Vocabulary levels in English as a third
language
An analysis of correlations between immigrants’ Swedish L2 and
English L3 vocabulary levels
Per Andersson
2021
Student thesis, Bachelor degree, 15 HE English
Eng 802
Supervisor: Henrik Kaatari Examiner: Kavita Thomas
FACULTY OF EDUCATION AND BUSINESS STUDIES
Department of HumanitiesAbstract
In an ever-changing modern world people migrate, to a greater extent than ever before. Sweden has faced a huge influx of new immigrants in the past 10 years, many of which have had to study for many years before being able to work, including learning English as a third language.
This study looked at foreign-born adult learners (Swedish L2) of English and compared them to Swedish-born (Swedish L1) learners of English in terms of English vocabulary. This study also compared factors of age, gender, length of stay in Sweden and Swedish language level completed in Swedish L2 learners learning English; the study found that there is a small positive correlation between length of stay, younger age and completed Swedish studies with the results in the English Vocabulary Levels Test (VLT). As expected, the female students within both Swedish L1 and L2 learners outperformed the male counterparts.
The study found a significant correlation between attainment on a Swedish vocabulary test and the English vocabulary test. A learner with a higher score on the Swedish test generally performed better on the English vocabulary test, to a statistically significant degree. This, most likely, is because Swedish is typologically close to English with a high number of cognates between the languages. Surprisingly enough, the results produced pointed to very limited vocabulary learning by moving through the Swedish adult education system as the increase in both Swedish and English vocabulary was negligible.
Keywords
Contents
1. Introduction………. 1
1.1 Aim and Research Questions……….………... 2
1.2 Hypothesis………. 2
2. Literature Review……… 3
2.1 Second Language Acquisition Theory………...3
2.2 English Vocabulary Tests………. 5
2.3 The Vocabulary Levels Test………. 9
2.4 Learning an L2 and an L3………. 10
3. Method……… 12
3.1 Procedure……….………. 12
3.2 Participants………14
3.3 Data & Method of Analysis……….. 14
3.4 Validity and Reliability………. 15
4. Results………. 17
4.1 General results and by gender………... 17
4.2 Results by age and length of stay in Sweden……… 18
4.3 Results by proficiency in Swedish……… 19
5. Discussion……… 22
5.1 Conclusion………. 24 Appendix A – English vocabulary test
1
1 Introduction
The Pandemic in 2020-2021 made it clear to the world that it is no longer separated, people move about freely and cultures merge, mix and mingle. English remains a strong
world-language with approximately 1.34 billion speakers worldwide (Collins 2014). Europe has seen an influx of immigrants, unprecedented in modern days, many with limited or very limited knowledge of English. In 2016 the number of immigrants that came to Sweden reached a staggering 163,005, to drop back down to a mere 82,518 in 2020 (Statista 2021). In 2015, the number of pupils in the Swedish school system who had not been born in Sweden reached 12.5% (Regeringskansliet 2017).
Many of the immigrants arriving in Sweden as adults have limited school background and therefore have to attend school in order to be fully immersed in society. English is a language widely used in Sweden and required as a field of study throughout; some schools teach in English primarily; universities offer many courses delivered in English; and, workplaces often have international business connections. Knowledge of English is vital to be a productive member of society
As a language teacher for adults, at Komvux1, it was obvious that many
immigrants have poor English (and Swedish) vocabulary skills which in turn makes every aspect of English difficult: reading, writing, speaking and listening. Words or phrases that a Swedish-born student would be expected to understand immigrants often struggle with, even though those students are on the same course and should have similar English skills. This is intriguing and something worth researching. Whether Swedish-born learners (Swedish L1) of English as a Foreign Language had a greater vocabulary than immigrant learners (Swedish L2) of English, and if, by learning Swedish vocabulary, the learner would perform better in an English vocabulary test is the focus for this essay. Vocabulary is not explicitly taught in Swedish schools to any greater extent any more as the curriculum and language instruction is built on the communicative principle and vocabulary is the bi-product of learning a language. Therefore, the communicative approach is not in line with general language acquisition theories: “It has been acknowledged for some time that vocabulary knowledge is a good predictor of general proficiency in a foreign language” (Milton 2013, p. 67).
_________________________________________________________________
1 Komvux, Kommunal Vuxenutbildning, is the commonly used name for the municipal education system in Sweden where adults can study
2
1.1 Aim and research questions
This study will look at learners of English 5, the lowest English course at upper Secondary School (Gymnasiet), and compare Swedish-born and foreign-born learners. The aim of the study is to determine whether there is or is not a link between immigrants’ Swedish
vocabulary (L2) and their English vocabulary (L3). For the study to yield clear results the foreign-born learners participating in this study had not studied English prior to moving to Sweden. The investigation will focus on a few key questions comparing the two groups’ receptive vocabulary size in reading.
1. Are there correlations between an immigrant’s vocabulary size in Swedish and their English vocabulary size?
2. Do other variables such as age, gender and length of stay in Sweden influence the size of a learner’s vocabulary?
3. Is there a significant difference between Swedish-born learners’ and foreign-born learners’ English vocabulary?
By answering these questions this essay may provide an insight into the possible obstacles that immigrants face in terms of learning English compared to native Swedes, and possibly some ways teaching could be improved. Further research in this field could develop teaching methods to bridge the gap and give immigrants access to life and education on equal terms as Swedes born in the country.
1.2 Hypotheses
English 5 is the first-level course at Swedish High School. To be accepted into English 5, a pupil, Swedish-born and foreign-born alike, must obtain a pass grade in English from the Swedish elementary school year 9 curriculum, which precedes High School. All participants in this study must therefore have obtained a pass grade. In this study some predictions about the outcome, based on previous research, have been made:
1. Foreign-born pupils who have a greater Swedish vocabulary will have a greater English vocabulary as well.
2. Variables such as age, gender and length of stay in Sweden will have a small impact on the size of a learner’s vocabulary but not to a significant extent.
3
3. Swedish-born learners will have a significantly bigger English vocabulary than foreign-born learners.
2 Literature Review
In this section four key areas of previous research will be outlined. First, there will be an overview of theories outlining acquisition of a first and second language and how they differ and may be alike. Secondly, a review of the different types of tests will be presented. Thirdly, an in-depth presentation of the specific test relevant for this essay, VLT, and other
researchers’ results from carrying out similar tests. The final part of this section, 2.4, will show the different aspects of learning a third language, and how learners can benefit from that.
