Data collection

The current study will provide essential information on the acquisition order of phonological patterns that occur in the speech of children learning Kuwaiti-Arabic. The outcomes of this study will provide essential data on the

phonological development of children aged 1;4-3;7 and highlight the influence of the ambient language on the developmental patterns. What is special about Kuwaiti-Arabic is that it has a rich consonant inventory which combines the consonants found in MSA with dialectal variants and other sounds that result from the influence of loanwords. For example, Kuwaiti-Arabic shares the /ɡ/ and /ʧ/ phoneme with English. The similarities and differences between Arabic and English will form solid grounds needed for exploring the development of universal and language-specific patterns.

4.3. Data collection

4.3.1. Participants

The data collection took place in Kuwait. A total of 70 children, in groups of ten, were sampled from the general Kuwaiti population. Residents of Kuwait

originating from neighbouring countries were excluded. The subjects were randomly selected from variable ethnic backgrounds and social classes, concentrating on inner city population. The age groups were selected as a continuum to Ayyad’s (2011) study (age 4;0-5;0), which was in progress at the

112

time when this study took place. The sample was divided into seven gender-balanced subgroups with the following age ranges: 1;4-1;7, 1;8-1;11, 2;0-2;3, 2;4-2;7, 2;8-2:11, 3;0-3;3 and 3;4-3;7. According to Watson and Scukanec (1997), the six-month increments may not be specific enough to capture the phonological growth that takes place in children younger than three years;

therefore, three-month age increments were used in the current study.

4.3.2. Subjects recruitment

Informed written consent was obtained from directors of randomly selected Arabic speaking childcare centres in Kuwait. Information packages were sent out to parents of children attending the childcare centre. Parents interested in participating in the project were contacted by phone to arrange a meeting and answer further queries. Parents who took part in this research were asked to refer other families who may be interested to take part. Referred families were contacted over the phone by the researcher who provided further information and explained the aims of this research. Subject recruitment limitations are discussed in section 4.6 of the current chapter.

Written consent was obtained prior to the planned recording session from the child’s parent or legal guardian for both video and audio recordings. All parents were provided with hearing screen test results and short debriefing meeting after the recording session. Case history information was collected concerning birth and developmental history, health, and social information to ensure the population was representative.

4.3.3. Elicitation procedure

Spontaneous speech sample was audio and video recorded on a single occasion. The parent was instructed to interact spontaneously with his or her child for 30 minutes. A set to rubber toys and picture books were made available during all recording sessions. The parent/child spontaneous

interaction was digitally recorded with an Edirol R-09HR Handheld SD Recorder and a Shure PG14/PG185 Lavalier wireless microphone system attached to a custom made vest and connected wirelessly to the recorder. The vest was

113

made to hold the microphone transmitter on the child’s back and the lavalier microphone was attached approximately 10-15 centimetres below the child’s chin.

All children were offered a screening hearing assessment in the form of free-field audiogram to ensure adequate hearing necessary for the child’s speech development. The researcher used an Interacoustics PA5 Kamplex KS5 portable paediatric audiometer to perform the screening test in a quiet room prior to recording the speech sample. They each passed a 25- or 30-dB pure tone hearing screening test at 500, 1000, and 4000Hz.

A number of picture books and rubber toys (duck, fish, frog and turtle) were made available for the parents to use during the recording session. The parents were encouraged to use the provided toys and books when the child is not cooperative, especially near the end of the 30-minute session.

The recording sessions took place mostly in the family home, however, on several occasions where a quiet space was not available at the family place, the recordings were done in a quiet room kindly provided by Al-Khurafi Activity Kids Centre. Al-Khurafi Activity Kids is a charity run centre that provides

disabled children with the opportunity to learn through play and fun within a safe and enjoyable environment. The manager of the centre was approached and kindly offered a room within the premises to be used for the research purposes.

All data were exported from the digital recorder to a laptop, then onto an external hard disk for safe back up.

4.3.4. Maternal interviews and mothers’ role in data elicitation:

Naturalistic observational method

Parents were asked to play with the child as they normally would on a typical day. During the session parents were asked to elicit the words/utterances that they think the child already produces where possible. Parents and children in Kuwait often speak English. This practice is often socially acceptable in the

114

current Kuwaiti culture. However, the parents were informed that the study was on Arabic speech development and were encouraged to use Arabic as much as they can.

The children’s utterances were spontaneous. The type of interactions that were captured in the recordings were either free play with the parents, elicited speech while playing with toys, or elicited speech while viewing picture books.

The parents were asked to use picture books to elicit speech by asking the child to name pictures where possible.

It was emphasised that the parents should not produce the words themselves, otherwise the child may simply imitate, and for that the researcher proposed a few ideas on how to elicit these utterances in an indirect way. For example, the parent may point to a familiar item (toy or object) and ask the child to name it;

then ask the child simple questions to elaborate (e.g., what is this? Yes, it is a fish. Where does it live? What colour is it?). The researcher left the room if the child was being distracted by her presence.

4.3.5. Data preparation

The children’s emerging phonological inventories were documented while keeping track of their developmental patterns and individual differences. The data sessions were orthographically and phonetically transcribed before analysis. All recording were transcribed using the PHON computer program (Byrne et al., 2008). The audio recording was used primarily for the phonetic transcription while the video recording was used to clarify unintelligible audio recordings. The sound file was segmented into the different utterances produced by the child prior to actual transcription.

