I had an Iraqi roommate this summer. Let's call him Ali. His cultural story is interesting and comes with all sorts of other-than-linguistic things to muse about; he's very humble and is a sous-chef for one of the eateries on Grounds here. He was going to enter engineering school in Germany when the war broke out. He speaks broken German, more fluent Turkish and some Kurdish -- and very impressive English for two years of immersion.
Ali, trying to go back to school and enter the American college system what with its SATs and all, entreated me for help on reading, writing and speaking. I do think his progress in two years is very laudable -- some other members of the refugee community have been here for a year or more, yet cannot speak more than a smattering of English.
"My biggest problem is reading," he once said, struggling to pronounce that "-ng" sound. "English is a hard language to learn, harder than Turkish, German or Arabic. In Arabic, you spell the words like they are pronounced. Like this..." He writes out an example for me, which I struggle to read.
"English is kinda written as it is pronounced," I try to reassure him, struggling to summarise the vast body of work linguists and philologists have done over the years. "It's just that the pronunciations have changed over the years, and many spellings have yet to be updated. You know Arabic dialects, right? Aren't they spoken much differently than Standard Arabic is? It's kinda like that." It's like speaking those dialects but writing them in the system of formal Standard Arabic.
But I already knew there were complications with this analogy. Often, the colloquial dialects -- if they are written at all -- are transcribed exactly the way the sound. This was the case with the Vulgar Latin's descendants -- the Romance dialects, that diverged into the Romance languages like (Old) French, Italian and Spanish. (Modern French writing is an interesting story and shares some parallels with English.) With the Chinese languages, many characters are in fact, composites of other characters, where some characters have been borrowed for sound rather than meaning. After that, the borrowed character and the word being represented diverge in sound in some dialects -- and incongruence develops à la English. Often some of the diverging dialects will simply adopt different characters -- but elsewhere other dialects will keep the original (but incongruent) analogy; and of course the most conservative dialects do not even see an incongruence at all.
An immediate concept that helped Ali, but wasn't taught very often in ESL classes, was the concept of stress. At first this didn't have anything to do with reading or writing -- simply the way he pronounced the words that made him hard to understand.
"Sometimes, you say the words correctly... correct vowel and everything, but you don't put the correct stress on the word, so it sounds as though the vowel isn't there. And sometimes you say the stress correctly, but you use the full vowel on unstressed syllables or the reduced vowel on stressed syllables."
This was new to him, because in the two years of ESL help he had received, "stress" was a concept that had appeared to escape ESL teachers' minds. A fundamental concept in English phonology, it's kind of hard to teach both fluency and reading/writing if you ignore it. Stress, you see, is a historical innovation in the history of the English language. Unlike many other languages, it's neither predictable (i.e. it's phonemic) but neither is it explicitly marked in writing (as is the case with Spanish).
But don't worry -- stress is something easy to memorise, because the stress of each word is usually memorable. Many grammatical function words like "the", "a", "of", or affixes "-ing", "-er" etc. have no stress at all. Ali, I discovered, had already internalised many stresses -- he just didn't know it was that important. If you randomly quizzed him on the stresses of "bigger", "party", "telephone", "ambulance" or "reduce," he would get it right, down to the primary and secondary stresses. This is because stress is something that is quite easy to memorise and he had already done it unconsciously, even as a non-native speaker.
There are complications of stress, like the fact that the relative strength of a word's stress compared to other word's stresses will change depending on where it is in a sentence (syntactical stress), but that was for a more advanced lesson. Some words' syllables can be stressed when emphasised, such that when you quiz them in isolation, they become stressed, e.g. the "re-" in "reduce" or "the" gets pronounced with an "-ee" (/i/) sound rather than a reduced vowel -- but are usually unstressed when you use them in a sentence. But I told Ali not to worry for now -- it's not a big error for a second language learner to sound extra emphatic.
Stress came in handy, I discovered, because Ali had a real trouble with the letter "e". It can be pronounced like the "e" in "me" (/i/), the "ei" (/e/) in "weigh", the "e" (/ɛ/) in "bed", the reduced vowels /ə/ and /ɪ/ (when unstressed, I treated them like the same phoneme for Ali's sake, but some English dialects -- including mine -- do not merge them). And oh yeah, if you want to be really finicky, there's the stressed rhotic /ɝ/ (like in "rehearse") as opposed to the unstressed rhotic /ər/, but I didn't go there.
