Zero past tense verbs in -st?

A regular reader of this blog at the University of Wisconsin, one of the Wisconsin Englishes Project folks, passed along this chunk of an email from a student (with the student's permission, and shortened here):

I have had several people tell me that I say "taste" for the past tense of "to taste." So if someone asked me "How was breakfast this morning?" I might reply, "It taste really good." Today I noticed, for the first time, that I did the exact same thing with "to last." I said a sentence like, "It last about thirty minutes."

Some thoughts about what is going on:

1) On analogy to verbs ending in a "t" such as hit, shit, pet, let, I am simply using the infinitive form for present, past and participle.
2) I am losing the schwa part of the -ed past tense marker, and the cluster reduces to "taste."
3) Everyone in my dialect says this? I don't think I would even blink an eye if a professor said "it taste really good." Maybe I have been hearing this my whole life?

Now, the person who passed this on mentioned that he talked about the point briefly before class started and that other Wisconsin speakers reported that it didn't sound odd at all. I've asked a couple of folks from there who say it sounds pretty weird to them. So, (3) is probably right -- it's a dialect feature, albeit probably a variable one. (1) might play some role in promoting this -- though the past tense of to pet is petted .... beautiful evidence that this speaker truly has these zero past tense forms!), and on (2) the cluster reduction thing probably plays a role too, at least in getting this going, but it's apparently not limited to -st+ed forms, given that he treats to pet the way he does.

One interesting angle here is that for many dialects of English, coda cluster reduction appears less frequently when the cluster is carrying morphological information. So, I think Labov, Wolfram and others have reported that final clusters are more often simplified when they are monomorphemic (like find) rather than bimorphemic (like killed). The process for this speaker clearly doesn't show that kind of morphological sensitivity.

But we get more: this person and presumably others seem to have a verb system where at least some t/d-final verbs (pet) have moved into the class of verbs like not only hit, set, let, but also a set with -st: burst, cast, thrust. (I'd say thrusted, but I think thrust is the standard past form.) But note that this class appears to be expanding in the standard too: broadcast, telecast and forecast, narrowcast, simulcast and I assume to podcast. For these verbs the past is reported to vary between Ø and -ed. I would have figured that such new forms would inevitably be 'regular' (i.e., with -ed), like Marcus et al. argued in their classic 1995 piece in Cognitive Psychology – the way that we get flied out to center field, saber-tooths, Toronto Maple Leafs, etc. Looking back now at that article, though, I don't see that they particularly predict that new derivations in -cast would become regular. If that's right, it's really cool: This little opening for new (and pretty high-frequency) -st verbs with zero past forms might prime the pump for this extension of the pattern.

I know this little blog doesn't have many readers (yet?), but does anybody know more about this? Regional patterns in the Upper Midwest or elsewhere with this particular pattern of reduction? Ideas for how this fits into the bigger picture of English verb inflection?