The film is two years old now, and still riveting.

Two 6-year-old girls, faces slight, are rearranging furniture in a large dollhouse. They have short brown hair, little print dresses and eyes that scrunch with concentration as they examine each new piece of furniture. Their conversation, to the untrained ear, sounds like this:

"Genebene manita,"

"Nomemee."

"Eebedeebeda. Dis din qui naba."

"Neveda. Ca Baedabada."

"Ga."

The film was made in July 1977 at a San Diego hospital, and was the first recorded study of two animated twin sisters whose case has fascinated experts ever since the bewildered parents first brought the girls in for help.Grace and Virginia Kennedy, apparently healthy and energetic identical twins, spoke to each other in a rapid-fire language that nobody else understood.

It was not English.

It was not German, which was their mother's native language, and which their grandmother had spoken to them while caring for them during the day.

Somehow, in the extended privacy of a world without regular visitors, the sisters had made a language of their own - a "twin language," which occurs fairly often in very young twins, but rarely in children so old, and almost never to the exclusion of any other tongue. The Kennedy girls' only concession to English was an occasional request for certain items ("Want water," they would say, or "Want juice"), although it was obvious that they understood both English and German when someone spoke directly to them.

They called each other Poto, for Grace, and Camenga, for Virginia. They had never been to school, or played much with other children. Until they arrived two years ago at the San Diego Children's Hospital Speech, Hearing and Neurosensory Center, the girls had never been examined by speech experts; their father, then an unemployed accountant, had been referred to specialists only after he told the state unemployment office, in response to a routine question, that his daughters were not in school because they could not talk to others.

And when strangers spoke to them in their own language, after careful transcription of what seemed to be the twins' words, the girls looked utterly blank - "like we are crazy," one of their therapists said.

"They call me Camenga," Virginia Kennedy says, bending over a sheet of paper in her sister's bedroom to print the word.

Why do they call her that?

"Cause."

Because what?

"Poto," Ginny says, pointing at Grace.And that is all she will say about that. The twins know a photographer is coming to take their picture - "When is the camera man coming?" Grace keeps asking - but they do not seem to know why. Their language is leaving them, and if they understand their extraordinary history at all, they do not talk about it with strangers.

They are curious, charming and roaring with energy. They dive into their visitor's purse, pulling out pens and notebooks, asking what each thing is and whether they can keep it, fighting over the calculator which delights them.

"Where is your house," asks Grace. "I can see your house? I can see your house right now?"

She is told that the house is too far away to see right now, that it would be a very long drive. "I tell you when it's your house," Grace says firmly. "I sing all the way to your house."

At 8 1/2, the Kennedy twins are second graders in special San Diego public school speech handicap classes. They have been intentionally separated attending different schools and the language they now speak, both in school and to each other, seems to be mostly a simplified and very fast English - a sort of speeded-up pidgin. Tenses, conjugations and subordinate clauses - all the stubborn agonies of high school Latin and French - still give them trouble, and there is no way of knowing how quickly, if ever, they will pick up the more complicated nuances of standard English speech.

Breaking the Code

For the last two years, ever since the twins' much publicized arrival at the San Diego hospital, linguist and speech pathologists have been examining their private language in great detail, trying to understand where it came from and how it works. Three linguists at the University of California's San Diego campus have listened over and over to the videotapes of the twins, spending as much as an hour on each minute of tape, unscrambling vocabulary and charting syntax like the military analysts of some complex maritime code.

Was this an "invented language," in the most literal sense - the language whose vocabulary and sentence structure bore no resemblance to any known tongue - and if so, how had the words been formed? What were its vowels? Its consonants? Was it caused by some speech impediment, some inability to pronounce words in the people the twins heard talking?

The findings, so far, are tantalizing but inconclusive. To this day, nobody is certain about why, for example, a 6-year-old Grace Kennedy said, "Cabengo padem manibadi peetu," and her twin sister answered, "Doan nee bada tengkmatt." The explanation may rest with neurology or environment, or some apparently unprecedented combination of the two.

Grace and Virginia Kennedy were born in Columbus, the first children of a Georgia-born accountant named Thomas Kennedy and the German woman he had met while traveling in Munich. (Kennedy has three other children from a previous marriage.) They day after the twins were born, as Kennedy remembers it, Grace suddenly raised her head and stared at him as she lay in her hospital crib. She was having a convulsive seizure. Ginny had a similar seizure the following day.

"The pediatrician went into the brain area to see what was causing the seizures," Kennedy says. He says the doctor's tests showed a slight accumulation of fluid on the brain of each baby. The fluid was released and although the twins were given anti-convulsion drugs, they continued to have seizures off and on for the first six months of their lives.

Then at about six months the convulsions stopped in both girls. Neither the Kennedys nor their doctor understood exactly why, (and there is still no certainty what effect the seizures may have had) but Christine Kennedy says the doctor, very tentatively, pronounced both girls healthy. "He said they were too small to actually say they would come out normal," she says. "He said it would take all the way up their fifth or sixth year before they (the doctors) could see."

