Accent in daughter languages of my main protolanguage

[MysteryMan23]

So, I'm working on the main proto-language of my world. It's basically an ersatz of Proto-Indo-European, right down to spawning my world's biggest language family by number of speakers.

Now, the daughter languages of PIE seem, at least to me, to have a wide variety of accent systems. You have Proto-Germanic, which developed a predictable stress accent on the first syllable, Proto-Balto-Slavic, which created a contrast of pitches where there wasn't any originally, and other languages kept the system pretty much intact.

What I want to know is, what kind of accent system would my main proto-language have to have in order to explain this variety of results? And how exactly do accents tend to change over time, anyways?

[Mâq Lar]

AFAIK, pretty much any system can change into pretty much other, much the same as other sound changes. For example, a language could start with a stress accent, where the intensity and volume are the markers of accent. There are probably also differences in length and pitch of the stressed vowel, but speakers don't key on those as the salient features when they listen. But in different areas, the pitch distinction could exaggerate over time, and the difference in intensity or length could reduce, until the speakers see pitch as the primary correlate of stress and you have a pitch accent. Long and short vowels could then have a range of different tone melodies. (Or think of Welsh - originally the stress accented syllable had high pitch, but then they decoupled and the stress accent moved while the pitch accent stayed on the final syllable). The stress could easily move around - it might start as rigidly predictable on one particular syllable, but then it might be attracted to heavy syllables, or long vowels, or tense vowels. It might start as unpredictable - lexically determined - but then regularise differently in differently daughter languages (eg always penult, always ante penult, always first syllable, always first heavy syllable, always last heavy syllable....). And don't forget there will be many cycles of these changes, which will be affected by other changes - eg if stress is always on the antepenult syllable, and a sound change deletes final vowels, will stress move to the new antepenult, or remain on the (now penult) same syllable? It could do both in different daughters.

I personally would start with a stress accent because it's the one I am most familiar with. I'd start with a predictable stress accent, then I'd play around with different phonetic correlates of stress (volume, intensity, pitch etc) and which is the most salient, and different ways stress could move to create the daughters.