3.4 Measures
3.4.5 Phonological Processing Measures
Phonological processing refers to “the use of phonological information (i.e. the sounds of one’s language) in processing written and oral language” (Wagner & Torgesen, 1987, p.192). Two mea- sures in the present study were categorised as phonological processing, including the Spooner- isms subtest of the Phonological Assessment Battery, and the Rapid Digit Naming and Rapid Letter Naming subtests of the Comprehensive Test of Phonological Processing, discussed below in turn. Although the categorisation of RAN as phonological processing is contested (Wolf, Miller & Donnelly, 2001), the decision was made to retain it under this category, since RAN does involve processing of phonological information at some level (Bowey, 2005).
3.4.5.1
Phonological Assessment Battery (PhAB; Frederickson, Frith & Reason,
1997) – Spoonerisms
The Spoonerisms subtest of the PhAB measures children’s ability to substitute individual phonemes and syllables in real and pseudowords. After 3 initial practice items, Part 1 comprises a partial spoonerisms task in which examinees are asked to substitute individual phonemes in words, e.g. fun with a b gives bun (a score of 0 or 1). Part 2 requires substitution between two words, e.g. fed man gives med fan (a score of 0, 1 or 2 as 1 point is awarded if only one of the words is correct). Children are allowed a maximum of 20 seconds for each item and while each part has no discontinuation criteria, children over the age of 7 who do not score on part 1 do not progress to part 2. The PhAB is standardised on a population of 628 children in primary school Years 1 to 9 in the U.K. (age 6 to 14), with children learning EAL comprising 3.6% of this sample. The Spoonerisms subtest has a high internal consistency reliability coefficient of α = .93 for the 8;0 to 9;11 age band.
3.4.5.2
Comprehensive Test of Phonological Processing (CTOPP; Wagner, Torge-
sen & Rashotte, 1999): Rapid Digit Naming (RDN) and Rapid Letter Naming
(RLN)
The rapid naming subtests of the CTOPP assess speed of letter and digit naming. Examinees are presented initially with practice items and asked to read the numbers 2, 7, 4, 5, 3, 8 and letters a, t, s, k, c, n as quickly as possible. Each subtest consists of two parallel forms with a matrix of four rows of 9 items each, which examinees are asked to read aloud, from left to right, as quickly as possible. Testing discontinues if examinees are unable to name all of the items on the practice sheet and if more than four errors are made on either form; errors include incorrect pronunciation and omission of items. Upon skipping a line, the first item is scored as incorrect and the examinee is redirected to the appropriate place. Scores for each subtest are calculated by summing the time taken in seconds to name stimuli on both forms. The scaled scores of each subtest (with a mean of 10 and standard deviation of 3) may be summed to form a rapid naming composite score.
The CTOPP is normed on a sample of 1,656 individuals across thirty U.S. states aged be- tween 5 and 24 years, with rapid naming subtests being appropriate for those aged 7 years and over. Average content sampling coefficients across all age bands for RDN and RLN are .87 and .82, respectively. A subgroup of 30 individuals aged 7 to 18 years was assessed on two occa-
sions no more than two weeks apart. Test-retest coefficients of r = .80 and .72 for digit and letter naming, respectively, indicate an acceptable level of temporal reliability.
3.4.6
Literacy Measures
The decision was made to focus primarily on passage reading skills due to the age of the children in the study (Year 4; age 8-9 at t1) as well as the literacy demands of the national curriculum in this particular educational phase (i.e. a marked transition in focus from word- to passage-level reading skills).
3.4.6.1
Test of Word Reading Efficiency (TOWRE-II; Torgesen, Wagner & Rashotte,
1999)
The TOWRE-2 is a timed assessment in which examinees are asked to read aloud lists of real words (Sight-Word Efficiency) and non-words (Phonemic Decoding Efficiency) of increasing diffi- culty, for instance, is, up, cat, . . . morning, resolve, describe, . . . calculated, alternative, collective in sight-word reading, and ip, ga, ko, . . . prain, zint, bloot, . . . strotalanted, prilingdorfent, chun- fendilt in phonemic decoding. As such, the TOWRE is a measure both of the ability to recognise and produce real words varying in frequency and regularity of spelling, as well as to apply grapho- phonemic knowledge (Tarar, Meisinger & Dickens, 2015). Practice lists of 8 words are presented initially in order to encourage familiarity with the test format. Examinees are asked to read as many words as they can, as quickly as possible, within 45 seconds; after hesitations of up to 3 seconds, the examinee is prompted to continue onto the next item.