2.1 Second Language Acquisition Theory
The field of Second Language Acquisition is a complex field to study. Over time, research has shifted from believing that errors in a learner’s second language is because they are not
paying enough attention, to the belief that rote learning would eradicate any mistakes an L2 learner makes. Today, however, it is clear that how one acquires a second language is heavily influenced by the person’s first language. In this paper a learner’s first language will be called L1, a learner’s second language L2, and third language L3. Two of the influential fields of study are: the universal grammar theory and the cognitive approach to language learning. The theory of universal grammar is considered to be a prominent one for both monolingual language learning as well as bilingual language learning. It builds on the premise that, regardless of language and culture, L1 children seem to learn language at a similar rate and in a similar way; the brain seems to have a predisposed linguistic architecture already established (Mitchell et al. 2013, p. 62–97). The idea is that all languages share the same principles (not rules); for instance, that all languages have something called specifier – head - complement (principle) but different languages have different ways of ordering or specifying them (the parameters of the specific language). The opinions on the role of Universal
Grammar in L2 acquisition are more diverse; some believe that learning an L2 is not
constrained at all by the innate Universal Grammar; others believe that they are constrained by it but some features not used in the L1 may impact L2 learning, i.e. if a set of parameters is not used in the L1 it may be inaccessible in the L2. The most recent research into L2
4
Universal Grammar are available to L2 learners but certain subsets and parameters may be harder to attain as they vary from language to language (Mitchell et al. 2013, p. 86–89). Contrary to its name, Universal Grammar, it is very much a theory to keep in mind when reviewing the theory surrounding vocabulary tests in sections 2.3-2.5. Vocabulary is, by no means, purely individual words with completely different meanings. Words are made up by units, phonological and lexical features, endings etc.; rules simply put. Applying the
Universal Grammar theory to vocabulary learning could indicate that certain features of words used in one’s native language, could be easily accessed in learning another language, while other features are not. An example of the sound /b/ and /p/ for Somali speakers springs to mind. Due to their lack of a /p/ sound in their native language they cannot accurately, or without difficulty, distinguish or produce those sounds in words, even when they have learnt a language for some time (Conway 2008, pp. 20-26).
In addition to the Universal Grammar theory there is the Cognitive approach to Language Learning as well. This approach is essentially built around two opposing views; one that states that L2 learning is mostly explicit; and the other view is that L2 learning is more implicit in nature (in the same way as L1 learning). The first subset of the Cognitive Approach, implicit learning, essentially states that L1 and L2 acquisition is similar and
learning takes place through implicit processes, i.e. without conscious though. The reason that an individual does not learn his or her first and second language in the same way is explained by the fact that they are older when learning their second language. Whilst learning an L1 they are younger, have a limited view of the world, and limited cognitive thought processes. However, whilst learning an L2, they would be older and have more experience of life and a different, more mature, thought process (Mitchell et al. 2013, p. 98 -129). In the second subset of the cognitive approach, explicit learning is a driving force in L2 learning. In this theory the premise is that there are processes at work when learning an L2 that are not available to L1 learners: some memory systems and superior working memory, knowledge about language forms, and finally awareness and attention when acquiring a language. These all come from the fact that the acquisition of the L2 comes after the individual has grown cognitively and is more aware of the process of learning a new language (Mitchell et al. 2013, pp. 130–159). Both approaches have one thing in common: learning is not identical for L1 and L2.
5
2.2 English Vocabulary Tests
The English language consists of a vast number of words and phrases; Collins dictionary (2014)1 in 2014 contained 722,000 words and counting; The Swedish language is commonly
accepted to be much smaller in size, with an estimated 125,000 words in the Swedish
Academy’s Dictionaryin 2014. There may be some discrepancies between how the
dictionaries choose which words to include, so the relative sizes may differ. With that in mind, the question is how to measure vocabulary and does it have an impact on a learner’s English proficiency?
Many different ways of defining one’s vocabulary ’size’ exist; for example: receptive, productive, aural or written? Different studies performed have focussed on different aspects. For the purposes of this essay the receptive vocabulary has been selected as the area to be studied. In the receptive language acquisition field, there are three main aspects to consider, called ’dimensions’ (Gyllstad 2013, p. 16), when defining vocabulary knowledge: length, breadth and depth. They can briefly, in the above order, be described as; ’lexical fluency’, i.e. ability to use the words in context and at speed; ’vocabulary size’, meaning knowing a number of individual words outside of a context; and lastly ’word knowledge’, i.e. understanding concepts and associations to a word, quality over quantity. What dimension one chooses to study will most certainly have an impact on the results. Gyllstad (2010, p. 7-9) is of the opinion that the two aspects, breadth and depth, have been more influential and researched.
The term depth can be associated with the word quality, that an L2 learner understands all aspects of a word. An example could be a small child using the words “buy” and “sell” in the same way as an adult would use “take” and “give”; they would not
understand that the buy and sell must convey some form of payment. Gyllstad’s research found that the depth dimension can have the following three distinctions: precision of meaning, having a concise understanding of the word; comprehensive word knowledge, having phonological, morphological knowledge etc.; network knowledge, ability to link to and distinguish it from other words (Gyllstad 2013, pp. 16-18).
The term breadth refers to the number of words an individual knows, i.e. vocabulary size. The real challenge here is how to measure the number of words a learner knows. For example, call, calls, and calling could count as one word or three, depending on definition. The word called in “I called at your house yesterday” could count as the same as “I called you yesterday”, or two separate words as they carry different meaning. In tests with the
6
breadth dimension, it is vital to state clearly how the test counts words; be it word form, lemma or word family. Most tests, including the VLT, count the word once from the same word family, form or lemma. Lemma is a word that represents all other words with the same meaning but in different forms, i.e. build is the lemma form and represents builds, building, built etc. (Gyllstad 2013, pp. 19-21).
In contrast to Gyllstad above, Read & Chapelle (2001) believe the third
dimension, length, which they call lexical measures and lexicalized language is a more fitting approach to measuring vocabulary. The breadth and depth dimensions are most commonly tested through a multiple-choice test in a decontextualized way. A lexical measures test could, however, be constructed in running text with word blanks or partial deletions, i.e. the
participant is given a part of the word and must fill in the rest. Read & Chapelle explain that there is a need to develop lexical tests: “There should be a more sophisticated approach that draws on contemporary theory and practice in test validation” (Read & Chapelle 2001, p. 3). They also state that there is a shortage of useable tests that have been proven to be reliable: “What is lacking, though, is any comprehensive basis for evaluating current tests and developing new lexical measures for the future” (Read & Chapelle 2001, p. 3). It seems the lexical approach is in need of more research before it is reliable.
