Unique words/vocabulary - Sufficient or necessary condition to determine original source?

Question

Suppose we have two texts that are two different languages of the same story. One is the original, one is a translation. We wish to determine which one was the original. Among the more useful data points we can construct from these hypothetical documents is that Text A has vastly more unique words than Text B: 5,000-10,000 more unique words than Text B.

My intuition is that the text that has the larger vocabulary would be the original source: adding more specific words while translating seems disingenuous. The fact that Text A has so many more unique words leads me to believe that it's not impossible for Text B to be based off an earlier version of Text A that has not survived. That is to say, it's very hard to reconcile the number of unique words in Text A if it is merely a translation of Text B.

Question

Is a more sophisticated vocabulary a smoking gun to conclusively determine among literary fragments which is the source and which is a translation?

Methodology

We're not splitting hairs here. Assume we can all agree on what constitutes "unique words." Text A is said to have specific words that have their own roots, not declinations, plural forms, typos or other arbitrary things to inflate the unique word count. Text B is said to have a more impoverished vocabulary: it uses only general words, such as "animal" whereas Text A would have the genus and species for that part of the text.

Answer 1

You'll not be able to devise a real methodology without knowing the languages in question, because translation is not a mechanical process, and so you'll never get two identical translations of a long text. You might get a translator who employs a lot of flowery language not in the original. You might get a translator that tries, as best as possible, to stick to a single word to translate each unique word. You might get a translator who uses a lot of words to describe a single word that doesn't exist in the target language, and those small words can be more faithfully returned.

You also have idioms, and these really can break your hypothesis.

To get into the details, words don't map neatly onto concepts, and every language maps these words differently. In English, "I run the show" and "I run a race" use the same main verb, but in Latin these are completely different verbs.

But if you translate "I run the show; I run the race" into Latin and then back into English, you could also end up more words, taking into account the versatility of the Latin words.

Mapping this out with the above example, you can see how you can add more vocabulary in the translation than in the original.

Original English:
    "I run the show. I run a race."

Latin translation process
    "I run the show" is an idiom for "I am in charge". "I run a race" is more straightforward.

Latin translation
    "Imperium habet. Studium curro."

English translation:     "I have the power. I run a race."

--
Original unique word count: 6 (run x2, I x2, the rest once).
Translated unique word count: 7 (I x2, the rest once).

In the above scenario, the translated English gained a unique word, which isn't what your hypothesis expected.