While I don't have access to the original research, I must say I am highly skeptical of the assertion that generational shift, i.e., "the language is changing," is a likely explanation for the use of the solecism "on accident" among the young. As I understand it, the conclusion is based on the observation that in one usage inventory, "on accident" was most commonly observed in the under-ten age group, least commonly observed in the over-35 age group, and the 10-to-35-year-olds showed an intermediate incidence of using "on" as opposed to "by." So that means that same-age English-speakers born after, say, 1995 are more likely to use "on accident" that those born before, say, 1970, right? Well…no. The assumption that generational shift is the underlying cause can only be made when comparing usage among speech populations that exhibit like stages of language development. In fact, cognitive development and language education are far more likely explanations for the observed difference, in my view.
Children under the age of ten have yet to enter the fourth and last stage of cognitive development. They really only communicate in informal register no matter what the venue of speech. From a psycholinguistic standpoint, grouping everyone from the age of ten to the age of 35 into a single affinity group makes no sense at all. It almost seems as if that age bracket was created by the researcher to promote a presupposed conclusion of generational shift. There is a lot of language development in the first half of that age bracket and very little in the second. Adolescent language users are being educated in the use of formal register which is much more highly codified than informal speech. During this period, they come to recognize that some of their habitual language patterns are nonstandard and they learn to adopt the grammar and conventions that formal register requires. Eventually they will learn to code-switch and will continue to use nonstandard utterances when informal register is acceptable. But during this critical development period there is also a lot of global modification of language patterns: modifications that have to be made for the sake of learning formal register often transfer, consciously or otherwise, to informal register as well and hence become permanent changes in habitual usage. So the Occam's razor explanation for the prevalence of "on accident" is simply youthful naïveté…it's essentially a pattern error and one likely to be erased by education. We'll know whether or not there's any generational shift in acceptability of "on accident" in another 20 years or so.
I was less interested in the quality of the research than the hosts' attitude: they seemed to say, "Well, it's a fact that this change is taking place...there's no right or wrong here...get used to it." As someone who routinely reads university students' essays, and who has noticed that all understanding of prepostitions seems to have disappeared, I thought it worth pointing out that there is something in the suggested paralleling of "on accident" with "on purpose" as an explanation of the noticed change. "X (the event in question) was done on purpose" properly opposes "X happened by accident". The use of "on" carries with it the idea of control by a subject--X was brought about by the subject's purpose or intention (something "internal" to the subject). We say "X happened by accident" to express the notion that something "external" brought about X; that is, the origin of the event is "outside" any subject. To connect the accidental nature of X's generation with "on" invites the idea that there is less difference between subject action and externally generated events; that is, human action and mere causal events are of the same type. The "on/by" distinction denies this. This could be a subtle recognition of the feeling that individual responsibility is less clear...that what happens--what started with me as actor--really was not my fault or responsibility (it was someone/something else's fault).
While talking with students I have been made very aware of how difficult it is to talk about prepositions (and what I have written above struggles with getting you to see what is there to be seen). This parallels other changes in language (what brings this about is a book-length enterprise), for example, the demise of the "may/can" (permission/ability) distinction. That one has to do with promoting the idea that what one can do (is able to) is that which one may do (is permitted). To express both notions by "can" invites this blurring. There just might be more to these shifts in language than the program discussion suggested.
This all leads to one of the most fundamental dichotomies in linguistics: prescriptivism vs. descriptivism. Education is, by its very nature, prescriptive. The aim is to develop pragmatic competence and artistic form in the use of language (primarily formal register). That requires the diagnosis of defective usage and a prescription for correction. Social linguists, on the other hand, tend to view language as an evolving organism that yields to selective pressure in the speech communities that use it. For them, usage is not something to be controlled but simply something to be described with an air of regulatory detachment. This view is perhaps particularly prominent in the case of English, which lacks any widely-recognized body to serve the function of, say, the Real Academia Española or the Académie Française in making authoritative prescriptions.
I don't have access to the original research
I think you should have stopped there. You might have avoided this ignorant comment: "It's essentially a pattern error and one likely to be erased by education."
lux rationis said:
While I don't have access to the original research, I must say I am highly skeptical of the assertion that generational shift, i.e., "the language is changing," is a likely explanation for the use of the solecism "on accident" among the young. As I understand it, the conclusion is based on the observation that in one usage inventory, "on accident" was most commonly observed in the under-ten age group, least commonly observed in the over-35 age group, and the 10-to-35-year-olds showed an intermediate incidence of using "on" as opposed to "by." So that means that same-age English-speakers born after, say, 1995 are more likely to use "on accident" that those born before, say, 1970, right? Well…no. The assumption that generational shift is the underlying cause can only be made when comparing usage among speech populations that exhibit like stages of language development. In fact, cognitive development and language education are far more likely explanations for the observed difference, in my view.
Children under the age of ten have yet to enter the fourth and last stage of cognitive development. They really only communicate in informal register no matter what the venue of speech. From a psycholinguistic standpoint, grouping everyone from the age of ten to the age of 35 into a single affinity group makes no sense at all. It almost seems as if that age bracket was created by the researcher to promote a presupposed conclusion of generational shift. There is a lot of language development in the first half of that age bracket and very little in the second. Adolescent language users are being educated in the use of formal register which is much more highly codified than informal speech. During this period, they come to recognize that some of their habitual language patterns are nonstandard and they learn to adopt the grammar and conventions that formal register requires. Eventually they will learn to code-switch and will continue to use nonstandard utterances when informal register is acceptable. But during this critical development period there is also a lot of global modification of language patterns: modifications that have to be made for the sake of learning formal register often transfer, consciously or otherwise, to informal register as well and hence become permanent changes in habitual usage. So the Occam's razor explanation for the prevalence of "on accident" is simply youthful naïveté…it's essentially a pattern error and one likely to be erased by education. We'll know whether or not there's any generational shift in acceptability of "on accident" in another 20 years or so.
First of all, how old is that research? A large number of the people I know in my age group (40's) have been using the combination "on accident" for quite some time. So, although this was noted as being used in the younger group, when exactly did they test this younger group? Could it be that this younger group is now in their 40's? You said that they would grow out of it, however since many of the people I know say this "on accident" which I realize is incorrect does that mean that we are all still going to grow out of a habit of speaking just because we know the truth of what is correct? I personally realize there are a lot of things we say that are incorrect but they become the normal way of speaking for people when everyone around them speaks that way too. For example, my children are now saying the words "Lawl" or "LOL" when they think something is funny rather than laughing. LOL stands for Laughs Out Loud (internet slang) and now is making its way into everyday speech.