What can be said about the evolution of syllable stress in related languages?

Question

Remembering a Czech song I once learnt I remembered a short Czech crash course I had and the teacher who said:

In Czech, stress is always on the first syllable.

This got me thinking and I remembered Russian, also a slavic language, which is well known for its unpredictable stress pattern and where stress could appear anywhere in a word (to the best of my knowledge). And then I remembered something else, looked it up, and yes, Wikipedia tells me that in Polish, stress is typically on the penultimate syllable.

These languages are all related to each other, i.e. stem from a common ancestor, yet seem to have distinctly different stress patterns.

By contrast, French typically stresses a words final syllable, Italian usually the penultimate although the final, third-last and fourth-last syllable may be stressed and Spanish seems to follow the same rule as Italian (again the latter according to Wikipedia). This all seems to indicate that their common ancestor (Latin) had stress towards the end of a word. However, the connection is based on the assumption that stress evolves similarly to phonology, i.e. small changes are fine but significant differences would need to be explained.

Thus:

1. Is it generally accepted that stress co-evolves with words or are the romance languages more the exception than the rule?^[1]

2. If the answer to the first question is yes, how is the seemingly tremendous divergence of the slavic languages explained?

---

[1]: Since there have been a few questions on what I mean by co-evolution of words and stress, here is my rationale:
Words can be thought to evolve, e.g. a common precursor *ph₂tḗr evolved into Latin pater, Sanskrit पितृ ‎(pitṛ) and Germanic *fadēr; the first further evolved into French père and Spanish/Italian padre while the last further evolved into German Vater, English father, Icelandic faðir and others. For a series of phonologically related words, the evolution is typically similar, as with ÜIE *p- that evolved into Germanic *f-. I can imagine stress following a similar set of evolutionary rules, e.g. to invent an example the stress of a two-syllable word ending in -er moving from the second to the first syllable. Co-evolution would assume that the stress evolution was influenced by a phonological evolution (e.g. the final -er syllable being reduced to shwa) and vice-versa in some way.

Answer 1

Pronunciation, including intonation, including stress, is a prime example of a potentially areal feature, rather than a feature that is losslessly passed down via language families.

You gave the example of Czech stress. Regular initial stress is a tendency in Europe in a contiguous area spanning Czechia, Slovakia, Hungary, and Vojvodina. (As you know, Hungarian is not a Slavic nor even an Indo-European language.) In Southern Serbia there is a transitional zone to Macedonian and Bulgarian, which are on the same dialect continuum but do not have initial stress. Similar is true in Eastern Slovakia and somewhere between Czech and Polish areas.

So stress pattern borders and the language and language family borders are different.

Why or how did it happen? There are two main mechanisms: substrate and superstrate. There are more recent and obvious examples of substrate effects on intonation: Mexican Spanish, Irish English, Indian English and West Indian English. I will not speculate here on the forces at play in Central Europe and among Slavic languages but, given the borders in question, reasonable cases can be made for both substrate and superstrate effects.

Answer 2

I don't know what it means for stress or words to co-evolve. Most languages have relatively regular rule-governed stress systems, but Indo-European languages have more than their share of lexicalized systems such as found in English and Russian. The accent system of Proto-Indo European was rather messy, with morphological alternations and lexical classes as one finds in Lithuanian and Russian (some people even think it was a tone language of sorts). Various sub-groups regularized the system, hence Latin developed a regular right-end oriented system, and late Germanic developed initial accent. English then messed with that system to create the messier system that we have.

The Romance languages tend towards a word-final position of stress because they derive from Latin and Latin had such a stress system. The Latin stress system was easy to learn, thus is was largely preserved in Romance, though the loss of vowel length distinctions and other factors resulted in some irregularity. The Indo-European system was hard to learn, and tended to get regularized in favor of a more transparent rule-based system, as it did in West Slavic. This is an instance of a general trend in historical change: lexically and grammatically arbitrary rules are harder to learn than regular phonological rules.