There has been a substantial amount of research on language maintenance and shift (LMLS) and language variation and change (LVC) in New Zealand in the last four decades and most of this research has concentrated on exploring LMLS separately from LVC. Most researchers deal with these two topics as two different fields. For example, if they study LMLS (e.g., proficiency, domains and attitudes), they don’t focus on the speaker’s production of a language (e.g., vowels and consonants) within the same thesis. This thesis combined both LMLS and LVC in one thesis by employing questionnaires which were gathered from 99 Arab Jordanians and Palestinians to answer three research questions related to LMLS. The first research question related to reported language proficiency and the influence of generation (1st, 1.5 and 2nd) and length of residence (1- 10 years, 11-20 years and 21-30 years) on that. The second research question looked at language use in different domains (e.g., home, friendship and religion) and the influence of generation and length of residence on that. The third research question examined the participants’ attitudes towards both Arabic and English languages in general and New Zealand English (NZE) in particular and cultures and the influence of generation and length of residence on that. 20 of the survey participants who expressed willingness to be interviewed, subsequently participated in recorded interviews, which were used to investigate the realisation of particular consonants (ING and intervocalic /t/ and NZE short front vowels (KIT, DRESS and TRAP) in the speech of Jordanians and Palestinians in Christchurch New Zealand. The interviews aimed to answer four research questions. Two research questions related to the consonants (whether social factors influence the production of these two consonants and whether attitudes collected by questionnaire predict any of the linguistic behavior), and two questions related to the vowels (whether social factors and lexical frequency influence the production of these three vowels, and whether attitudes collected by questionnaires predict any of the linguistic behavior for these vowels).

By combining work in language maintenance and shift with work in language variation and change, this thesis aimed to reveal patterns which could be masked when each question was investigated separately. This is because LMLS and LVC are both driven in part by attitudes. I linked speakers’ attitudes in the questionnaires to their linguistic behavior and examined the influence attitudes have on the production of the variables ING, intervocalic /t/, KIT, DRESS and TRAP. The interviews also provide some explanations for the attitudinal significant correlations

found in the questionnaire and in their productions of the five variables examined. Principal component analysis (PCA) was used to find underlying attitudinal categories from the answers and build up an attitudinal index score for each speaker. The scores were used to evaluate the attitudes of one speaker compared to another toward Arabic language and culture and English language and New Zealand culture.

The results for the LMLS part of the study showed that there is a gradual language shift in all domains (e.g., home, friends and religion), most sharply in the friends domain, then religion and finally home domain among 1.5 and 2nd generations and 11-20 and 21-30 length of residence. In addition, clear regression in Arabic literacy skills among 1.5 and 2nd generations and those who have been in NZ between 11-20 years was found. However, the attitudinal results showed that Arabic Jordanians and Palestinians in Christchurch are very loyal and positive towards their ethnic language because it is intertwined with their Islamic religion and culture. They also showed positive attitudes towards English in general and New Zealand English (NZE) and culture in particular, due to its perceived usefulness as well as its status in the world.

The LVC part of the study used mixed effects logistic regression modelling to analyse the influence of different factors on the production of ING and intervocalic /t/. The study identified three possible variants for the ING variable in the Jordanians and Palestinians speech of English [ɪŋ], [ɪn] and [ɪŋɡ]. The variable ING patterns were similar to NZE in the production of the younger generations (2nd and 1.5). However, the older generation (1st generation) showed a mixture between prototypical NZE patterns and typical Jordanian Arabic realisations. The 1st generation was likely not to have completely acquired the NZE variants of the variable ING. Female speakers were more likely to produce more native-like ING features than males. Length of residence was also significant, with those who have been living in New Zealand between 11 to 20 years producing more NZE variants than other groups. Occupation also played a role, with ‘in-work’ speakers using more NZE variants than ‘not in-work’ speakers. The analysis showed that there was a significant positive correlation between Principal Component (PC2) (attitudes toward English) and the production of the NZE variants of ING. The results are discussed in light of positive vs negative attitudes, instrumental vs integrative attitudes and identity.

For intervocalic /t/ three possible variants were identified across Jordanian and Palestinian speakers in Christchurch (CANONICAL /t/, FLAP and GLOTTAL STOP) and social factors were found to play a significant role. For example 2nd generation participants produced the most FLAP and GLOTTAL STOP realisations, those who have been living in the country from 11-20 and 21-30 years were found to produce the most FLAP, while 1st generation participants and those who have been living in NZ from 1-10 years produced significantly more CANONICAL /t/. Attitudes were not found to have any significant bearing on the production of the intervocalic /t/ variants.

The other three linguistic variables investigated in the thesis were the three NZE short front vowels KIT, DRESS and TRAP, where I tried to determine if the speakers had adopted the shift happening in these three NZE vowels and investigated the effect of social variables such as generation, gender, word frequency and attitude on the speaker’s production of these vowels. A mixed effects model was used to analyse the influence of these factors on the vowels. The results showed that the social factors: generation, attitudes, word frequency and gender were significant factors affecting Jordanian speakers’ production of the three NZE short front vowels. Significant differences were found for DRESS F1, TRAP F1, TRAP F2, KIT F1 and KIT F2. The results also provide evidence for vowel shift in L2 speakers for the three NZE short vowels (KIT, DRESS and TRAP), particularly among 1st and 1.5 generations more than the 2nd generations. Attitudes were significant with DRESS F1, TRAP F2 and KIT F2 and discussed deeply in the thesis. Finally, the qualitative attitudinal results in the interviews offered some explanations for the consonant and the vowel results and matched them to the linguistic behavior (production of the vowels and consonants).

Overall, the results provided evidence that attitudes can link both LMLS and LVC and that the quantitative attitudinal results from the questionnaire likely match with the qualitative attitudinal results from the interviews and all are likely to predict linguistic behavior. The findings also suggest that the role attitudes play in LMLS and LVC can be very complex.