Integration of placenames in English

We constantly use words that come historically from other languages, of course, and those words have varying degrees of connection to the source language: cartouche screams 'hi, I'm French!' even though it's been around in English since at least 1548 (no doubt longer), while more often with even new borrowings we're not so aware of their foreignness. Aside from not being terribly common vocabulary for a lot of us and maybe being spelled a little funny, cartouche is striking for keeping final stress (we'd expect stress on the first syllable) and having a long or tense vowel followed by the 'sh' sound at the end of the word.

How we integrate loanwords phonologically is an interesting and important question — a burning one in phonological theory these days, in fact — but there's been relatively less attention to partial integration: We've integrated jalapeño into English to a fair extent, and most folks have [a] in taco, but English speakers almost never produce a dental or unaspirated /t/ at the beginning. Lots of people will produce an [x] in the name Bach, a sound foreign to Modern English. And if you're a little on the pedantic side (or wait tables in a trendy restaurant), you can getting away with nasalized vowels (instead of nasalized vowel plus nasal) in French words. But unless you're actually switching into German or French, I imagine you don't use a uvular /r/ in words from either language. But I don't know of rigorous ways of getting at which features are retained or not in particular cases.

This morning on NPR, Linda Gradstein did a report on the West Bank since the Six-Day War. She was reporting from a very famous place (pictured above), called in English Shiloh. This word is extremely well known in American English, from the Bible but also as a famous Civil War battle site, placename, first name, etc. According to Merriam-Webster, the only American pronunciation is with [ai] in the first syllable. And of course they do have an entry for the Middle Eastern placename. Same holds for answers.com.

Linda Gradstein was pronouncing this as [ʃílo] (basically 'SHE lo') repeatedly (but I think stressing both syllables in one instance). I thought, wow, I would have guessed this would have final stress in Hebrew. And if you're going to integrate the vowel ([i] instead of [ai]), I figured surely you'd keep the stress pattern. But then she talked to a settler living there who in fact did pronounce it with final stress. Where is the asymmetry here? Why would you move only half-way to the (apparent) Hebrew pronunciation? Even if this established in English in Israel (where Gradstein has presumably lived for the many years she's been NPR reporter from there), how did this come to be? Is it a general pattern? What am I missing here?