Writing assessments have taken two primary forms in the past two decades: direct and indirect. Irrespective of type, either form needs to be anchored to making decisions in the classroom and predicting performance on high-stakes tests, particularly in a high-stakes environment with serious consequences. In this study, 11th-grade students were given a classroom assessment in which they had 1 minute to think and 3 minutes to write. Student work was scored for correct word sequence (CWS), total words written (TWW), and correct minus incorrect word sequence (CIWS). Students were also given a high-stakes state test to determine eligibility for graduation. This study focuses on the relation between performance on the classroom assessment and the state tests, with comparisons made between the performance of students receiving special education services (SPED) and students in general education. In an age of accountability, test validity has become an increasingly complicated topic. The social consequences of assessments impact students and their educational experience.

Many students struggle on writing tasks with little success because writing is a complex task. Students with learning disabilities (LD) and students who are served in English to Speakers of Other Language (ESOL) programs generally perform at lower rates on writing tasks than their English-only speaking peers without disabilities. Several researchers indicate that students with disabilities may be able to improve their performance on writing tasks through the implementation of Direct Instruction writing programs. The purpose of the current study was to demonstrate the relationship between the implementation of an accelerated Direct Instruction program and student writing performance with students who have LD in written expression, and who are either English-only speaking or native Spanish-speaking. Specifically, using a multiple-probe across participants single-subject research design, two groups of students received instruction using the Expressive Writing program. One group of three students were concurrently served in both special education for learning disabilities in the area of written expression and in a program for students who were English Language Learners (ELL), and another group of three were native English speakers who had learning disabilities in written expression. Students were divided into two separate groups to determine the effect of an abbreviated instructional sequence on both groups of students as the language background differences between the two groups did not allow them to be examined as one distinct group. The effects of instruction were measured by analyzing the number of correct word sequences, the number of words, and the types of errors when students were given three-minute writing probes. Additionally, performance on the Test of Written Language (3rd edition) (TOWL-3) and a classroom generalization measure were examined. Results indicated that when only half of the total lessons were presented to the students in both groups, the number of correct word sequences and the total number of words written increased on within-program writing probes, the TOWL-3, and on a generalization measure. An implication of these results is that adolescent students with writing deficits may be able improve their basic writing skills using half of the total program. Such a finding is important because students at this level who have not yet acquired these skills must quickly acquire them to be able to develop the more sophisticated skills required of students at their grade level. Limitations included the lack of the in-program placement test being proctored at the end of the intervention and the use of a nonconcurrent baseline with the second group of students.

Conventional language models estimate the probability that a word sequence within a chosen language will occur. By contrast, the purpose of our work is to estimate the probability that the word sequence belongs to the chosen language. The language of interest in our research is comprehensible well-formed English. We explain how conventional language models assume what we refer to as a degree of generalization, the extent to which a model generalizes from a given sequence. We explain why such an assumption may hinder estimation of the probability that a sequence belongs. We show that the probability that a word sequence belongs to a chosen language (represented by a given sequence) can be estimated by avoiding an assumed degree of generalization, and we introduce two methods for doing so: Minimal Number of Segments (MINS) and Segment Selection. We demonstrate that in some cases both MINS and Segment Selection perform better at distinguishing sequences that belong from those that do not than any other method we tested, including Good-Turing, interpolated modified Kneser-Ney, and the Sequence Memoizer.