This is the third of four categories explaining why translations differ.
3) How does the translator resolve ambiguities on the word-level?
Hebrew writing did not indicate doubled letters (which are significant) or vowels until the 8th/9th century AD*, when Jews who had memorized the pronunciation of the traditional text came up with a system (three, actually) of indicating the pronunciation in the text with marks above, below, and inside the consonantal text. That, again, is a thousand-year gap. Scholars vary in how much weight to give the vowel-pointing (niqqudot, or just "pointing"), but at times, greater sense can be made of a text by repointing a word or two. If we have GDSNWHR in God's appearance to Moses, and the tradition said "GoD iS NoWHeRe" we might consider that a bit odd for a believing Israelite to say, particularly as God was there appearing to him. A scholar might repoint or redivide as "GoD iS NoW HeRe" since it fits the context better. Just as BT in English could give us BuTT, BiT, BaT, ByTe, BuT or BeT, many words vary only in their pointing. (For an LDS example, Don Parry's BYU Studies article "Temple Worship and a Possible Prayer Circle in Psalm 24" repoints dor "generation" as dur "circle." He also discusses the history of pointing.)
One of the more common (and complex) examples involves debating whether lo "to him" or lo' "not" is the correct reading. There's also the difficult case of lo' meaning "no, not" (originally l?' ) and lo' meaning "indeed, certainly" (originally lu'), since they look exactly the same in Biblical Hebrew (references here).
Here's what a text looks like without pointing, although technically, there should be no dot over the right-most branch of the pitchfork letter. Sin (?) and shin (š) are distinguished only by the location of the dot.
Same text WITH pointing.
Same text with the pointing AND accents indicating how to "sing" the text (which is what the "cantor" does in a modern synagogue).
Let's assume the traditional pointing is largely accurate (as it probably is in most cases), and raise another issue. We look to a word's usage to determine meaning. The more usages we find, the more contexts we have to help us. However, the Old Testament does not have a lot of words (<7000, many of them related to each other), and many words have multiple meanings. Add to the small sample size the fact that usage (and therefore meaning) shifts over time, and sometimes it's quite difficult to know just what a word means in a given passage. We can't willy-nilly assume a legal or technical usage will have the exact same meaning in a narrative usage hundreds of years later. If a word is rare, we then turn to the versions as well as comparative Semitics; does a similar word exist in (to pick a few) Aramaic, Ugaritic, Arabic or Assyrian/Babylonian? Do the usages there shed any light on its usage in the Old Testament? (I've written before about some Ugaritic examples, and this Ensign article from 1983 gives some as well.) Poetic texts tend to use rarer vocabulary and use it in less concrete ways. Job is arguably the most difficult text in the Hebrew Bible, with a high concentration of hapax legomena (words that occur only once and nowhere else) and many other rare words. The bottom line is, even with years of scholarship, the versions, excellent lexicons, Jewish tradition, etc., sometimes we just don't know what a word really means in a particular case. For Jews, that has serious implications. When you read through the list of animals that aren't kosher in Leviticus 11, look down at the bottom of the page of the Jewish Study Bible and note that we're really not sure what birds these refer to. But again, translations have to say something, and good scholarship recognizes and admits its own uncertainty.
At both the word level and higher, the structure of Hebrew lends itself to ambiguity, multiple meanings, puns, allusions, etc. It's great for poetry, but this also makes it infuriatingly difficult at times both to understand and to translate. I recall one of my professors saying that every word has at least four meanings- the primary meaning, its opposite, something having to do with sex, and something having to do with camels. He exaggerated, but not much. Arabic is even worse than Hebrew in this regard, and though my Arabic poetry exam suffered a bloodbath of red ink, somehow I still had an A.
* Hebrew originally began being purely consonantal, being based on a rebus principle. Then, the semi-vowels y and w (and perhaps h) began to be used to indicate some long vowels at the end of a word. Yet later, but still in the pre-NT period, they were used to indicate long vowels inside a word. For example, David's name, daw?d in pre-exilic texts is written DWD (w being consonantal there), but in post-exilic texts DWYD. The DSS continue this trend, and began using them sporadically to represent certain short vowels, but we don't find that usage in the MT, to my knowledge.