Last spring, Julia Kim’s students with disabilities at Fairmount Elementary in San Francisco were ready to take a new standardized test. They were excited that it had been built especially for them.

In past years, students with visual perception disorders had test questions read out loud. This time, the students sat in front of their computers awaiting the new technology designed to help them complete the test on their own for the first time.

But as soon as the first question appeared, students complained that the print was too small.

The color contrast tool, which used a background to minimize visual distortions, had been developed for the Common Core test to make it easier for special education students to see. But in practice, the tool prevented the one student in Kim’s class who used it from reading questions and marking answers. “I can’t see it,” he told Kim. It was too dark to read.

The Common Core tests, which are based on learning goals adopted in 43 states and the District of Columbia, offer many state-of-the-art technological tools to level the playing field for special education students. But Kim’s students were not alone. School employees across California have reported glitches in the tests’ enhancements for students with disabilities.

A field test administered in 2014 was meant to iron out the kinks. As a result, a noise buffer and closed captioning were added, according to an email sent last April on behalf of Michelle Center, who is now the California Department of Education’s director of the Assessment Development & Administration Division.

Still, according to teachers and administrators, special education students across California spent days last spring toiling over computerized tests that their teachers say often made it more difficult, not easier, for them to access the material.

“The majority of my students weren’t able to process any of the tests,” Kim said.

In San Francisco, one school found that text-to-speech tools read passages too quickly for students to follow, so teachers had to jump in and read the text out loud – distracting other students. The California School for the Blind found that different accessibility tools, such as Braille, could not be used at the same time as text-to-speech. In the Santa Ana Unified School District, curriculum specialist Gabriela Aguirre said she was concerned that the text-to-speech voice was distracting to students because it sounded robotic.

Precisely how many problems occurred with the tools known as accommodations last spring is not known. The California Department of Education didn’t specifically track accessibility glitches, Pam Slater, who worked as a CDE spokesperson until recently, said in an interview last fall.

Kim administered the exams to 14 students with disabilities in third through fifth grades. She and other teachers said they had problems with the accommodations. Those glitches only worsened anxiety about a test they had already worried was going to be especially difficult because of the tougher new standards.

Test scores for students in general, including those with disabilities, were low, as the state announced last fall. While there is no way to know what effect the lackluster accommodations might have had on the results, it’s clear that tools meant to help students with disabilities take tests just as effectively as their peers need a lot of improvement.

Of the more than 300,000 students with disabilities who took the tests in California, 88 percent did not meet achievement targets in English language arts and 91 percent did not meet targets in math, according to data on the state’s testing website. Fifty-two percent of California’s general education students failed to meet achievement targets on the exam in English language arts and 63 percent failed in math. Students with disabilities across the country similarly had lower scores than their peers on the new tests.

Slater said the California scores reflect a “starting point from which to make improvements in the coming years.”

There is no question that the tests are innovative. The new test promised technology unavailable to students with disabilities during the old paper-and-pencil exams. Instead of teachers reading the test aloud to students with disabilities, the new test had headphones and a dictation tool. And instead of an interpreter standing in front of the class to sign for students with hearing impairments, the new test provided videos of interpreters and closed captioning.

But for many, the upgrades were a letdown. The dictation tool used a robotic voice reading the text at a fast clip. In the videos, signs used by American Sign Language (ASL) interpreters were often indecipherable because the interpreters were wearing light-colored clothing or because they were using an ASL dialect unfamiliar to students. The text-to-speech tool did not work for students taking the test who were also using Braille or magnification. Students using Braille were constantly interrupted while waiting for Braille printouts of sections of the test, one teacher said.

Students with disabilities in Oregon and Washington had similar problems with accommodations, according to teacher surveys by the Washington and Oregon affiliates of the National Education Assn., a teachers union.

The Smarter Balanced Assessment Consortium’s executive director, Tony Alpert, stands behind his firm’s sign language videos. “Smarter Balanced used a very specific process to produce high-quality videos that use standard American Sign Language consistent with best practices and the audience of experts,” he said. He added that often the problems weren’t with the test itself. “In some cases, there are temporary local technology issues that may cause a problem with a test.”

The text-to-speech function, Slater said, will be corrected for this spring’s test so that students will be able to control the pacing. And they will have the option to choose a more human-sounding voice.

Also this spring, blind students who use Braille will have the option to take the test entirely on paper or in an online “fixed form” – a test in which everyone answers the same questions – according to Bill Ainsworth, a California Department of Education spokesman.

Tweaks to the test beyond the accessibility features also affected students with disabilities. For the first time, the exams use a technology called computer adaptive testing to tailor the difficulty of questions for each student based on his or her previous response. The tests can assess material at two grade levels below or above the student’s actual grade, according to Kelli Gauthier. Gauthier is the director of policy analysis and communications for the Smarter Balanced Assessment Consortium, which developed the Common Core exam taken by students in California and 17 other states.

That the questions get harder as students answer correctly is a “double-edged sword for students with disabilities,” said Matthew Navo, the superintendent of Sanger Unified School District in California’s Central Valley. “Students with disabilities deal their whole lives with trying to overcome life’s barriers,” Navo said. “To give them an assessment that gets more difficult as they get better, they tap out sooner.”

Stephanie Herlich, an assistive technology specialist at the California School for the Blind in Fremont, Calif., said that she would have liked to see an even greater span in the grade levels tested. “We’ve had tears,” she said. “We’ve had kids turn off that didn’t want to do it anymore.”

Maureen O’Leary Burness, who served until recently as the co-executive director of the Statewide Special Education Task Force, said last spring’s test might not have yielded accurate scores. In its first year, she said, it was “a test of the test.” But going forward, she added, “the big picture is we really do need to be accountable to all students in California.”

This story was produced by The Hechinger Report, a nonprofit, independent news organization focused on inequality and innovation in education.