The TOWRE-2 is normed on a nationally representative U.S. population of 1,717 individuals aged between 6;00 and 24;11 years. The sample is representative of the U.S. school-aged pop- ulation in terms of gender, geographical distribution, parental education and income, Hispanic status, and prevalence of learning difficulties. The assessment consists of four parallel forms, A, B, C, and D, which are shown to have delayed test-retest reliability coefficients ranging from r = .89 to .93, and very high inter-scorer reliability coefficients of r = .99 for sight-word efficiency, phonemic decoding efficiency, and total word reading efficiency. Form A only was selected for use in the study.
3.4.6.2
York Assessment of Reading Comprehension Primary 2
ndEdition (YARC;
GL Assessment, 2011)
The YARC Primary is an individually-administered, graded reading comprehension assessment for primary school pupils aged 5 to 11 years. The assessment consists of an interactive Beginner passage, as well as parallel forms (A and B) of 6 test passages which examinees are asked to read aloud. Form A only was used in the present study. Passages are a mixture of fiction and non-fiction text and increase in difficulty in accordance with year groups and requirements of the national curriculum. The YARC yields raw, ability, and standard scores (mean=100, SD=15) for passage reading rate, accuracy, and comprehension. It is recommended that testing begin at the level appropriate for the child’s school year (e.g. Level 4 for Year 4), although it is advisable to begin on the previous level if testing takes place within the first half of the academic year. After
the beginner passage, an examinee then continues backwards if he or she scores 4 or fewer marks for comprehension or reaches near the limit of 20 reading accuracy errors, or forwards if either of these criteria is not met. Examinees progress through the assessment until they reach discontinuation criteria or complete the sixth and final passage. Scoring is calculated using the two highest, consecutively completed passages. Given that children in the present study were assessed at the beginning of Year 4, Level 3 was administered as a beginner passage.
At t1, children showed great variability in the passages they read, potentially creating difficul- ties for fair comparison between the monolingual and EAL groups over time. Thus, the decision was taken at subsequent time points to assess all children on the same two passages (2 and 3 at t2 and 3 and 4 at t3), then allowing them to continue forwards or backwards until discon- tinuation criteria were met or until they reached the final passage. Other than the choice of the minimum passages to be read at time points 1 and 2, the YARC was administered fully in accor- dance with instructions in the manual. Analyses in Chapter 4 will make reference both to YARC standard scores for rate, accuracy, and comprehension, as well as raw scores derived from pas- sage 3 alone. Passage 3 was chosen for the analysis of raw scores in order to allow within- and between-subjects comparison over time holding passage constant. As stated above, all children read this passage at t2 and t3, and the majority read it at t1 (n=75 out of 81 children).
During administration of the YARC, the examiner provides assistance for decoding difficulties but does not provide correct answers for comprehension questions. Questions tap both literal and inferential comprehension skill, and prompts such as ‘can you tell me more?’ are given for vague or incomplete answers. Generally, answers are considered correct if the gist matches exemplar responses in the answer key of the manual, however some answers require explicit mention of certain key words. Questions that require two-part responses are marked correct if and only if both parts of the response are given, as no half marks are permitted.
The YARC was designed and standardised using a nationally representative U.K. primary school population of 1,376 children, 14% of whom did not speak English as a first language. The test is designed to be gender- and culture-neutral. The Accuracy and Rate components of the YARC have the highest reliability, with correlations between parallel forms of .75 to .93 for accuracy and .91 to .95 for rate. The Comprehension component, on the other hand, has a lower range of reliability of between r = .48 and .77 across all levels and alternative forms. However, reliability calculated from pair passages (e.g. scores of performance on Level 1A and 2A, and so on) is higher, with coefficients between r = .71 and .84. Taken together with the varied nature of passage content and types of comprehension questions, then, there is evidence for good reliability and validity of the YARC Primary.
3.4.6.3
Writing: Bespoke Task
For ease of comparison with oral narrative language measures such as number of utterances and MLU, a bespoke writing measure was employed in which examinees were asked to produce a short piece of writing based on topic prompts such as ‘my favourite place’ and ‘my favourite thing to do’. Before writing, examinees were prompted by the examiner to think of examples, and then given a maximum of five minutes to write as much as possible. Examinees were prompted to pay attention to accuracy of spelling and grammar.
Similar to the Peter and the Cat oral narrative measure, children’s writing samples were tran- scribed manually into T-units, defined as a main clause plus its subordinate clauses (Bishop & Clarkson, 2002; Hunt, 1970). As well as total number of T-units, additional variables included the mean length of T-unit in words (MLTw), and the mean number of spelling and morphosyntactic er- rors per T-unit. Words were categorised as spelling errors simply if they did not conform to correct spelling, and as in analysis of oral narrative data, morphosyntactic errors included those in agree- ment (e.g. there was lots of exsiting [sic] things), past-tense morphology and over-regularisation (e.g. when I saw the Queen I taked a photo with her ), as well as omission of obligatory elements such as pronouns, prepositions, or determiners (e.g. I got Xbox, or however I wanted to go Japan).