Language teaching and instruction have moved away from explicitly teaching vocabulary. In many countries a communicative approach prevails, where a vast vocabulary is the bi-product of learning a language. This is not in line with current research into Second Language Acquisition (Milton 2013).
In academic circles, the place of vocabulary in language learning has been significantly revised over the last decade and current academic thinking is very much at odds with much classroom and textbook practice. Far from being an element which is merely incidental to language learning, current thinking advocates that vocabulary may be crucial to the development of language performance overall /…/ The acquisition of vocabulary items in sufficient quantity triggers the setting of universal grammatical parameters (Milton 2013, p. 58)
There is a difference between productive and receptive word knowledge. Some words can be recognised if they occur in someone’s writing or speech (receptive) but a learner would not be able to use the same word in writing or speech (productive). Some words are recognized when spoken (phonological) but cannot be recognized in writing. Milton accepts that it is logical to think of language as having four skills; two receptive skills in reading and listening; and two productive skills in writing and speaking; the receptive vocabulary is greater in size. In reality, though, they are not completely separate but do, in fact, intertwine. A bigger
7
vocabulary is expected to result in better language performance as well as a better range and control of one’s vocabulary (Milton 2013).
/…/ the most important conclusion that emerges from the research is the importance of vocabulary knowledge in being able to understand and communicate in a foreign language /…/ showing a moderate to strong relationship between vocabulary measures and the ability to read, write, listen, and it seems, also speak /…/ (Milton 2013, p. 71)
He further writes: “Current methods and approaches to language teaching fail to consider how vocabulary should be systematically built into the curriculum” (Milton 2013, p. 73). It is interesting to note, although vocabulary clearly is a strong indicator of, possibly part of, a learner’s language proficiency it is not explicitly taught in many countries’ curriculums nor in the Swedish curriculum. Vocabulary in Sweden is mainly taught through a text (or a
video/audio clip) where the teacher and pupils pick out words that are found in the text. They may do some grammar work around them; using them in speech; translate them into Swedish (and possibly other languages); but the key factor is that the text (chosen) decides what words or phrases are taught rather than a vocabulary driven curriculum.
Webb & Chang carried out a study of Taiwanese learners’ English vocabulary growth in 2012 and found that the progress was not efficient and that “the results indicate that an institutional program for vocabulary learning may help to optimize learning” (Webb & Chang 2012, p. 122).
Gyllstad (2007) developed and carried out tests on Swedish advanced learners of English, called COLLEX and COLLMATCH, which tested collocations rather than singular words in a traditional vocabulary test.Collocations are words that are commonly used together to create meaning, such as make the bed or make up your mind.This type of test differs from other traditional vocabulary tests in that it shows the word in a context or sentence, rather than a singular word that should be matched with a word of similar or the same meaning. He developed and trialed several versions of the tests in his thesis, and the final versions are named COLLEX 5 and COLLMATCH 3. One of the research questions in his study was to find out whether there is a relationship between Swedish L2 learners’
vocabulary size and collocations. The results from Gyllstad’s study can be directly compared with this study as both studies use the word-levels set out in the Vocabulary Levels Test, detailed in section 2.3. The word-levels outlined in the VLT are comprised by the most commonly used words in the English language, grouped into levels with the most common words in the first tier, 2000-level.
8
In a traditional vocabulary test the word make would be tested in a manner similar to the one in table 1. The participant is expected to tick one correct answer. In the VLT the traditional test has been expanded to include several words to prevent guessing (see Appendix A for examples).
Table 1. Example question of a traditional vocabulary test. Make
Create Bird Sauce
The word make in a collocation, in Gyllstad’s test, would be tested in a manner similar to the one in table 2. The participant is expected to circle the correct phrase and tick the box if they had guessed.
Table 2. Example question of Gyllstad’s collocations test.
Set the bed Make the bed I am guessing
As part of Gyllstad’s efforts to create a collocation test and prove its validity he compared his results with more traditional tests. One interpretation of the results, Gyllstad (2007, p. 201) reports, is that traditional vocabulary tests, such as the VLT, test similar but not the same aspects as the COLLEX and COLLMATCH tests. He suggests COLLMATCH tests lexical knowledge of depth and breadth. The reason for that could be, he argues, that it takes a larger vocabulary to recognise collocations. Gyllstad carried out a battery of tests testing
collocations, breadth and width vocabulary tests and collated the results. The results showed a clear correlation that high results in the breadth and depth tests corresponded to an equally high result in the COLLEX and COLLMATCH tests. In comparing the tests, he found that
The first implication is that one could argue that COLLEX and
COLLMATCH, and WAT [word association test], are all influenced by vocabulary size to a great extent. In terms of the WAT, Wolter (2005) points to the fact that some of the words featuring in the version used in the present study are fairly low-frequency items, and that vocabulary size is therefore believed to have a considerable influence on test-takers’ performance (Gyllstad 2007, p. 212)
9
2.3 The Vocabulary Levels Test
After reviewing the research in the field of vocabulary it became clear that the length dimension was untested (Read & Chapelle 2001). This left only the breadth and depth dimensions. The most commonly used and most reliable form of test was the breadth test, used by Gyllstad (2013) Milton (2013), Webb & Chang (2012) and Molnar (2008, 2010). This has, as found in the previous section, been found to correlate well to other tests; in Gyllstad’s research he found they considerably correlated with the results of other tests.
The process of selecting an appropriate test for L2 learners was fairly straight-forward. As Schmitt et al clearly state: “The closest thing the field has to such a vocabulary test is the Vocabulary Levels Test (Nation, 1983, 1990)” (Schmitt et. al. 2001, p. 55). The Vocabulary Levels Test, VLT, is comprised of different levels corresponding to how
frequently the words are used in the English language. The first 2000 words are thought to be for simple spoken usage and the next level, 3000, is when L2 learners are beginning to read authentic texts. The 5000-word level is the level where a learner ought to be able to read most texts and the 10000-word level is where the learner has a wide vocabulary. The words chosen are from the most commonly used words in the English language, ’high frequency words’. The words most used in the English language are included in the first level, less used words in the next and so on. When designing the test one word can only occur once from the same word family. If the word from the same word family occurred more than once on any level the researchers chose the word that had the highest occurrence, and only included it in the test once (Schmitt et al. 2001, p. 59).
Since the conception of the Vocabulary Size test in the eighties by Nation, it has undergone some validation studies and other researchers have created similar tests based on the same data as Nation. The first ones were called VLT tests A-D. In an attempt to validate the results further they created tests called E and F, to finally call them Version 1 and 2 respectively. The main differences between the old test, VLT, and the new versions, VLT 1 and 2, were that they had had some changes in vocabulary on the 2000-word level as well as a different academic word list. Both versions were trialled on learners and finalized after a test-run and amendments. Schmitt made conscious decisions in an attempt to make the two versions the same level of difficulty, but with different words (Schmitt et al. 2001).
