Investigating the Phonetic CLI on Swahili TLA The acquisition of nasal plosive by L1 Mandarin speakers learning English (L2) and Swahili (L3)

. Swahili is an increasingly important language in China with its utility in the Belt and Road Initiative. A number of previous studies in third language acquisition (TLA) applied the concept of crosslinguistic influence (CLI), but there are scant studies focusing on the possible phonetic CLI in the phonetic acquisition of Swahili. The present study investigates possible phonetic CLI from Mandarin as L1 and English as L2 on L3 Swahili. Participants were 35 Swahili learners possessing the target language sequence. In this study, a 26-item questionnaire is used to collect data. The main findings of this research include the acquisition of Swahili nasal plosive consonants blocked by Mandarin and English for different reasons such as CLI source deficiency from Mandarin and negative transfer from similar consonant clusters in English. In addition, the results show that phonetic CLI is greatly caused by English for its similar “nasal stops+plosive” consonant clusters. The findings can fill the research gap of Swahili TLA field and Swahili learners with a deeper understanding of acquiring Swahili nasal plosive consonants.


Introduction
Tanzania is an international cooperation partner of China's Belt and Road Initiative, which plays an everimperative role in Chinese diplomatic policies. To have better relationships and cooperation, China is encouraging more students to learn Swahili, the national language of Tanzania. Therefore, how Chinese students who speak English as L2 acquire Swahili phonetics is worth exploring. However little research has been carried out in the field of Swahili acquisition as a third language since Chinese students usually speak Mandarin as L1 and acquire English as L2. Besides, hardly any Chinese scholar utilized crosslinguistic influence (CLI) to analyze the factors that boosted or blocked Swahili phonetic acquisition. Therefore, this study is of significance politically and academically to fill the blank. Phonemes in Swahili have miscellaneous features. Some of them are distinctive from those of Mandarin and English while some others share similarities with Mandarin or English. The title has applied the similar structure of another research carried out by Lin focusing on L2 and L3 studies [1]. This study devised a 26-item questionnaire that was designed to investigate the possible CLI from Mandarin as L1 and English as L2. The questionnaire has arranged 15 five-level scales to estimate the difficulty of acquiring various Swahili phonemes and the magnitude of influences on different phonemes given by L1 Mandarin and L2 English. Then statistical analysis was carried out by utilizing descriptive analysis and Pearson Correlation analysis in order to see which phonemes have stronger CLI and the potential source of it. The findings would provide references for Swahili language teaching in China.

Literature Review
Crosslinguistic influence (CLA) is the impact of people's knowledge of one acquired language on another language which was initially used merely in the field of second language acquisition (SLA) instead of spreading to TLA [2]. In the years between the 1970s and 1980s, many scientists like Dulay and Burt claimed about inviability of CLI in certain sub-fields of linguistics, which had later been proved by Odlin that transfer could happen in all linguistics systems including phonetics [3][4]. Third language (L3) acquisition (TLA) was begin assumed to be the sub-field of second language acquisition (SLA) meaning that the second non-native language acquired after the first, which changed in the early 2000s when researchers started to regard this phenomenon as a separate field of inquiry [5]. Recent years have seen a dramatic growth of research in the field of TLA, ranging from psycholinguistics and sociolinguistics [6]. Among them, a greater amount of research on L3 phonology and phonetics has been presented at many international conferences [7]. In phonology it seems that all learners (including advanced learners), retain an L1-based accent (and intonation) in their speech, L2-transfer appears to be relatively rare [8]. The team of Kopečková believes that people who studying L3 have two previously acquired languages which may unconsciously mix with other languages as a source of CLI during their L3 acquisition [9]. Geoffrey Schwartz has observed an asymmetry by which voiceless stops were less susceptible to phonetic CLI than voiced stops [10]. Team of Amengual examined how the voiceless velar stop /k/ was realized in English, Spanish and Japanese, and discovered that different bilingual and trilingual groups had varying VOT (voice onset time) values [11]. In addition to claiming that L1 influences the extent to which L2 and L3 values resemble native norms, the Llama team discovered that L3 phonetics is an inherent occurrence of L1 qualities in the L2 as well as the L3 [12]. The nasal plosive, which is uncommon and mainly seen in African languages, has also received little attention from researchers.
Even though Swahili has been greatly documented compared with other African languages, the opinions about the number and examples of Swahili consonants vary considerably (table 1) [13]. According to the team of Awino, the version of Nchimbi (1995) is more agreeable by scholars of Swahili for many years [14]. Therefore, the study would also apply the version of consonants claimed by Nchimbi.
As conclusion, CLI has been attracting the attention of many scholars of TLA field in recent years. Plenty of findings have been made about the phonetic interactions among L1, L2, and L>=3 involving many consonants [8][9][10][11][12]. However, research about L3 Swahili nasal plosive phonemes acquisition by focusing on CLI has hardly been noticed.

