processes using the six different types of negative sentences listed in the methodology section as models. The participants are selected randomly from the list of the level two students participating in the study:  Replacing 'nothing' with 'thing' 11 Nobody can't cheat me S12 Nobody can cheat me.
Dropping of the second negative 'not' S13 Nobody can beat me.
Dropping of the second negative 'not' S20 Nobody can beat me.
Dropping of the second negative 'not' 6 No one will love nothing when it gets cold.

S4
Nobody loves when it is cold.
Dropping of 'nothing'

S8
Nobody likes the thing when it is cold.
Replacing 'nothing' with 'thing' S11 Nobody will like it when it is cold.
Replacing 'nothing' with the pronoun 'it' 9 Never none shall be mistress of it.

S30
Not possible to be mistress to me.
Dropping of 'never' S24 Don't be obsessed by him.
Dropping of 'never' S10 Nobody must cling to things.

Dropping of 'never'
16 The price of the car isn't insignificant.

S28
The price of the car is not important.

Replacing
The price of the car is not important.
The price of the car is not costly.
Replacing 'insignificant' with 'costly' S13 The price of the car is not costly.

Results for Level Four Students
Moving to the level four data, these results were more or less the same as for level two; the level four students translated 353 sentences with double negation incorrectly (78%). The number of correct translations of double negation was 66 (15%), and the remaining 30 were left untranslated (7%). The remaining sentence was translated twice by the same student, both correctly and incorrectly. Checking the correct translations, 24 of the 66 correct translations were literal (36%). The following exemplifies the literal translations and translations of the understood meaning, respectively. S11's translation of sentence 11 in Table 2 is explained in (33).

| Studies in English
The sentence was translated as: Ladei kulu ʃaeiʔ have-1 st p.s every thing 'I have everything' Considering the last two sentences which instantiate negative concord, it was not expected that the level four students would make 32 correct attempts of 60; sentence 19 was translated 12 times correctly, and sentence 20 was translated 20 times correctly. The remaining attempts were either wrong or left untranslated.
It is also interesting to note that among the incorrect attempts to translate NC sentences was the attempt of some students to preserve two negative elements and drop the third one. For example, S5 literally translated sentence 19: This also exists in the data of the level two students. S1 translated sentence 19: (36) Nobody ain't doin' nothing' wrong. as La aħad la yuxtiʔ Neg body Neg 3 rd .s.mas-do wrong 'Nobody does not err' (meaning 'everybody errs') Interestingly, the incorrect translations with multiple negation produced by the level four students revealed the same processes applied by the level two students. Table 3 illustrates the processes using the six different types of negative sentences listed in the methodology section as models. The participants were selected randomly from the list of the level two students who participated in the study.  Nobody can't cheat me S17 Nobody is able to cheat me.
Dropping of 'not' S20 Nobody is able to cheat me.
Dropping of 'not'

S30
Nobody is able to cheat me.
Dropping of 'not' 6 No one will love nothing when it gets cold.

S18
Nobody will love a thing when it gets cold.
Dropping of 'nothing' S28 Nobody will like something when it becomes cold.

S4
Nobody likes something when it is cold.
Replacing 'nothing' with 'something' 9 Never none shall be mistress of it.

S24
Nobody will be able to master it.
Dropping of 'never' S12 Nobody is able to do it. Dropping of 'never' S10 Nobody adores that. Dropping of 'never' 16 The price of the car isn't insignificant.

S5
The price of the car is not a lot.

Replacing 'insignificant' with 'a lot' S8
The price of the car is not important.
The (price of the) car is, somehow, low in price.
Dropping of 'not' S16 The price of the car is nominal.
Dropping of 'not'

