Are there any rules of thumb for number of "steps" between two points in a language's natural development?

Question

Let's suppose that you want to create Conlang B. You want this conlang to be derived from Conlang A. Let's also suppose that you have some arbitrary period of time in between A and B. Are there any rules of thumb for how many steps—how many major grammar/vocabulary alterations, etc.—would generally manifest between A and B as a general guideline? Is it entirely up to the conlanger's prerogative, or can patterns from natural languages be found?

Answer 1

It’s relatively hard to say. First, some general numbers:

Over 1000 years, Old High German split into the various southern German dialects, including Swiss German and Standard German. While differences are very noticable, many developments were parallel and speakers usually don’t take long to get accustomed to the differences. A similar time depth can be claimed for English vs Scots. Old High German itself is not readily intelligible to the modern speaker, but many words are recognizable.

2000 years ago, we had Latin and Proto-Germanic, the ancestors of the Romance and Germanic branches of Indo-European. While the modern languages all show clear similarities in most regards of the language, they are as a general rule not mutually intelligible and speakers need to invest a lot of time to learn another of those languages - but still much less than with an unrelated language. Proto-Germanic and Latin are both quite different than the modern languages, showing vast differences in morphology and syntax, as well as having undergone many sound changes since. Nevertheless, with some knowledge on etymology, words can readily be recognized for what they are in a written text.

About 4000-6000 years ago, we have PIE. At this point, the language is no longer immediately recognizable as being related to a modern European language. The grammar is very different - just as an example, PIE verbs inflect for three persons and three numbers, but not directly for tense or aspect, which are instead marked with various more-or-less productive derivational affixes. The phonology would also look very out of place in modern day Europe. And for most roots, even if they didn’t change much phonologically, there was probably a semantic shift somewhere between PIE and English. The average English speaker will understand exactly zero words in a passage in Serbian or Punjabi.

---

But there’s another side to this as well. General tendencies are fine, but what about individual languages? If you compare Italian and French to Latin, then it’s plainly obvious that French has changed more, at least with regards to phonology and morphology. Similarly, comparing English to Icelandic, the latter is clearly more conservative (though no where near as much as some people might make you believe). To me, French and English look like they are 500+ years further in the future than their contemporary siblings — in reality of course they just underwent some more drastical sound changes that in turn messed with their morphology, which in turn forced changes to the syntax. People try to ascribe those changes to various factors, usually external ones such as a large number of foreign learners, but most of this is not grounded in actual science and just speculation.

As such, no definitive answer can be given. I find the time depths I gave above to be good rules of thumb, but single languages may deviate from those quite a bit. To achieve such results, it’s important to apply changes not only to the phonology, but to all parts of the language — especially the lexicon.

Answer 2

As there is no one simple rule of thumb (or generally applicable rule at all), over history these have been messed up. If it would be practically doable, many linguists would want to rename those periods (but many intermediate forms would be lost because of undecided diachronics).

I'd recommend:

• Languages never written (fully reconstructed) usually span a long period of time (PIE spans 2000 years from 4500BCE to 2500BCE). If you'd follow the rule of "many sound changes", we'd be able to split PIE into 10 or more language segments because of sporadic sound changes everywhere.

• Usually before a big separation of languages. Proto-Germanic separated into many languages quite fast (for a hypothetical one, at least). Middle Dutch was a nontrivial influence to both English and French, including my favourite words: bodge and feague.

• In our sample of the PIE western world, we see a recurring theme: PIE > sub > old > middle > modern. English, Dutch, German fit this perfectly, and with some fiddling (Counting Latin and proto-Italic as one sub, for no other proto-Italic still exists and Latin was quite early (Proto-Germanic even hypothetically borrowed some words)) the Romance languages. This theme can be also applied to Russian, with some bashing and bodging even the Chinese languages, if I recall correctly.

• Big spelling changes, not by a regulation's suggestions (or laws—think France) but by linguistic fashion changes.

• When a language sees itself as 'newer' and 'better' than its predecessor.

• Just to fit. Calling Old and Middle English 'Old English' collectively would be 'weird' as too many changes have occurred for it to be one language. So they drew a line where they thought it would be applicable.

I don't think it can be reduced to one single rule of thumb, like in jknappen's answer. Some sound changes affect many words, but don't split the language. For example: when -e's at the end of words became silent, many words changed pronunciation (many underused words have an -e, which are less common but more numerous). Such 'common exceptions' make it difficult to have one single rule.