Upon analysing the final results of the two new versions of the test the results were similar overall, as expected, but when Schmitt broke the results down into the categories Romance Speakers, Non-Romance Speakers and Slovaks, he saw tendencies that
Romance-10
Speakers received slightly higher results in version 2 on the 3000, 5000 and 10000 word-level, whereas non-Romance speakers and Slovaks received higher results on the academic level. The results on the 2000-word level were indistinguishable (Schmitt et al. 2001, pp. 77-79). This ultimately influenced the decision on what test to choose for the study.
In his collocations study Gyllstad (2007) carried out some comparable research to validate his findings. He had used the VLT 1 test in order to compare the results between his COLLMATCH test and the vocabulary size test. He carried out his test on Swedish students at university level, not currently studying English. His results are interesting as a comparison to this study as the students tested in his study had studied English for another year beyond the learners tested here. His results from the VLT 1 test are detailed in table 3 below as raw scores. Each level has a maximum score of 30. The Academic word list, below, is not based on the most commonly used words in the English language, like the levels, but a separate vocabulary taken from words used in an Academic setting.
Table 3. Test scores for Gyllstad’s study (2007, p. 208)
2000 3000 5000 Academic Max score
29.3 29.2 26.2 28.2 30
2.4 Learning an L2 and an L3
Both Swedish and English are Germanic languages. Germanic languages originally stem from the Indo-European language group, which includes English. English and Swedish are
Germanic languages but Hungarian is a Uralic language. This means they share a common core vocabulary and many words are similar. An example is Fader in Swedish, Father in English and Vater in German, another Germanic language. Inevitably, Swedish speakers should find it easier to recognise words and sentences than, for instance, a language such as Arabic where there are essentially three different words for father, depending on whose father is spoken about.
Szabo (2016) looked into, amongst other things, the vocabulary size in multilingual individuals, i.e. those with an L3, in this instance Hungarian (L1), Romanian (L2), and English (L3). Although Szabo set out to determine whether or not specific teaching methods could increase L3 vocabulary sizes amongst Hungarian native speakers, a baseline test had to be carried out. The findings indicated a higher vocabulary in English (the L3) by Hungarian speakers who had higher vocabulary scores in Romanian compared to Hungarian speakers with lower scores in Romanian. It is interesting to note that Romanian, Hungarian
11
and English are all part of different language groups but the reasons for the results, Szabo states, is that Romanian and English do have a high number of cognatesin common. Cognates are words in two languages that have similar meaning and spelling. An example could be introduction (English) and introduction (Swedish).
”/…/ there is a sizeable correlation between L2 Romanian and L3 English lexical knowledge, suggesting that the higher vocabulary size in one
language, the higher it is likely to be in the other language”. (Szabo 2016, p. 20)
Chen et al. set out to measure English as a Second Language in young children with Spanish L1 and Chinese L1 respectively, compared to monolingual learners of English. Their research predicted that cognate recognition in English would be higher amongst Spanish L1 speakers than Chinese L1 speakers and therefore yield higher scores in vocabulary tests. Their findings were that monolingual speakers of English received the highest results on the vocabulary tests, considering that their first language is English that was expected. The second result was that the gap between the Spanish L1 learners of English and English L1 (monolingual) learners was smaller than between Chinese L1 learners of English and Spanish L1 learners of English. Chen et al. drew the conclusion that: “/…/ cognate awareness is effective in reducing the gap in vocabulary development for Spanish-speaking ELLs” (2011, p. 2011).
Chen et al. also compared their results with length of stay in Canada and found that, for Chinese L1 learners of English L2, their knowledge of cognate words increased significantly. They found that:
Exposure to a language is crucial for vocabulary development in that
language, especially when learners’ first language has little overlap with the second language (Chen et al. 2011, p. 2013)
This effect was not seen amongst the Spanish L1 learners of English L2 to any greater extent in Chen et al’s test, as they already had a high number of cognate words to rely on.
Molnar set out to research a way in which having an L2 could positively influence the vocabulary size in an L3 through the explicit teaching of cognates between L2 and L3. The findings were that, on test levels with many cognates, the results were
significantly better for the test group with an L2 than the control group with only an L1. On test levels with fewer cognates the results were smaller. However, the learners should have sufficient knowledge of the L2 in order to benefit in their L3 (Molnar 2010, pp. 346-347).
In an older study, Molnar (2008, p. 2) set out to compare L3 language acquisition when the L1 is not typologically closer to the L3 than the L2. Molnar held the
12
view that there is most definitely a positive transfer between similar L1 and L2 languages as well as if the L3 is similar to both L2 and L1. However, the question is whether the L3 is learnt through accessing the L1, which the learner is more proficient in, or the L2, which is more closely connected with the L3. His study found that transfer occurs from the native language but, most likely, from the language most similar to the L3 (Molnar 2008, pp. 3-4). He subsequently presented findings that monolingual learners had a statistically significant lower score than the bilingual learners where the L2 was more closely linked to English (Molnar 2008, pp. 3-4). The view that bilingual L3 learners perform better than monolingual learners is confirmed by previous studies by Jacqueline (1998), Molnar claimed.
In two further studies it was also found that the L2 influenced the L3 to a much higher degree than the L1 when the L2 was more typologically similar to the L3 than the L1 (Lasagabaster 2000, pp. 170-172). In a similar study in Sweden between Finnish L1 learners of English and Swedish L1 learners of English, with corresponding L2 languages, it was found that regardless of the learner’s L1, Finnish or Swedish, the transfer took place from Swedish to English (Ringbom 1986, pp. 155-156). This means that Finnish L1 learners use their L2 (Swedish) to access their L3 (English). This is because Swedish is typologically closer to English than Finnish. Swedish is part of the same language family as English, Germanic languages, whereas Finnish is part of the Uralic language group.
3 Method
This aim of this study is to compare English vocabulary in native Swedish speakers with non-native speakers of English as well as relating proficiency in English to proficiency in Swedish for the native speakers. It is hypothesised that Swedish L1 learners will outperform non-Swedish L2 learners, and it is further hypothesised that a better non-Swedish vocabulary will result in a better English vocabulary. In this section, the procedure, participants, collection of data and validity and reliability will be presented.