DISCUSSION
The results presented in the above section showed that DN can be a problematic area of English Grammar for EFL learners and translators. This may be due to three facts. First, DN is a neglected area in both English and MSA grammar, teachers and students may regard it as a structure which emphasises the negative meaning of the sentence. Second, the second negative marker emphasizes the first negative marker or the structure of multiple negation in MSA is not like the structure of multiple negation in English and some other languages.
We assume that most of the incorrect translations of sentences manifesting DN resulted from the students' attempts to produce a single negation though the positive final reading in some of the sentences was clear. Sentence number 11 is a strong proof of this assumption because the positive interpretation can be easily understood. This might be attributed to the differences in the structure of DN in MSA and English as the latter allows the co-occurrence of the two negative elements in the same clause whereas MSA as shown in section 2.3 does not allow such a construction. This characteristic in the structure of DN in MSA aligns with Giannakidou (2000) whose proposal that negative concord items cannot be approved through an indicative clause boundary implies that negative items in DN can only be approved across clause boundaries.
An interesting point to note is that in most of the cases in which a negative element is dropped or replaced, most of the students tended to drop or replace the second negative element in the structure. However, this was not the case when the students translated sentence number 9 in Table 2 and Table 3. The students in both levels dropped the first negative element 'never'. This finding is interesting because 'never' is considered an N-word in MSA which functions as a negative polarity item and not a negative particle.
Another interesting point is that the students never attempted to translate DN constructions in English as NC constructions though, as shown in section 2.3, NC in MSA allows the co-occurrence of negative elements within the same clause. They translated them as single negation. This might be accounted for if we consider Alanazi's (2013) proposal. Alanazi (2013) claims that N-words in MSA are slightly different from their counterparts in strict NC languages because they function as negative polarity items and not as negative concords. Besides, in their attempts to translate NC constructions, some of the students interestingly dropped or replaced one of the three negative elements constituting the NC sentences. In many of these cases, the students dropped the negative concord item 'nothing' or replaced it with the polarity element 'thing'. We assume that the students did so because NC constructions in MSA do not constitute a negative concord element. We assume that the replacement of the negative concord 'nothing' with the positive polarity 'thing' support Watanabe's (2004) assumption that these elements are inherently negative.
A point to highlight concerns the students' correct attempts at translating sentences with negative concord. We initially assumed that most of the students would fail to translate these sentences because the structure is informal and includes three negative elements. This could make the processing of the meaning more difficult. However, sentence 20 was correctly translated by the level two students 26 times, and 20 times by the level four students.
Unlike our expectations, the intensity of exposure did not have any positive impact on the students' performance. Most students in level two and level four translated most of the sentences with double negation incorrectly, with respectively 79% and 78% of the sentences being incorrectly translated. This is also manifested in the students' attempts at translating NC sentences; level four students produced 24 out of 60 incorrect translations, but level two participants produced 16 out of 60 incorrect translations.
What is more interesting is that all the students could have translated the sentences with double negation literally to obtain a structure closer to English if they found the structure new and difficult to understand; however, as the findings showed, 70 out of 113 correct translations were literal. Both levels of students translated some of these sentences according to their understanding of the semantic aspect of the sentence because we assume, they found the literal translation uncommon and weak.
Another point to note is the high scores in the number of correct translations of sentences 16 and 18. As shown, these sentences were the only ones with the structure 'not + aux ---un/in + adjective'. Although this result could have led to the claim that the second negative element, which occurs as a prefix attached to an adjective, was easier for the students to process, on checking the translations it appears that sentence 16, which was translated correctly 38 times, was translated 22 times literally. We assume that literal translation usually indicates that students are unsure of the meaning because they do not know the double negation rule, and thus they resort to an easier method of translation and that is the literal translation.

CONCLUSION
Generally, we conclude that DN in MSA is distinct from DN constructions presented in the literature. Moreover, the distinction between DN and NC construction in MSA is simple and straightforward because the latter never constitutes a negative concord element in its structure.
Based on the findings of the empirical study, the following conclusions can be drawn. Clearly, most of the participants did not know how to use DN, and a significant majority lacked the theoretical background on DN structures as used in English. Moreover, the large number of errors in choosing the right equivalent in MSA revealed that the students were influenced by the structure of the single negation itself. It is also pertinent to note that DN usually occurs in informal language styles, with which the students are unfamiliar as they are used to the more formal language structures in academia. The language structures in formal styles are usually straightforward, without double negation, making the content easier to understand. Perhaps the most important finding is the presence of DN in MSA in both formal and informal situations. However, this area of grammar is neglected in formal teaching, and therefore it may be concluded that although many students recognize the difference between single negation and DN in sentences, they cannot produce their exact equivalent in MSA.