Reading Rate
Students read at the “very slow” mean reading rate of approximately 57 words per minute. An earlier study by Lesgold, Resrick & Hammond (1985) investigated the reading rates of “phonics” and “non-phonics” taught groups of “normal” beginning readers in America. It was reported that both groups achieved a mean rate of 40 to 45 wpm by the end of Grade 1, although the “phonics” taught group took longer to achieve this. The “very slow” mean reading rate achieved by New Zealand students reading at band 0 in 2000 (44.46 wpm) equate with the younger American readers in the earlier study. These findings, however, do not support the sole use of interventions to increase reading rate in an attempt to bring about corresponding improvements in reading proficiency or comprehension. The positive relationships between reading rate and oral reading performance descriptors (expression, clarity, clause structure and sentence structure) imply instead that these variables are separate observable problems facing students with reading difficulties.

Error Rates
The error percentage rate was just over 9%, of which substitutions were the most common error type, followed by omissions and insertions. Frankel Tal & Siegal (1986) investigated errors made by “dyslexic” and “poor” grade 4-5 readers, compared with “younger normal” Grade 2-3 readers, in Canadian schools. The results of this study coincide with Frankel Tal & Siegal’s findings that nearly 50% of errors were vowel substitutions, followed by consonant substitutions, and deletion and insertion errors. Differences in the type of substitution errors, however, are worthy of further consideration. Vowel and consonant substitutions formed a substantial proportion of errors made by students in the earlier Canadian study, while whole word substitution errors were made only infrequently. For the purposes of this NEMP probe study, substitutions of a sound in a sound were classified as “sound substitutions”, regardless of whether a vowel, consonant or blend was involved. This sample of New Zealand primary school students reading below expectation achieved a mean substitution rate of 6.18% of words read (the mean error rate was 9.21%), but more words were substituted than sounds. It is probable, however, that this difference stems from the selection of reading material for the two studies. Lists of pseudowords were read in the Canadian study, whereas texts in the NEMP oral reading task words were presented within an authentic fiction, non-fiction or non-book context which provided an opportunity for students to use context-based cues when deciphering unfamiliar words. This difference reflects the greater emphasis placed on the use of context cues in New Zealand reading programmes and the use of authentic contexts in NEMP tasks.

Strategy Rates
Students paused to employ an observable strategy for fewer than 1 in every 10 text words read. Context-based strategies, especially “guessing” or “clarifying the meaning” of words, were the most common strategies used, followed by decoding (typically “sounding out” of the initial letter sound only). Coping strategies (typically an “emotional response” or “no/masked attempt”) and mixed strategies were used infrequently. The British study by Masterton, Laxon & Stuart (1992) found that “normal” beginning readers in their second year at school used a mixture of “word recognition” and “assembled phonology” strategies to read familiar and unfamiliar words, in a similar manner to “older skilled” readers. This NEMP probe study observed New Zealand students employing a wider range of strategies when pausing to decipher unfamiliar words. The tendency for the students in this sample to rely on contextual cues may reflect the greater emphasis placed on their use in New Zealand schools and/or a lack of ability in the use of phonological decoding skills. This view is supported by the strong tendency for students to “sound out” only the initial letter or letter blend of words on the occasions when they did attempt to sound out unfamiliar words. These findings may also reflect differences in teaching approaches across countries. Greaney’s 1992 New Zealand study found that “young normal” readers achieved better on rhyme awareness and analogy tasks than did “older poor” readers matched for word recognition achievement, but that both groups scored significantly better when a reading list of words was ordered by rhyme patterns. Greaney also found that teaching “older poor” students to use analogy-based cues was a more effective tutoring procedure than emphasising the use of context-based cues. Kirk (2001) found that the severe reading problems exhibited by the 6 severely reading-disabled students in her study resulted from difficulties in using accurate and complete letter-sound information and integrating this information with the use of contextual meaning to decipher words. The findings of both New Zealand studies support the view that students reading below expectation are deficient in the knowledge that allows them to effectively apply decoding strategies based on letter-sound information. Similarly, Frankel Tal & Siegal (1996) concluded that “disabled” readers, regardless of whether they were “dyslexic” or “poor” readers, lag behind “normal” readers in phonological decoding skills. Remedial reading programmes would benefit from inclusion of decoding strategies based on letter-sound information, especially for those students whose comprehension and fluency is compromised by very slow reading and frequent pausing.