3.1 Procedure
After reviewing the research in section 2.1, it became apparent that the most reliable way of proceeding was a vocabulary levels test of ’breadth’. The other forms of tests were either too new or unreliable (Gyllstad’s collocations tests) or too complex to carry out during a lesson (the lexical tests). The choice had been between a test of depth and a test of breadth. The depth test was discarded as a viable option as it was unsure it would show anything of value at
13
the chosen learners’ level of English. The choice most suitable to the study was a traditional vocabulary levels test. A traditional vocabulary test (breadth) gives the participant a word that needs to be matched to a word of similar meaning (synonym).
The next step in the decision-making process was how to design the test. It needed to be built on proven scientific research and also have a progression of difficulty built in. Most of the research in section 2.1 has either in full or in part used the Vocabulary Levels Test based on research by Nation in 1983 and 1990. The word bank (the words that are the most frequently used in the English language) created by Nation is well-known and widely accepted in the literature to be accurate, and was thus chosen.
The next step in creating a test was to decide whether to use the VLT, VLT 1 or VLT 2 test. The original VLT test by Nation was discarded as it used an older vocabulary list and had not been validated as extensively as the VLT 1 and 2 tests. Schmitt et al. (2001, p. 77-78) found that Romance speakers on the 3000, 5000 and 10000 level had slightly higher results than non-Romance speakers in the second version of the test, apart from
the academic level where the reverse was found. As the intended group of foreign-born learners would be mostly non-Romance, i.e. Arabic-speaking, learners with few cognates in their language, it was intriguing to see whether the same would be found in the results of this study, so the VLT 2 test was chosen, which was obtained digitally from Lextutor1.
The final step was to select whether to use the whole VLT 2 test or parts of it. The research would have to be carried out in classes where there was a finite amount of time, therefore the best option was to keep the levels of the test intact but limit the number of questions and levels, thus making sure all learners had time to complete the test in one lesson, as well as maintaining focus throughout. As this study wanted to look at comparing L2 learners’ progression in Swedish linked to their progression in English some background questions were added on the front as well as some Swedish vocabulary questions at the back for Swedish L2 learners to complete (Appendix A and B). The tests created for this study followed the word levels outlined in the VLT 2 and used the same words as that test, but fewer words. The VLT 2 had 10 questions per level but this study only included 6 questions per level due to the time constraint.
The Swedish test had 21 question and was designed with the same layout as the English test, as not to confuse the learners, but with a different word bank. Half of the words ____________________________________________
14
were taken from recent diagnostic tests for year 9 pupils and half of the words were taken from the university admission’s word test (Högskoleprovet). This made sure there was a spread of difficulty. It was unfortunate that there is no reliable Swedish word bank to base the test on. Both test used the VLT 2 system where you had 6 words on the left that corresponded to three words to the right. Each question could yield 3 points (see appendix A and B).
3.2 Participants
To obtain valid results it was vital to be careful to compare learners with similar levels of English to each other. The logical choice was to compare learners at a level where it was certain that they had previously been instructed in English to a similar level. A decision was made to select learners from upper secondary, where foreign-born and Swedish-born alike have had to have a pass grade from elementary school in order to be accepted onto the course. This would make sure, to the greatest extent possible, that the playing field was level.
The idea to carry out the research in the same school for both groups was contemplated, but there would have been great difficulties in making sure that the foreign-born learners’ Swedish and English had not developed or been influenced too much by attending a Swedish school, as that would have influenced the results. Therefore, the decision was made to carry out the test for foreign-born learners at Komvux, school for adults, where they had been exposed to less Swedish than in a regular school. The test for Swedish-born learners was carried out in a regular school in the same town. Both groups were given the test towards the end of the English 5 course they were studying.
In the group of Swedish L1 learners there were 32 participants, all of which considered themselves (by marking the appropriate box) as having Swedish as their first language. There may be some who marked this box although they do not meet the criteria for a Swedish L1 learner so they may in reality be bilingual. The Swedish L1 participants
consisted of 24 females and 8 males attending Upper Secondary School (17-18 years old). In the group of Swedish L2 learners there were 37 participants, of which 30 had Arabic as their first language and the ages ranged from 21 to 35. The remainder were a mix of Spanish, Eastern European and other languages. Similarly to the Swedish L1 group, this group also had more females (22) than males (15) but not to the same degree.
3.3 Data & Method of Analysis
The tests were collected anonymously and corrected. For the Swedish-born learners (Swedish L1/English L2) the data collected was limited to their test scores. The additional data
15
collected in the test for the foreign-born learners’ (Swedish L2/English L3) tests were: length of stay in Sweden, level of Swedish language studies completed, a Swedish comparison test, age, gender and L1 language. The results were tallied and compiled in Excel and compared by each subset mentioned in this section. The data for the Swedish test and the English test was put into Excel for a correlation test and each individual test result was also put into a
scatterplot to visualize a potential correlation. 3.4 Validity and Reliability
In order to evaluate how well a test, in this instance, measures what it is meant to measure the test designer needs to show the test result is valid (validity) and reliable (reliability). There are two types of validity; external validity which is where the results can be applied to other situations; internal validity is where it is shown that the results in this particular study were not influenced by other factors. In order for the results to be reliable the results need to be able to be replicated under the same conditions (Scribbr 2021).
In sections 3.1-3.3 the reasons for choosing the VLT 2 test and how the test for this study was designed have been discussed, as well as a description of the participants. The complete test is attached in Appendix A and B. This information makes it possible to replicate the study and therefore the results are reliable. “External validity refers to the extent to which results from a study can be applied (generalized) to other situations, groups or events”
(Scribbr 2021). The results in this study were also compared to the study of Gyllstad (2007) where the scores in his VLT 2 test were similar to those in this study, as well as Chen et al. (2012), Molnar (2008, 2010) where the results were similar. It should be noted that the low number of learners means that the results cannot be formally validated externally but could be seen as an indication that the results may be in line with the result of Gyllstad.
The tests were carried out on a small number of learners, where one or two strong or weak learners could skew the results, impacting the internal validity negatively. To investigate that this had not happened an analysis was made of the spread of the results. It was clear that by removing the weakest or strongest results in each group the results would
change, but only slightly. It was deduced there still may be an issue with internal validity as other factors or variables could not be ruled out: “Internal validity refers to the degree of confidence that the causal relationship being tested is trustworthy and not influenced by other factors or variables” (Scribbr 2021). Some of the remaining internal validity issues of age, degree of exposure to Swedish/incidental exposure and educational background will be examined below.