When the girls began to talk, they did what most children do. They rattled along, making noises that sounded like language, and they said "mommy" and "daddy" in distinct English. By the time they were six, when the family had moved to California, they were still doing precisely that. They stayed at home most days, cared for by a German-speaking grandmother who attended to their needs but apparently did not talk to them much. Kennedy says both he and his wife would spend their days out looking for work, and that when they came home and watched their daughters in animated but unintelligible conversation with each other, they simply did not know what to think.

"We had been cautioned that they might be mentally retarded, and we wouldn't know until they were six years old," he says. "We just thought it was a childhood thing between them."

It was not until the Children's Hospital therapists first talked about private languages, Kennedy says, that he began to think about what he had sometimes seem as the girls played together. Grace - the first born, by five minutes, and the more dominant of the two - would say a word to Ginny while holding up an object, as though naming it.

"Our best guess," says Chris Hagen, chairman of the speech pathology department that took on the Kennedy twins' therapy, "would be that it had something to do with the communicative environment."

"Or the lack thereof," adds Donald Krebs, director of the speech, hearing and neurosensory center.

The Origins of Speech

There is still considerable mystery about the origins of speech. Many linguists believe, for example, that children who are deprived of human contact before puberty (like the classic "wolf children") may never learn to speak normally, although there is no consensus about why this should be so. Still, it seems likely that a child who learns to talk starts with a basic neural framework - the brain has to be in working order - and then, from the first afternoon when "Mama" or "cooky" sends the parents into ecstasy, some give-and-take system shapes the developing language. "Urglap" says the child, holding up a toothbrush, and the mother says, "No. Toothbrush."

Twins, and sometimes siblings who are close in age, commonly slow up this process by reinforcing each other's invented or mispronounced words. ("Urglap, Urglap!") If you put lively, identical twins in a household with limited outside conversation and correction, suggests Hagen, they might end up talking like Grace and Ginny Kennedy.

"Two human beings were there - Robinson Crusoe and Friday were there on that island - they did what was normal," Hagen says, stressing that this is only speculation. They rewarded each other. They were, in a sense, almost like the father and the mother and the child, all in one."

What the children apparently did, as far as speech pathologists and linguists can determine, was to latch onto the English and German sounds they heard spoken around them and reshape the familiar noises into words of their own. It took months of listening to the children's voices, sometimes replaying tapes again and again, before certain words began to make sense: "pinit" meant "finished," "gimba" meant "camper," "buda" meant "butter." The girls could pronounce words quite differently from one moment to the next, which made understanding them even harder; Richard Meier, a University of California, San Diego, psycholinguistics graduate student who worked on the language, recorded 26 different pronunciations of the twins' word for "potato" - ranging from "puhted" to "pandaydooz" - in one 15 minute videotape.

Linguistic Spaghetti

There were some words that took longer still. "Toolenis," for example, had them stumped for a long time. They had figured out by watching the children that the word meant "spaghetti," but they could not imagine why. Finally the pathologists asked Mrs. Kennedy where the word might have come from, and she brightened immediately. While cooking spaghetti, now and then, she had sung "O Sole Mio."

Their syntax, Meier says, was basically simple English - subject, verb, object - with a few striking exceptions. When the girls used the word "anmet," which seemed to be a distortion of the German word anmachen, meaning to fasten or to fix, they stuck the verb at the end of the sentence, German-style.

Some words remain untranslated. Meier is still not sure what "nunukid" means or "pulens." "Mis" and "nea" seem to be pronouns, but he is not certain about that, either.

Ginny and Grace still report regularly to the center for speech therapy, which is provided at no cost (although there was a recent lapse of several weeks because the Kennedys said they could not get the gas for the trip.)

The two speech-language pathologists who have worked with the twins for the last two years began with what Hagen calls "incidental learning" - talking to the twins in English as they played - and gradually began working more directly with the girls, describing each activity in English words and asking Ginny and Grace to repeat the English. The pathologists also used "stimulus cards" to work on vocabulary and sentence structure - one sequence of six cards, for example, shows a boy on a bicycle heading perilously close to a log, then rubbing his knees on the ground, then being ministered to by his mother.

Is there some concern now that the twins will revert to their private language, despite the best efforts of pathologists and teachers? Alex Romain, one of their pathologists, says no. She says their world has expanded "so much that even if the girls wanted to use the old language, which seems unlikely, they would not be able to find the vocabulary or structure they need.

"We had always hoped, actually, that they would retain it, and be bilinqual, or trilingual," Romani says (the girls understand German but do not speak it.) But only a few of the original words remain, and when the pathologists recently sat the twins down to watch one of the early videotapes, the girls showed no interest at all - "as if it's completely foreign to them," says Romain.

The most maddening part of the Kennedy twins' story is that they may never be able to explain it either. There is no way to tell whether Ginny and Grace will ever remember the sound or the secrets of the private language - or whether they have any idea, right now, about why these large people fell over themselves just to hear twins converse.

When Ginny picks up the toy telephone with a stranger watching, all she says is "Poto and Catenga are twins."

But her eyes are so alert, so full of mischief, that it seems for an instant as though she must know. CAPTION: Picture 1, The Kennedy twins, by Stephen Kelley, for The Washington Post; Picture 2, Mr. and Mrs. Thomas Kennedy with twin daughters, by Stephen Kelley, for The Washington Post