Vord length distribution

[MarcoP]

Hi Anton, thank you for your comments!
I am not an expert, so my answers may not be very reliable. Anyway, here they are.

In the context of this discussion, when we speak of "syllables" in a language, do we mean phonetic syllables or their written representation? I have an impression that these two are messed, which is probably methodologically incorrect.

The graphs and comments in my previous post were based on the written representation of syllables. As you say, this is most likely incorrect. Sadly, I am not knowledgeable enough to do any better. I am sorry. 

How can one syllable contain two (or more) vowels? In your example of "thuong" (I omit the modifiers), would not that be two syllables (thu + ong) instead of one?

As far as I know, in English and Italian consecutive vowels are not split into different syllables. See also what wikipedia says about You are not allowed to view links. Register or Login to view..
For instance, Shakespeare wrote 10-syllables verses and "doubt" (which includes two vocalic sounds) counts as a single syllable:

I doubt whether their legs be worth the sums
1    1     2      1    1    1    1    1   1

I guess it's the same in Vietnamese.

How can Vietnamese contain only single-syllable words? Means all their words are very short?

As can be seen from Stolfi's distribution curves, there are no long words in Vietnamese. The longest words I find in Stolfi's VIQR file are 9 characters long (e.g. "tru+o+`ng").

Then how do they manage to express complex notions like "hydraulics" or "referendum"?

According to this online dictionary, complex notions are expressed by multiple words:

hydraulics - động thủy học
referendum - trưng cầu ý dân

In short, when it comes to syllables, I'm at a loss. 

Yes, me too! I will leave further research to linguists!