4.3.6. Transcription issues

The author who is a native speaker of KA transcribed the data sessions orthographically and phonetically using the International Phonetic Alphabet (IPA) revised 2005. The data was transcribed and analysed using a computer program PHON (Byrne et al., 2008). Seven samples (10%) were sent to an

115

experienced phonetician in Kuwait who is a native speaker of Arabic; the inter-rater reliability was 95% with the author’s transcriptions for consonants.

Reliability was considered only for consonants for this thesis; vowels were deferred for future studies.

4.3.7. Word Identification

Words were identified using the criteria proposed by Vihman and McCune (1994). The proposed criteria accounts for; the context in which the vocalisation may occur, the vocalisation shape and how it relates to the adult form; and its relation to other vocalisations.

For the context-based criteria, vocalisations were identified as words when their meanings were clearly identified in contexts or by the mother, or when the child used them more than once with similar phonological shapes across different uses. For example, when a child produced the word /mi.jaw/ ‘cat-sound’ while pointing to a picture of a cat, this vocalisation was counted as a word. However, when the child imitated a response of a verbal stimulus, the utterance is not considered as a word.

For the shape criteria, vocalisations were counted as words if more than two segments were matched to the adult form, or when the prosody of the vocalisation matched the adult target. For example, when the target word /s4l.ħʊ.ˈfaːt/ ‘turtle’ is realised as [f4ʊ.ˈhaːt] the word is counted despite the apparent syllable deletion and assimilation error.

Vocalisations were identified as words when vocalisations were instances of imitation produced with apparent understanding, when all instances of vocalisations shared the same phonological shape, or when all uses of vocalisations occurred in contexts that plausibly suggest the same word. For example, one child produced an unknown word /ˈɡuː.ma/ in five different

occasions during a 30-minute session, each time the word was produced it was understood and repeated by the mother. The video recording showed that the child was pointing to a bird-like toy each time she produced the word. The

116

vocalisations were considered a word as they had a consistent phonological shape in all five instances.

4.3.8. Language tagging

Kuwaiti Arabic speaking children tend to use both Arabic and English languages in their normal environment. A few children still showed preference for English words (e.g. dog, cat, car) in addition to character names (e.g. Superman, Spiderman, McQueen). For the purpose of data analysis all English names were identified as Arabic words as they do not have Arabic equivalents, and were produced by adults with Arabic-like phonetics. For example, Spiderman /spaɪ.d4ɹ.mæn/ often produced as [sbaj.deɾ.maːn] by Arabic speaking adults, because the voiceless /p/ does not occur in Arabic andis often realised as voiced [b]. All other words of English origins were tagged for selective analysis.

All words were included in the frequency analysis, whereas English words were excluded from the accuracy analysis. Onomatopoeic words such as /mi.jaw/

‘cat-sound’, /haw.haw/ ‘dog-sound’, /ʔɑm/ ‘food’ or ‘eat’, and /baːʕ/ ‘sheep-sound’ are commonly used in child directed speech, therefore such words all counted as Arabic words and were included in data analysis.

4.3.9. Geminate transcription

Geminates occur frequently in Arabic. The term geminate in phonology refers to a long or doubled consonant that contrasts phonemically with its shorter or singleton counterpart (Crystal, 2003). Geminate consonants are sometimes transcribed by (a) a sequence of two identical phonemes; (b) with single

consonant followed by the IPA length mark; or (c) with two identical consonants separated with a syllable boundary marker. In this study, the geminates were transcribed as a single consonant followed by the IPA length mark, and demonstrated in the following examples:

(a) (b) (c)

/sad.rejja/ /sad.re.jːa/ /sad.rej.ja/ ‘baby bib’

/ħ4.mmaːm/ /ħ4.mːaːm/ /ħ4m.maːm/ ‘bathroom’

/fi.ʧʧa/ /fi.ʧːa/ /fiʧ.ʧa/ ‘open it’+ masculine

117

The geminate contrast is commonly found in languages such as Arabic and Italian, as exemplified by the minimal pairs in (1) and (2), respectively

(1) Arabic geminate contrast:

/ħa.ˈmaːm/ ‘pigeons’

/ħa.ˈmːaːm/ ‘bathroom’

(2) Italian geminate contrast (Davis, 2011, p.837):

/fato/ ‘fate’

/fa.tːo/ ‘fact’

4.3.10. Tap and trill distinction in transcription

The /r/ phoneme production is variable in manner and place of articulation in different dialects of Arabic. Omar (1973) treated all Egyptian Arabic /r/ as taps, whereas Ayyad described all /r/s as trills in Kuwaiti Arabic. Saleh et al. (2013) examined speech samples of 50 adult speakers of Egyptian Arabic.

Spectographic analysis of the samples showed that taps and trills are often influenced by word context (e.g. position, adjacent vowel and consonants).

They found that /r/s are produced as trill in geminate environments, in word-medial syllable-final position with preceding open unrounded vowels (e.g.

/ba.ra/ ‘outside’) or followed by back vowel (/ba.ruːħ/ ‘I will go’). In the current study the distinction between the alveolar trill /r/ and tap /ɾ/ was based on the word context where it occurs. The alveolar trill was used for all /r/ geminates;

whereas the tap was used for all other occurrences. For example, in the minimal pair /ˈma.ra/ ‘once’ and /ˈma.ɾa/ ‘woman’, the first /r/ was treated as a geminate and the second as a tap.