16
The main factor where the results could have varied is that it was too troublesome to account for the background of the learners, i.e. their schooling. For the Swedish-born pupils it was fairly straight-forward, on the English 5 course they had had 9 years of compulsory schooling previously with a certain amount of English instruction from year 2. For the Swedish L2 learners their school background varied, from someone who had no schooling and was completely illiterate a couple of years ago (before they came to Sweden), to those who had attended university in their home country. It is possible that the results would have been different had only those with a directly comparable school
background to the Swedish L1 participants been selected, but that was not possible. This could affect internal validity of the findings through selection bias when comparing the results of the two groups (VeryWellMind 2021).
Within the group of Swedish L2 learners of English the issue of a directly comparable background may be an internal validity concern. Although the learners were randomly selected and a representation of a group of immigrants at the same level of study at this point of time for this area, they may not be respresentative of the Swedish L2 learners of English as a whole (Scribbr 2021). However, the study also compared the Swedish L2 learners’ Swedish skills and correlated them directly with their English results, making the factor of school background less important in the third research question, but may be an internal validity concern when it comes to the factor of length of stay in Sweden, which is part of the second research question.
Although the Swedish L1 learners most likely had studied, or were studying, a third language this was not part of the study, but may have influenced the results. Equally, the Swedish L2 learners may have learnt another language previously as well, possibly with substantially more cognates with English and Swedish, and this may also have had some limited impact on the results affecting internal validity to some extent.
The two groups also differed in terms of age, the Swedish-born test-takers were approximately 17 years old whereas the foreign-born test-takers were older, 21-35 years old. The Swedish L1 learners of English had studied English for a considerable amount of time. They had also been exposed to English through radio, TV, video games and generally in society. Meanwhile, the Swedish L2 learners of English were older and had studied English for a considerably shorter time. Both factors may have had a negative impact on the results of the Swedish L2 learners of English and the validity of the findings in research question 3.
In relation to the first and second research question, comparing the results within the Swedish L2 group, factors that could have impacted the internal validity were the question
17
of immersion and exposure to Swedish (and English). None of the Swedish L2 learners of English had been in Sweden less than 2 years. In that time learners will have had incidental exposure to Swedish (and English) through factors such as having worked, lived in an area with many or few Swedish people, and having Swedish friends or not. These factors cannot be accounted for accurately and may have influenced the internal validity.
4 Results
The results from this study will be presented in this chapter. In 4.1, the results from the Swedish L1 and the Swedish L2 English test will be presented. In 4.2 and 4.3 the results from the Swedish L2 learners will be presented in more detail.
4.1 General results and by gender
In this study, 32 learners with Swedish as L1 participated, of which a significantly higher proportion were female (24) than male (8). As expected, the female learners’ scores were higher than the male learners’ scores on all levels, but showed an interesting tendency to have narrowed at the highest level, 5000, see Table 4. These results are very much in line with the results presented by Gyllstad (2007, p. 201) in his study of Swedish university students. Table 4. Results from Swedish L1 participants by word level and gender, raw score.
Word level Female Male Combined genders Max score
2000 17.5 16.9 17.3 18 3000 16.5 15.6 16.3 18 5000 15 14.4 14.8 18 Academic 7.9 7.8 7.9 9 Total all levels 56.9 54.7 56.3 63
In the study, 37 learners with Swedish as their L2 participated, of which a higher proportion were female (22) than male (15). In the same way as the Swedish L1 learners, the results narrowed between the genders at the 5000 level instead of increasing, see Table 5. This may be due to the skewness of gender participants or the small number of participants.
Table 5. Results from Swedish L2 participants by word level and gender, raw score.
Word level Female Male
Combined
18 2000 12.3 10.7 11.7 18 3000 10.1 9.4 9.8 18 5000 8.2 7.7 8 18 Academic 4.2 3.9 4 9 Total all levels 34.8 31.7 33.5 63
Comparing the data from table 4 and 5 (in Figure 1) the differences are quite substantial, given they were studying English at the same level. Some of the results may be explained by the higher achievement by females than males in English. The other factors to account for are the educational background of the participants. The combined gender results on the lower level, 2000, differ with approximately 35.1% but as the difficulty increases the attainment gap also widens, to 36.2% at the 3000-word level and finally a 37.8% gap on the 5000-word level. This percentage is calculated by subtracting the lower from the higher number and dividing it by the total for the level. This is similar to Schmitt’s el al’s (2001, p. 78) findings that there was a high number of cognates on the 3000 and 5000-word level, influencing the results on those levels. On the Academic word level, which is a different set of words than the high frequency words tested on the other levels, the Swedish L1 learners outperformed the Swedish L2 learners with in excess of 43%.
Figure 1. Comparison of L1 and L2 learners’ English test results, combined genders. 4.2 Results by age and length of stay in Sweden
As discussed in section 2, length of stay (residency) had not been proven to be a deciding factor in the Canadian study of learners with Spanish or Chinese L1 respectively. The findings in this study confirm that it is not a vital factor in determining the learners’ vocabulary size. Out of the 37 participants with Swedish as their second language none had been in Sweden less than 2 years. 21 had been in Sweden 3-4 years and 14 had been in Sweden for more than
0 20 40 60 80
2000 3000 5000 Academic Total all levels
19
5 years. The results showed a small tendency for a higher score on the high frequency words the longer they had been in Sweden (on the 2000, 3000 and 5000-word level), but the reverse was found on the Academic word level, see Table 6. This is in line with the research on cognates, where they were found to be more frequent in the high frequency word lists than the Academic word list.
Table 6. Swedish L2 learners’ raw scores by length of stay in Sweden.
Up to 2 years 3-4 years 5 or more years Max score
2000 None 11.4 11.8 18
3000 None 9.7 10 18
5000 None 7.6 8.3 18
Academic None 3.3 2.8 9
Total all levels None 32 32.9 63
Another factor that could have had an impact was the learners’ age. In this study, the youngest Swedish L2 participant was 19 and the oldest was 37. No significant results could be noted, other than the youngest learners got a slightly better result, see Table 7.
Table 7. Swedish L2 learners’ raw scores by age.
18-25 years 26-30 years Aged 30 or above
2000 11.8 11.5 11.4
3000 9.9 9.8 9.7
5000 8.4 7.7 7.6
Academic 3.3 3.1 2.9
Total all levels 33.4 32.1 31.6
4.3 Results by proficiency in Swedish
The most interesting results from the study are those of a possible correlation between proficiency in Swedish (L2) and that of the target language English (L3). All participants in this study have had to complete elementary school Swedish (svenska grund) before applying for English 5, which is the course where this study was performed. Therefore, all of the
participants had at least a working understanding of Swedish. All participants have also had to complete elementary school English (engelska grund) as well, before applying for English 5.
