Studies on native speakers of Arabic have explored the acquisition of Arabic sounds by children in different developmental stages, as discussed previously in Section 2.3.7 (Al Amayreh, 1994; Amayreh, 2003; Amayreh & Dyson, 1998, 2000; Dyson & Amayreh, 2000).
Additionally, Abdul-Kadir and Sudirman (2011) conducted a study on bilingual children L2 learners of Arabic. The authors investigated the difficulties of MSA phonemes spoken by Malaysian primary school children aged from seven to eleven years old. The researchers analysed recordings of 25 children, where each pronounced 25 Arabic consonants. The study concluded that the fricatives were the most difficult sounds to pronounce, especially the phonemes /ðˁ/, /ɣ/, /ħ/, /x/, /ʕ/, /h/.
The difference between L1 and L2 acquisition can be seen clearly in the conclusions of Abdul-Kadir and Sudirman (2011) and Amayreh and Dyson (1998), who showed the role of speakers’ language backgrounds. For example, monolingual children acquired the sound /ħ/ in their early period (Amayreh & Dyson, 1998), while bilingual children found it difficult to pronounce the sound /ħ/ accurately (Abdul-Kadir & Sudirman, 2011). This also applies to the sounds /ɣ/, /x/, /h/, which monolingual children in the transitional period had the ability to acquire quickly, while bilingual children faced difficulties in pronouncing these sounds.
With regard to adult L2 learners of Arabic, very few cross-linguistic studies have discussed the accuracy of perceiving and producing accurate Arabic segments,
including vowels and consonants, at different levels of proficiency (e.g. Al Mahmoud, 2013; Alosh, 1987; Alsulaiman et al., 2014; Alwabari, 2013; Kara, 1976). For instance, Shehata (2015) has questioned the claim about the difficulties of Arabic consonant phonemes in both perception and production by 107 adult native English speakers residing in the US. An online questionnaire was used, which consisted of questions about participants’ backgrounds and self-rating of the level of difficulty or easiness of all Arabic consonants in both perception and production. The results revealed that 90 out of 107 participants (84%) indicated the importance of learning Arabic language. Moreover, 97 out of 107 participants (90.6%) agreed that there is a difficulty with Arabic consonant sounds for native English speakers, and they considered the sounds /ħ, dˤ, tˤ, sˤ, ðˤ/ the most difficult consonants to perceive and produce. Furthermore, the results showed that the sounds /ɣ, ʕ, q/ are semi challenging and the sounds /k, j, s, b, f/ are the easiest. Shehata (2015) mapped the easiness of some Arabic sounds to their existence in English and the difficulties of the pharyngeal and pharyngealisation sounds to their absence in the phonetic system of English.
Shehata (2015) gave an insight into the attitude of L2 learners of Arabic with respect to the position and importance of learning Arabic language and the difficulty of Arabic sounds. However, learners’ self-evaluation of their performance did not capture the actual difficulties they are facing in perception and production. The author
discussed results from a previous study, which she claimed to be contradictory to the findings of her study. This study was done by Asfoor (1982), who examined the most intricate Arabic sounds pronounced by native speakers of English. To this end, 34 students were recorded pronouncing 10 Arabic sounds, rated by 24 Arabic instructors
as the most difficult Arabic consonants. The results concluded that English speakers found difficulties in pronouncing the Arabic stops in general.
To have a closer look at the methodology of Asfoor (1982), instructors assigned 10 out of 28 sounds to be the most difficult sounds, and only those sounds were examined, which included /x/, /ɣ/, /ʕ/, /dˁ/, /sˁ/, /q/, /ʔ/, /tˁ/, /dˁ/, and /ðˁ/. The stops that were chosen by the instructors did not include all stops but only emphatics. The stops /b/, /t/ and /k/ were not included in the study. The author generalised his conclusion to include all stops.
The validity and reliability of the methodology used in Asfoor (1982) are questionable for two reasons. First, as commented earlier, a limited number of sounds were examined, based on difficulties assigned by native speakers of Arabic and not by L2 Arabic learners. The author did not mention the issue of pharyngealisation but generalised learners’ deficiencies in pronouncing the emphatic stops to all stops, which is considered a mistake. Second, the author tested students’ pronunciation based on imitating native speakers. The trial included recordings of native Arabic speakers pronouncing words, and the author asked the participants to imitate the recordings, which ignored the implicit knowledge of the learners. The validity of this method is questioned because this kind of test leads participants to merely repeat the words verbatim without using their own knowledge of production (Ellis, 2009). Although Shehata (2015) pointed out in her study that the results of Asfoor (1982) were incompatible with what other researchers found, the results were actually similar because the stops that Asfoor found difficult were emphatic stops and not all stops.
One of the studies in the production of Arabic phonetics by adult L2 learners was conducted by Alsulaiman et al. (2014), who investigated the variation in the
pronunciation of Arabic sounds amongst 33 non-native speakers of Arabic with
different nationalities, Pakistani, Indonesian, Nibali and Indian. Their primary intention for conducting this study was to develop an automatic error detection system for L2 learners of Arabic. The instruments were both read text and casual conversation, and the scripts were isolated words, digits, sentences, and paragraphs.
A number of errors were found in the production of each emphatic sound (/sˤ/ =69, /ðˤ/=52, /dˤ/=48, /tˤ/=13). Interestingly, the authors noticed that there were many commonalities in the types of errors among L2 learners of Arabic, regardless of their language backgrounds. The errors in the production of emphatics were:
• /dˤ/ → /d/ = alveodental voiced emphatic stop substituted with alveodental voiced non-emphatic stop.
• /tˤ/ → /t/ = alveodental unvoiced emphatic stop substituted with alveodental unvoiced non-emphatic stop
• /sˤ/ → /s/ = alveodental unvoiced emphatic fricative substituted with alveodental unvoiced non-emphatic fricative.
• /ðˤ/ → /ð/ = interdental voiced emphatic fricative substituted with interdental voiced non-emphatic fricative.
Moreover, there were unexpected and uncommon errors found in the data for the emphatics /ðˤ/ and /sˤ/ which were:
• /ðˤ/ → /z/ = interdental voiced emphatic fricative substituted with alveodental voiced non-emphatic fricative.
• /sˤ/ → /θ/ = alveodental unvoiced emphatic fricative substituted with interdental unvoiced non-emphatic fricative.
Alsulaiman et al. (2014) accounted for the pronunciation errors by referring to language transfer. The uncommon errors, on the other hand, were explained by confusion in the way the letter was written in the test. In fact, the written form of the sounds /ðˤ/ (ظ) and /sˤ/ (ص) is entirely different from /z/ (ز) and /θ/ (ث), so there might be another explanation for those errors. These uncommon errors require further
investigation to find a reasonable explanation for this substitution.
The speech recorded by participants was different from one speaker to another, which means that the frequency of emphatics in participants’ speech varied.
Unfortunately, the numbers of each emphatic pronounced by each participant or language group are unknown. In addition, the characteristics of participants and their backgrounds were also unknown.
Although Alsulaiman et al.’s (2014) study was conducted mainly for the purpose of developing an automatic error detection system, the results derived from it are useful for L2 Arabic phonology research and serve as a base for what to expect from different non-native speakers of Arabic in their pronunciation of Arabic sounds. This study is the only study found to test the production of adult L2 learners of Arabic from different language backgrounds.