Participants
The questionnaire was sent to WeChat (a prevalent social media in China) groups full of Chinese people who learn Swahili as foreign language to make sure they speak Mandarin as L1. The total participants in this questionnaire survey are 37 people learning Swahili through many methods including formal schooling, working abroad and institutional training for different years ranging from 1 year to more than 8 years. Then 35 valid samples among all have withdrawn after selecting participants with the target language learning sequence (Mandarin L1, English L2 and Swahili L3). About 88% of the participants are collage students majoring in Swahili language who contributed the most reliable and valuable information about the true difficulty and influences when acquiring the accurate Swahili phonemes since they have been taught through formal schooling. Only 1 person studied Swahili through working in Africa and 2 people through institutional training. Swahili language skills of participants also vary a lot, with 62.86% have learn Swahili for three years, 17.14% for four years, 8.57% for one year, 5.71%, 2.86% and 2.86% for two years, five years and more than 8 years respectively.

Measures
(1) Background Information Questions An 11-item questionnaire was designed to collect basic personal information including gender, age, phone numbers, sequence of language acquisition, the major way of learning Swahili and English, whether knowing other languages, how long have been learning the two languages, and the time spent in these languages' environment.
(2) Difficulty Scale Five scales ranging from point 1 to point 5 (from 1 being "very easy" to 5 being "very hard") has devised in the questionnaire aiming to access how difficult participants feel in the acquisition of specific phonemes involving vowels, a combination of two vowels, semivowels, nasal plosives, and post-alveolars.
(3) Influence and Similarity Scale Eight scales estimating how strong the influence of Mandarin or English on the target phonemes acquisition ranging from point 1 to point 5 (from 1 being"strong positive influence" to 5 being"strong negative influence") were created in the questionnaire. Two 5-level scales with 1 being "not similar at all" and 5 being "similar a lot" to evaluate how similar phonetically Mandarin or English is compared with Swahili.

Data Analysis
The SPSSAU project (2023) was used to analyze the data. This study compared the values of Pearson's correlation coefficients on the sets of data and made descriptive analysis to show the details [22].

Overview
The following passages are going to analyze the result of the 26-item questionnaire according to three main aspects: An overall analysis of how difficult learners feel in Swahili phonemes acquisition, a statistical descriptive analysis of data collected and several correlation analyses of the significance in data. After that, a discussion would be carried out about how Mandarin and English see differences in affecting L3 Swahili phonetic acquisition and the reason behind it.