20
The first data compared is that of test scores compared to completed studies of Swedish as a Second Language 1 or 2 (Svenska som Andraspåk 1 or 2; SvA 1 or Sva 2) or
those who had only completed elementary school Swedish (Svenska grund). Those are the names of the high school Swedish courses for immigrants. Unfortunately, none of the participants had completed Swedish as a Second Language 3 (Svenska som Andraspråk 3). For the purposes of comparison, the scores from the different test levels were combined in Table 8.
Table 8. Swedish language studies level completed related to average English raw test scores.
Swedish level completed Elementary SvA 1 SvA 2 SvA 3 Max Score
English Test score 33.8 34.6 35.1 None 63
The difference between levels is small, between Svenska grund and Svenska som andraspråk 1 it is 0.8 points and even smaller to the next level, 0.5 points. This would suggest that there is limited connection between an increased proficiency in Swedish and increased vocabulary in English.
The second set of results are far more interesting. A correlation test was run between the English test scores and the Swedish test scores. The results indicate that there is a very strong correlation between a good result on the Swedish vocabulary test and the English vocabulary test. The correlation coefficient was 0.76. This value resulted in a P-value of 0.046 and thus there is a statistically significant correlation between the two tests. The data for each Swedish L2 learner, their Swedish test score and their English test score, had been correlated above and was subsequently plotted onto a scatterplot (Figure 2). The X axis is the learner’s English test score (out of 63) and the Y axis is the same learner’s Swedish test score (out of 63 also). Using the scatterplot and the correlation coefficient is vital as the raw score for the Swedish test is significantly higher than the raw score for the English test.
21
Figure 2. Scatterplot of correlation between the Swedish and English language tests Looking at the trendline in figure 2, moving left to right, it is clear that the correlation
between scores on the Swedish test and the English tests is strong. Most results are within one or two points of the trendline, which indicates that most participants make similar advances in both Swedish and English vocabularies.
With the conflicting results above (Table 1 and Figure 2), having expected both a learner’s Swedish level and a learner’s results on the Swedish test to correspond positively to the English test results, it was necessary to look at the Swedish test results per Swedish study level. The max score on the Swedish test in this study for L2 learners was 63. There was a progression of only 1.8 points between the elementary level, Grund, and the highest level represented here, SVA 2. This confirmed a suspicion that learners did not acquire huge amounts of vocabulary whilst attending Swedish Language studies. It seems that the small progression in English test score (Table 8) is mirrored in the small progression of Swedish test scores, as shown in Table 9.
Table 9. Swedish raw test scores per study level.
Grund SvA 1 SvA 2 SvA 3 Max Score
43.2 43.7 45 None 63 30 35 40 45 50 55 20 25 30 35 40 45
English Test results
S w ed is h te st r es ul ts
22
5 Discussion
Learning a new language is not exclusively about learning words, but having a greater vocabulary will enable a person to express and understand things in a better way; words are the building-blocks of language. An ability to converse or read/write in a global language, English, is a valuable skill and Gyllstad (2007) pointed out that vocabulary is a predictor of other language skills also. The discussion follows the research questions and deals with each question in one paragraph each, starting with the third research question.
The third research question was: Is there a significant difference between Swedish-born learners’ and foreign-born learners’ English vocabulary? It was an expected outcome that the Swedish-born learners of English outperformed the foreign-born learners of English, which is what the study found. Most foreign-born participants in the study were Arabic-speaking, approximately 85%, and therefore their L1 was a language not typologically close to English, as Swedish is, nor a language sharing many cognates with English, as
Swedish does. The study found there was a big gap between Swedish L1 and Swedish L2 learners of English. Swedish learners typically perform well on English tests, and the scores in this study were very similar to those of Gyllstad’s (2007) results, where most Swedes in the study only made minor mistakes. In the study by Molnar (2010), it was found that
multilinguals performed better than bilinguals. The conclusion of this study should, therefore, have been that the Swedish L2 learners of English, multilinguals, should have outperformed the bilingual Swedish L1 learners of English. This, however, is not the case in circumstances where the target language is typologically closer to the L1, such as Hungarian and Romanian (Szabo 2016; Molnar 2008). So, the results in this study, where the Swedish L2 learners were predominantly Arabic-speaking, were as predicted. It was interesting to note that the
Academic level results narrowed in other research but in this study the gap increased rather than narrowed. In research this was explained by the higher number of cognates on the 3000 and 5000 levels, and fewer on the Academic level. One reason the reverse was found in this study could be the Swedish education system, which trains pupils early to use certain words in their writing. These words could possibly be closer to the Academic word list than the
languages in the research section (Chinese, Spanish, Hungarian, Romanian). Essentially, pupils in the Swedish school system are trained to write in an Academic way early on.
The second research question was: Do other variables such as age, gender and length of stay in Sweden influence the size of a learner’s vocabulary? The results in terms of gender produced in this study confirm other studies that females perform better and also
23
receive better grades in English. The study results show that both Swedish L1 females and Swedish L2 females performed better in the English test than their male counterparts,
although there were few males represented amongst the participants of both language groups. Age seemed to have a smaller impact on the results as well, possibly due to the fact that many of the participants started learning English as adults. Some research points to the fact that older learners acquire language faster than younger, and could be a further field of study to see the effects this has. It was more surprising that the study showed that length of stay in the country had practically no effect on the results in the VLT. This was, however, also shown in Chen et al.’s (2011) study of Chinese and Spanish children learning English. They showed that it was not how long an individual had been in the country but rather the exposure to language that determined language growth. In this study no data was collected to determine whether or not or to which degree a learner had been exposed to Swedish, but it can be
deduced that an individual who participates in Swedish society fully would increase his or her Swedish vocabulary, and subsequently their English vocabulary. There may be a validity issue here due to the fact that it cannot fully be accounted for how integrated in Swedish society the Swedish L2 participants were at the time they took the test. However, it can be deduced that a targeted approach with specific vocabulary teaching would boost an immigrant’s overall language in Swedish and therefore make English easier to learn.
