It seems to me that in an ideal alternate history story, assuming that
it is set in an English-speaking country and the divergence point is
within the last 200 years or so, one ought to extrapolate the history
of English language change in this alternate timeline just as one does
the history of  political, social and technological change, and that
at least the dialogue if not the narration ought to reflect this
alternate chronolect.  That ought probably to mean reinventing any
words that have entered the language since the divergence point, and
being careful about words which existed before the divergence but have
changed their meaning or extended to new senses since then.

Of course, there are probably limits to readers' tolerance for this.
I throw out the 200 years figure as a ballpark estimate, as that seems
to be around the time limit where writers of historical fiction stop
trying to accurate write period dialogue because perfect historical
accuracy would be inaccessible  to too many readers; but perhaps with
alternate history the limit should be smaller, as we need to count not
only the years backward from now to the divergence, but the years
forward from the divergence to the time the story is set, to see how
far the language of the story (of its dialogue at least, and maybe its
first-person narration) is from our English.

Some amount of parallel development in lexical change is plausible, of
course, but not as plausible as parallel developments in science and
technology.  If we look at several timelines that diverged from ours
around 1930ish, it's plausible that many of them would develop
rocketry, atomic fission, antibiotics, birth control pills, computers,
and so forth in the next few decades, though not at the same time and
maybe not in the same order we did: the science is there waiting to be
discovered, and barring a collapse of civilization, it probably will
be.  It's less plausible that many of those timelines would coin the
same slang term "nuke" for an atomic bomb, or "box" for a
microcomputer.  It's plausible that many of the timelines diverging
around that period would put an end to racial segregation and
decriminalize homosexual sex, some sooner than our timline and some
slower; it's less plausible that they would all use the terms "African
American" and "gay."

Most alternate history writers pay attention to big lexical changes
like that, which are tied in to their alternate world's social and
technological history -- for instance the alternate terms for gay
people in Jo Walton's trilogy beginning with _Farthing_.  But it seems
to me that some of them overlook lexical changes in less obvious
areas, in thousands of everyday words that are have changed or
extended their meaning over the last fifty or hundred or two hundred
years and which most people don't know about.

I'm trying to get this right in my current novel in progress, which is
an alternate history that diverged from ours around 1915 or a little
earlier.  I've been looking for words of possibly or obviously recent
origin to check, and searching my corpus of Project Gutenberg etexts
to see if those words are attested (in the sense I want to use then)
in books from before my divergence point.  Searching for idioms is a
bit harder; I managed to verify that "pulling someone's leg" to refer
to telling a lie as a practical joke dates to before my divergence,
though it seems to have had a wider range of meanings in the early
20th century than it does now.  Figuring out whether a common word
already had its modern sense by 1915 is harder, involving searching
through lots of irrelevant hits.  I'm still not sure how to check
whether a recent syntactic construction existed before 1915.  And the
hardest thing of all is knowing what to look for -- I probably
unconsciously assume that many words have been around or had their
current sense for 100+ years which haven't, in fact.

Mary Robinette Kowal, in writing her Jane Austen pastiches, created a
spell-check dictionary consisting of all the words in the Austen
corpus; anything flagged by her spell-checker, then, she did
particular research on to verify its period accuracy, but that failed
to catch a number of words whose meanings have shifted over time.  I
plan to do the same by putting together a more selective corpus of
etexts dating from the mid-19th century through my divergence point.

Can anyone suggest syntactic changes in English over the last 100
years or so I need to watch out for?

That describes what an alternate history writer (in my view) ought to
avoid; what to do instead is harder still: to coin alternate terms for
most or all of the scientific discoveries and inventions and social
innovations occurring after the divergence, alternate slang including
new words and new senses for old words, and to use those in dialogue
and narration in such a way that their meanings are all obvious from
context; ideally to do the same with a few small changes in syntax.
It's small-scale conlanging similar to the various "Future English"
projects, with the constraint that it has to be immediately readable
for speakers of modern English without exposition or explanation.

Jim Henry