An overall analysis of how difficult learners feel in Swahili phonemes acquisition
After collecting opinions from participants on the difficulty of learning different phonemes, the study found the process of phonetic acquisition turned out to be relatively easy for English L2 Mandarin speakers since most of the feedback is ranging from 1 ("very easy") to 3 ("ordinary") with the average point at about 1.83. Even though the general points tend to be low, different phonemes vary considerably. The result shows that the same phonemes shared by Mandarin and Swahili or English and Swahili appear to have the lowest points, having an average point of 1.47 for vowels, 1.74 for semivowels, and 1.79 for post-alveolar consonants like /ch/ (/t̠ ʃ/) and /sh/ (/ʃ/). More than half of the participants feel "very easy" in acquiring these phonemes, reaching 70.59%, 55.88%, and 55.88% respectively. The phonemes which are missing in both Mandarin and English have been proven to be the hardest in the process of acquisition. Among them, typical phonemes are consonants consisting of a nasal stop and a plosive consonant such as /mb/ and /nd/ which possess an average point of 2.24. The phonemes which are similar to those of Mandarin and English are ranked in the middle level. The articulation of two vowels attached together seems to be similar to diphthongs in Mandarin and English, which render it relatively difficult to acquire with an average point of 1.95. According to the feedback, the rate of similarity of 1 Abbreviation for translation of the original questionnaire: "How difficult do you think it is to learn the nasal sounds like /mb/, /nd/, /md/ in Swahili?" 2 Abbreviation for translation of the original questionnaire: How difficult do you think it is to learn the semi vowels /w/, /y/ in Swahili? 3 Abbreviation for translation of the original questionnaire: "How do you think of the influence from Mandarin on learning combination of two vowels in Swahili?" 4 Abbreviation for translation of the original questionnaire: "How do you think of the influence from English on learning combination of two vowels in Swahili?" 5 Abbreviation for translation of the original questionnaire: "How do you think of the influence from the phonetic system between Mandarin and Swahili is lower than that between English and Swahili.

Statistical descriptive analysis of data collected
The 26-item questionnaire collected various data from 36 participants who speak Mandarin as L1 and English as L2. The questionnaire utilized scales from point 1 to point 5 to estimate the difficulty and possible influences in the process of phonetic acquisition. However, some of the items turned out to be invalid with their max points failing to reach point 5 which then were canceled. The chart below presents the result of all the valid data (table 2): Mandarin on learning compound consonants like /mb/, /nd/, /md/ in Swahili?" 6 Abbreviation for translation of the original questionnaire: "How do you think of the influence from English on learning compound consonants like /mb/, /nd/, /md/ in Swahili?" 7 Abbreviation for translation of the original questionnaire: "How do you think of the influence from Mandarin on learning post-alveolar consonants like /ch/, /sh/, /j/ in Swahili?" 8 Abbreviation for translation of the original questionnaire: "How do you think of the influence from English on learning post-alveolar consonants like /ch/, /sh/, /j/ in Swahili?" As the chart shows, two scales that relevant to difficulty in phonetic acquisition were left valid with their average points both under 2.5, meaning that the difficulties of acquiring vowels and semi vowels are generally little. But the acquisitions of former phonemes (M=2.286) are obviously harder than the later ones (M=1.743). All phonemes in the chart see positive influences (less than 3) from both Mandarin and English. It is also noticeable that the consonants consist of a nasal stop and a plosive are the least susceptible (with their means up to 2.743 and 2.886 respectively) to effects from Mandarin and English with English effect also being the most stable (SD=0.867) item in the whole chart. These data indicate that it is commonly believed that the consonants consist of a nasal stop and a plosive have limited effect from Mandarin and English. While influence from Mandarin on post-alveolar phonemes like /ch (ʧ)/ or /sh (ʃ)/ is the most unstable data (SD=1.255) [18].