The first research question was: Are there correlations between an immigrant’s vocabulary size in Swedish and their English vocabulary size? As predicted by the results in the research section, an increased Swedish vocabulary meant that, to a statistically significant degree, participants performed better in the English VLT also. The surprising results were, though, that students progressing through the Swedish language courses (Svenska som
Andraspråk) did not increase their Swedish or English vocabulary to any great extent. In other words, how large a vocabulary an adult (Swedish L2 learner) has does not seem to be closely linked to attending school in Sweden. Ringblom (1986) had concluded, in his study of Finnish learners of English who also knew Swedish, that they relied on Swedish-English cognates, not their native tongue Finnish, in order to learn English. Lasagabaster (2000) found the same in his study. Starting with the understanding that it is the exposure to language (in this case Swedish), that will increase vocabulary rather than the length of stay in a country, makes it vital to make sure that Swedish language instruction is offered and expected in order for immigrants to also have easier access to English. Similarly to Taiwan, Swedish language instruction of vocabulary does not do well and the author of this study concurs with Webb and Chang’s suggestion (2002, p. 120): “[it] highlights a need for concrete vocabulary learning
24
goals within and between courses.” But, how are immigrants exposed to language and subsequently included in society to a greater extent? Possibly through increased and targeted vocabulary training, where both Swedish and English stand to gain, or for that matter any language with a high number of cognates and similar typological features. Logically, it could possibly work the same the other way around under the same conditions, i.e. pupils who have strong English skills could learn Swedish faster. This would be interesting to examine further in a separate study.
5.1 Conclusion
The aim of this study was to determine whether foreign learners of English would benefit from having a greater Swedish vocabulary when learning English vocabulary, and if Swedish L1 learners of English performed better on tests than Swedish L2 learners of English. The hypothesis was that participants with stronger Swedish skills would perform better in English vocabulary tests. An English test, the Vocabulary levels test (VLT), was carried out to
determine whether this was the case. The results of this study confirm that learners with a higher score on the Swedish vocabulary test also scored higher on the English vocabulary test, and Swedish L1 learners of English performed better than Swedish L2 learners of English. Due to to discrepancies in age, educational background and exposure to Swedish of the Swedish L2 learners of English there may be some internal validity issues with the findings.
The study also looked at whether other factors influenced the results on the English vocabulary test, such as age and length of stay in the country, for the Swedish L2 learners of English. The study confirmed that those factors played a smaller part in the results for thayt group, but due to the difficulty in accounting for the actual exposure to Swedish and English prior to the test those findings may be difficult to accurately assess.
This study found that Swedish L2 learners of English with a higher Swedish vocabulary also scored higher on the English vocabulary test, confirming research by Chen et al. (2012), Bardel & Falk (2007), Molnar, T (2008) and Gyllstad, H (2007) that a learner’s second language (Swedish) is important whilst acquiring a third language when they are typologically close to each other and have a high number of cognates.
25
References
Bardel C & Falk, Y. 2007. The Role of the Second Language in Third Language Acquisition: The Case of Germanic Syntax. Second Language Research, SAGE Publications, 23 (4), pp. 459-484
Chen et al. 2012. Comparing Vocabulary Development in Spanish- and Chinese-speaking ELLs: The Effects of Metalingustic and Sociacultural Factors. Read Writ, 25, pp. 1991-2020.
Collins Dictionary. 2014. Collins English Dictionary, 12th edition – FAQs. Available at: https://blog.collinsdictionary.com/language-lovers/collins-english-dictionary-12th-edition-faqs/ (Accessed 2021-03-05)
Conway, A. 2008. An Analysis of Somali Pronunciation Errors. [pdf]
https://www.hamline.edu/WorkArea/DownloadAsset.aspx?id=2147491027 (Accessed 2021-04-16)
Gyllstad, H. 2013. Looking at L2 vocabulary knowledge dimensions from an assessment perspective – challenges and potential solutions. Eurosla Monographs Series, 2, pp. 11-28.
Gyllstad, H. 2007. Testing English Collocations: Developing Receptive Tests for Use with Advanced Swedish Learners. Lund University, Sweden. Available at:
https://lup.lub.lu.se/search/ws/files/5893676/2172422.pdf]
Lasagabaster. 2000. Three languages and three linguistic models in the Basque educational system. In J. Cenoz and U. Jessner (eds) English in Europe: The Acquisition of a Third Language, pp. 179 -197
Lextutor. Vocabulary Levels Test v.2. Available at: https://www.lextutor.ca/tests/vlt2/ (Accessed continuously).
Milton, J. 2013. Measuring the contribution of vocabulary knowledge to proficiency in the four skills. Eurosla Monographs Series, pp. 57-79.
Mitchell et al. 2013. Second Language Learning Theories. Third edition. New York: Routledge.
Molnar, T. 2008. Second language versus third language acquisition: A comparison of the English lexical competence of monolingual and bilingual students. Toronto Working Papers in Linguistics(TWPL), Volume 33, pp. 1-16
Molnar, T. 2010. Cognate Recognition and L3 Vocabulary Acquisition. Acta Universitatis Sapientiae, 2, 2, pp. 337-349.
Read, A & Chapelle, C. A. 2001. A framework for Second Language Vocabulary Assessment. Language Testing, 18(1), pp. 1–32.
26
bakgrund. [pdf] https://eso.expertgrupp.se/wp-content/uploads/2014/12/ESO-2017_3.pdf [Accessed 2021-03-05]
Ringbom, H. 1986. Crosslinguistic Influence and the Foreign Language Learning Process. In E. Kellerman, & M. Sharwood Smith (Eds.),Crosslinguistic Influence in
Second Language Acquisition (pp. 150-162). New York: Pergamon Press.
Schmitt A, Schmitt D. & Clapham C. 2001. Developing and exploring the behaviour of two new versions of the Vocabulary Levels Test. Language Testing, 18 (1), pp. 55–89.
Scribbr. 2021. Reliability vs validity: what’s the difference?
https://www.scribbr.com/methodology/reliability-vs-validity/ (Accessed 2021-04-10) Statista. 2021. Immigration to Sweden from 2010 to 2020.
https://www.statista.com/statistics/523293/immigration-to-sweden/ (Accessed 2021-03-05)
Statista. 2021. The most Spoken Languages Worldwide in 2021.
https://www.statista.com/statistics/266808/the-most-spoken-languages-worldwide/
(Accessed 2021-03-05)
Svenska Akademiens ordlista. 2014. Ny upplaga av Svenska Akademiens ordlista (SAOL)
https://www.svenskaakademien.se/press/ny-upplaga-av-svenska-akademiens-ordlista-saol (Accessed 2021-03-05)
Szabo C. A. 2016. Exploring the Mental Lexicon of the Multilingual: Vocabulary Size, Cognate Recognition and Lexical Access in the L1, L2 and L3. Eurasian Journal of Applied Linguistics, 2(2), pp. 1-25
Thomas, J. 1998. The role played by matalinguistic awareness in second and third language learning. Journal of multilingual and multicultural development, 9, pp. 295-333 VeryWellMind. 2021. Understanding Internal and External Validity.
https://www.verywellmind.com/internal-and-external-validity-4584479 (Accessed 2021-06-01)
Webb S. A. & Chang A. C-S. 2012. Second Language Vocabulary Growth. RELC Journal 43, pp. 113-126
