Remember HAL, the onboard computer in “2001: A Space Odyssey,” struggling to sing “Daisy, Daisy…” while astronaut Dave disables older and older memory modules until just one shred of HAL’s artificial intelligence remains?

Who could forget … except that’s not how language loss really happens to humans on the verge of Alzheimer’s disease, according to a team of psychologists and linguists with a paradigm-flipping test to predict the disease.

“It is now known that Alzheimer’s disease may develop for years, silently, before appearance of symptoms leading to clinical diagnosis,” explains Cornell’s Barbara Lust. “We’re searching for early signs in the spoken language of individuals, before Alzheimer’s is actually manifest.”

Together with research collaborators at Cornell and three other institutions, Lust published surprising findings in the April journal Brain & Language, “Reversing Ribot: Does regression hold in language of prodromal Alzheimer’s disease?”

Theodule Ribot was the 19th-century French psychologist who proposed a law of regression or reversion - essentially that “structures last formed are the first to degenerate ... the new perishes before the old.” Ribot’s law predicts what is often observed in Alzheimer’s patients, that recent memories may be lost before older memories.

The new study inquired whether there are also changes in language that occur during the prodromal course of Alzheimer’s disease – changes that possibly could be predictors of the disease. Prodromal refers to the period before appearance of initial symptoms and the full development of disease. The stage of prodromal Alzheimer’s, before dementia sets in, is called Mild Cognitive Impairment or MCI.

The researchers asked whether the course of language deterioration in prodromal Alzheimer’s would systematically reverse the course of acquisition of language among children, in accord with Ribot’s prediction.

The study compared earlier research – on the course of language development of complex sentences in children under age 5 – with new research assessing language patterns of MCI adults.

Researchers also tested young adults and healthy aging adults. Adults were asked to imitate sentences with complex structures, including various types of relative clauses, just as the children had.

As hypothesized by the researchers – but contrary to common belief – the linguistic structures children develop first are the ones MCI adult subjects struggle with the most.

For example, individuals with MCI found it more difficult to repeat a sentence like, “The office manager corrected what bothered the summer intern,” often giving responses like “The officer ... uh ... inspected ... and um ... corrected the intern.”

For children, sentences like “Fozzie Bear hugs what Kermit the Frog kisses” were the earliest produced. Whereas sentences like “Scooter grabs the candy which Fozzie Bear eats” were late-developed by children – but easiest for MCI adults.

In MCI, a first-developed structure is being lost first and a last-developed structure is being retained longest – contrary to Ribot’s prediction ... and HAL’s experience.

Next, researchers hope to incorporate linguistic assessments into potential predictive tests for early-stage Alzheimer’s, as they further test their results with additional subjects for verification.

Collaborating with Lust, a professor of human development in the College of Human Ecology, were Cornell senior research associate Charles R. Henderson, Jordan Whitlock ’11, Alex Immerman ’08, Aileen Costigan and James Gair, professor emeritus of linguistics. Collaborators from other institutions were Suzanne Flynn, M.A. ’80, Ph.D. ’83, Massachusetts Institute of Technology; Janet Cohen Sherman, Ph.D. ’83, and Sarah Mancuso, Massachusetts General Hospital; and Zhong Chen, Rochester Institute of Technology.

Research was funded, in part, by Hatch Grants and Federal Formula Funds, as well as grants from the Bronfenbrenner Center for Translational Research, Cornell Institute for Translational Research on Aging, and Cornell’s Institute for the Social Sciences.