Correlation analysis on Swahili consonants consists of a nasal stop and a plosive
The study carried out an analysis on the correlation between influence on various sounds and the difficulty of acquiring them. Throughout the whole outcome, consonants that consist of nasal stop and plosive showed the most significance as the table 3. The correlation coefficient values between the influence from both language and the difficulty of acquisition are 0.489 (p<0.01) and 0.465 (p<0.01) respectively, indicating remarkable significance statistically. With the result from Mandarin and English turned out to be similar, it is rather challenging to recognize of the source of CLI because the articulation of nasal stop and plosive sounds may transfer from L1 and L2. Also, the deficiency of sources in CLI also led to the consequence. this result appeared probably because nasal plosive sounds are lacking in both Mandarin and English. According to Mandarin sources of phonological CLI of L3 Swahili acquisition, which is Mandarin applies a phonological system that consonants cannot be attached together, a strong negative impact of the L1 was found. While English has the similar consonant clusters but are relatively less frequently being used in English, so for 9 Abbreviation for translation of the original questionnaire: "How do you think of the similarity between English and Mandarin?" 10 Abbreviation for translation of the original questionnaire: "How difficult do you think it is to learn the vowel sounds in Swahili?" Mandarin native speakers with English as L2, nasal plosive sounds may be less familiar than other sounds, making it difficult to pronounce them accurately. In addition, producing nasal sounds requires mastering the correct pronunciation position and breathing technique, which may require more practice and patience for beginners. Therefore, Mandarin and English both see a relatively negative influence on the difficulty of nasal plosive sounds acquisition.
After analyzing data related to similarity among Mandarin, English and Swahili intending to find out the source of CLI, the remarkable result that indicating significance only appeared at English groups. The Pearson correlation coefficient was used in the study to indicate the strength of the correlations between the difficulties in acquisition and the similarity estimated by participants between English and Swahili as the table 4 shows: As the chart demonstrated, the correlation coefficient value between "Difficulty nasal stop+plosive" and the similarity between both languages was 0.416 and showed significance at 0.05 level, thus indicating a statistically strong positive correlation between these two factors. Participants who regard English and Swahili have bigger similarity tends to contend that consonants consist of a nasal stop and a plosive more difficult to acquire. Linguistic distance is an important factor of CLI meaning the extent of similarity between two languages. English and Swahili belong to different language families and have different sound systems [23]. However, these languages have a more similar phonetic system than that with Mandarin, such as all vowels and most consonants in Swahili can be found in English. Whereas Swahili and Mandarin only share a small number of phonemes. In other words, English has a closer linguistic distance with Swahili. Therefore, if a Mandarin L1 speaker is more familiar with English they may have an easier time learning the nasal stops and plosives (which are similar with the target nasal plosive) of the other language due to the positive effect of CLI led by closer linguistic distance [23]. 11 Abbreviation for translation of the original questionnaire: "How difficult do you think it is to learn the sounds of combination of two vowels in Swahili?" 12 Abbreviation for translation of the original questionnaire: "How difficult do you think it is to learn the post-alveolars sounds in Swahili?" However, even if two languages share some similar sounds, the exact pronunciation of those sounds may differ. In Swahili, for example, the /mb/ sound is pronounced with the lips fully closed (to produce a sound like /m/) and followed closely by a plosive sound, while in English there is no such consonant. What English has is nasal+plosive clusters which usually across the boundary of morphemes such as "dreamt", "home-grown" and "cranberry". That means even though English has three nasal stops (which is /m/, /n/ and /ŋ/), three voiced plosives (which is /b/, /d/ and /g/) and three voiceless plosives (which is /p/, /t/ and /k/), there is no consonant that combines nasal and plosive as a whole like /mb/ or /nd/ [24]. For English, the advantage of helping acquire the nasal stops and plosives clusters has turned to blocks for acquiring the accurate and correct Swahili nasal plosives due to the CLI. Because when English as L2 transferred, speakers tend to process the sounds as a consonant cluster (nasal+plosive clusters) instead of a single consonant. Furthermore, nasal plosive sounds can be challenging for learners of any language, regardless of the sources of CLI. This is because these sounds require a complex coordination between the oral and nasal cavities and may require specific tongue and lip positions.
Therefore, while the close phonetic linguistic distance between English and Swahili may help learners recognize the existence of nasal plosive in the other languages, it may not necessarily make it easier for them to master the exact pronunciation of these phonemes.

Conclusion
This study applied questionnaires to investigate the possible phonetic CLI from Mandarin as L1 and English as L2 on L3 Swahili. According to descriptive analysis, nasal plosive consonants are the least susceptible among the Swahili phonemes investigated. Besides the correlation analysis shows that Mandarin and English both have a relatively negative influence on the difficulty of nasal plosive sounds acquisition due to the deficiency of CLI sources from Mandarin and English. In order to step a further research in where exactly phonetic CLI comes from, the study made correlation analysis between difficulties and similarity and found significance appeared in English groups. Thus, a conclusion could be drawn that phonetic CLI mainly from English for its similar "nasal stops+plosive" consonant clusters.
The author gratefully acknowledge the financial support by the funding of China Scholarship Council.