And oh yeah. That silent-e issue, which Ali had learnt functionally but still had trouble with because his ESL teachers had taught him the rule imperfectly. But wait, stress can come to the rescue...
Long vowels, when pronounced long, are almost always pronounced stressed. That is, you rarely see unstressed long vowels. Plus, conjugations of verbs and their gerunds tend to preserve the stress of their parent verbs.
"Biking" preserves the stress of "bike" and is not pronounced "bick King"
"Corner" had the stress on the first syllable, so the e's in "cornered" cannot be stressed, and also cannot be pronounced "corneared".
"Cornea" in contrast has two stresses (primary and secondary). It's a good example of a word with syntax-dependent stress, because it is pronounced with three syllables when the second syllable is stressed (during emphasis, like if a med school lecturer distinguishing the cornea versus the sclera), but the secondary stress can be dropped to form a two-syllable word.
This got us into diphthongs and glides (when vowels get compounded together and one of the vowels get reduced). When not emphasised, the /i/ in "cornea" is reduced to a glide (/j/; it behaves like a consonant), and consonants cannot be stressed in English.
There are also "exceptions" that really aren't. For example, I gave this stress-based rule: Stressed terminal -y is pronounced like the vowel in "tie" -- e.g. fly, fry, why, rely, retie, deny
Unstressed terminal -y is pronounced like the vowel in "free" -- e.g. belly, patty, hearty
The reason why terminal -y is pronounced /i/ even in unstressed positions (like "party") has a historical reason behind it, because terminal -y doesn't come from historical "e" but historical "i", which historically split into /ɪ/ (bit) and /aɪ/ (bite) during the Great Vowel Shift.
For various reasons, historical "i" in this unstressed terminal position didn't become /ɪ/ and remained /i/, while stressed "i" followed the sound change to "/aɪ/". It sounds like a mouthful to explain, but makes intuitive sense to children who already have an implicit knowledge of English phonotactics. Some vowels (vowel sounds) are never found by themselves at the end of words, or heck, open syllables (syllables that do not end in a consonant). English simply forbids them to exist (or more accurately, the historical pressure for their existence never ... existed). This includes vowels like /ɪ/, /ɛ/, /æ/ and /ʌ/. Try it for yourselves: the vowels in "bit", "bed", "bad" and "but" never end words by themselves.
The reason even comes down to universal trends seen in all languages: if there is length distinction involved, long vowels prefer to be in open situations (syllables that end in vowel), and short vowels prefer to be in "closed" situations (syllables that in a consonant). It's sort of a length compensation. After the historical Great Reorganisation of the English sound system, consonant length is no longer as noticeable, but can be noticed in the following examples:
The /t/ in the word "fitting" is likely to be pronounced longer than the /t/ in "fighting"
i.e. a transcription making note of this might note "fitting" as "fɪt.tɪŋ" (CVC.CVC) but "fighting" as "faɪ.tɪŋ" (CVV.CVC).
The /f/ in "riffle" is likely to be pronounced longer than the /f/ in "rifle", hence
/rɪf.fl/ and /raɪ.fl/
This gives us a historical basis for why double consonants (in terms of consonants found in writing) tend to convert long vowels in verbs to their short counterparts, hence the distinction between "writing" and "written". If you say "rifle" like "rife-fel" as opposed to "rye-fel", you won't technically be wrong, but you will sound slightly strange. These subtle details can be confirmed by taking spectrograms of native speakers' utterances.
At some point -- to distinguish letters from sounds -- I had to introduce Ali to IPA and the linguists' English vowel chart, for both our sakes. I'm surprised that it isn't used in schools -- because there are some overarching rules that follow from it that was a big "OH! it makes so much sense now" moment for Ali.
** At this point I should also mention various important laws of logic in analysing sound systems. Like causality. Situation X may imply sound type A, e.g. there is a correlation, but that doesn't mean sound type A implies situation X. Similarly, sound type B may imply situation Y, but the causality may or may not be bidirectional.