Comprehension
Comprehension scores achieved by students in the NEMP probe study were satisfactory, given the majority of students (64%) answered at last 2 of 3 questions correctly. These results, however, reflect only literal comprehension as the answers were explicit in the text. Connelly, Johnston & Thompson (2001) used a more extensive assessment (Neale Analysis of Reading Ability-Revised, 1989) to measure recall of the main idea, sequence of events, other details, and some limited inference. It was found that “phonics” taught Scottish beginning readers (6 year olds) read more slowly than “non-phonics” taught New Zealand students matched for age and word recognition skills, but that they achieved higher comprehension scores and produced more contextually appropriate errors. Further comparative international reading research is needed to determine whether these differences also occur in older students.

Self-correction Rate
Students achieved a low self-correction rate of 2.46% when self-corrections were reported as a percentage of total text words read. This approach was used to avoid the confounding effect of measuring incidence as a proportion of errors. Thompson (1984) cautioned against too ready acceptance of the common interpretation that a high incidence of self-corrections is associated with high reading attainment. “An alternative interpretation is that self-corrections to some extent reflect incomplete processing that occurs with premature responding.” (Thompson, 1984, p. 53) Both data analysis methods resulted in low self-correction rates for this sample of students, suggesting that this strategy is not employed to any great degree by students reading below expectation. Whether the current teaching focus on increasing self-correction rates leads on to improved reading, however, deserves further consideration. Similarly to the oral reading performance descriptors, self-correction may offer an observable indicator of reading difficulties rather than a useful teaching focus to improve general reading proficiency or comprehension.

Oral Reading Fluency
Students generally spoke with “little/no” expression, while exhibiting “some” degree of control over clarity, clause structure, sentence structure and breathing. Young & Bowers (1995) studied the role of text phrasing skills of “poor” and “average” Grade 5 readers (mean age of 10.8 years) and found that “poor” readers were less fluent and expressive than “average” readers across all difficulty levels. It was indicated that parsing, the ability to reflect sentence structure through phrasing and inflexion, contributed to fluency in “average” but not in “poor” readers. Twenty-seven percent of students in this study exhibited “little/no” control over breathing or saliva flow, making noticeably loud noises as they gasped for air and/or sucked up excess fluids. The finding that more year 8 students spoke quietly may indicate that speech volume control and the awareness of appropriate speech volume levels for one-on-one interview settings increases with age and experience. Speech language therapy is a relatively specialised field of study and until recently many classroom teachers have accepted that oral language will develop naturally for most students through practice and maturation. Although the links between speech problems and subsequent reading, writing and spelling difficulties are recognised, more research and professional development for classroom teachers is needed to help students overcome this barrier to literacy.

Gender Differences
While boys tended to pause more often than girls to attempt to sound out unfamiliar words and exhibit less control over clause and sentence structure, statistically significant differences for gender were minimal. The findings of this study challenge common assertions that reading difficulties are primarily a gender issue. Students in this sample, regardless of gender, achieved similar achievement scores and brought a satisfactory range of attitudes and work habits to the learning task. Just as Eley (1999) warned that differences in reading achievement should not be interpreted as boys failing, the findings of this study warn against viewing reading difficulties as a gender, behavioural or attitudinal issue.

Text Type Differences
Students reading non-book texts employed significantly more strategies and exhibited less control over clause structure, sentence structure and breath control. It is possible that students lacked experience in reading this text type or that the fiction and non-fiction texts provided more recognizable contextual cues. As this material represents authentic text types that students encounter in community settings, students reading below expectation may benefit from initiatives to improve their reading of non-book texts.

The findings of this NEMP probe study provide insights into the challenges facing primary school students reading below normal expectation. Just as diagnostic assessment is a prerequisite to the successful implementation of programmes appropriate to the individual needs of students in classrooms, it is valuable to approach the issue of learning difficulties in a similar vein. Ongoing professional development and access to a wide range of research-based assessment and teaching resources would enable classroom teachers and literacy experts to more effectively meet the individual needs of their students.