If memory serves, at that point we had a semi-custom Z80 (actually it was a three processor system) S-100 bus based system running TurboDOS (at a blazing 8MHz per processor), an 8086+8087 off on daughter box connected to the main Z80 processor (running MS-DOS 1.x, for some value of x that I forget) fronted by a televido terminal with a switch (so I could speak to any of the Z80's directly, the 8086/7 was only available via software from the lead Z80) with a pair of Epson MX-80s being tortured into destruction (typesetting math documents using "Fancy Font" which was essentially a troff style derivative) and a MacPlus with a LaswerWriter.
I'm not sure how using those things would be better than an iPad for a child. Just the raw metal bits on the S-100 system (not inherent, just my suboptimal metalworking skills) makes me cringe to think of it as a child's toy. Oh, and the DecWriterII that we used for printing labels. Heavy enough to cause serious damage if toppled...
We did have a variety of language processors (but who really wants to teach a 2 year old how to program in Pascal, Fortran (any dialect, but especially the mutant that was Microsoft's ... until we got Lahey's much better product). Indeed, JRT systems Pascal (the first $29.95 compiler I found) computed x*0.0 !=0.0*x (one returned 0, one x). So it wouldn't even be a good tool for teaching math (however, a great tool for teaching the proper distrust of blindly assuming computers are correct).
OK, not everyone programmed up Kalman filtering software at home and wrote memos and reports. But the technology was there, and affordable if you compared it with, say, renting all the time on a Univac (or CDC, no one who had a choice picked an IBM 360 family for its numerics, but we did have clients who had them, so the code had to be portable to those platforms as well ...UTS on an Amdahl anyone?).
Personally, I prefer to have my kids playing with the iPad than, say, power tools.