Vowels are the flesh of words and consonants are their bones. Undoubtedly, the Spanish vowel system is much simpler than its English counterpart: the standard variety and most regional and social ones have only five phonemes, namely /i, e, a, o, u/, with very little if any phonetic overlapping among them. In turn, variants due to the linguistic context, such as nasalization, fronting or backing and so on, although detectable in acoustic analyses, are perceptually irrelevant (I will not discuss here whether the glides [j] and [w] should also be considered different phonemes in themselves and not mere allophones of /i/ and /u/ respectively.)
When it comes to adapting the various vowel sounds of English words into Spanish, speakers normally assimilate them to the perceptually nearest phoneme in their native language. There seems to be a strong influence of orthography in this process, since in the majority of cases the Spanish solution is in accordance with the vowel represented by the grapheme and not with the one that has a closer articulation. This way, English words with the phoneme /æ/, intermediate between Spanish /a/ and /e/, are normally adapted with /a/ [Λ], since <a> is the usual orthographic representation of that English vowel. For example: chat [’ʧΛt], bádminton [’bΛδmiņtɔn], sándwich [’sΛŋgwiʧ], etc. Even more clear-cut examples are the English words in which the grapheme <o> is pronounced as similar to Spanish <a> (in fact, as [ɒ]), but nevertheless are often assimilated into this language with /o/ [ɔ]: rock [’rɔk], pop [’pɔp]. Perhaps for the same reason, the English schwa [ə] tends to be adapted as /e/ in some words: single [’siŋgel], /u/ in others: surf [’surf], etc.
Anyway, we cannot postulate orthographic interpretation as a rule: in words like fútbol < football or béisbol < baseball the English [ɔ:] has in fact been adapted as Spanish [ɔ], in spite of the <a> grapheme. The prevalence of orthographic or of phonetic interpretations may be related to the primary channel (spoken or written) of diffusion of a specific English borrowing.
Also, Spanish does not phonologically distinguish between long and short vowels; lenghtening is for the most part used as a mere expressive device. Thus English long vowels are generally adapted as the most similar short ones in Spanish, as some of the above examples have already shown. Both [ı] and [i:] are transformed into [i], which can lead to homophony in some cases: bit, beat [’bit]. Spanish native speakers have well-known problems to distinguish between sheet or beach and certain similar words with [ı], which can result in some social discomfort.
Another significant phenomenon arises with the adaptation of words or syllables beginning with a glide. This is rather strange to Spanish articulatory patterns and very difficult to pronounce for native speakers; even words that should normatively begin with a glide, like hierba ‘grass’ or huevo ‘egg’, are almost categorically realized with a consonantal support: [’ʤerβΛ], [’gweβɔ]. The adaptation of English words follows the same pattern: jockey [’ʤɔkej], jazz [’ ʤaθ], western [’gweşter(n)], güisqui [’gwiski] < whisky, etc. It must be noted that, although güisqui is the orthographic variant recommended by the Real Academia, most speakers prefer to use the English form. In fact, I find it difficult to imagine a whisky brand being sold with the word güisqui written on it. For some time, the ever-undeterred Academia also recommended güeb as an adaptation of web.
As for diphthongs and triphthongs, the variety of schemes and their possible phonetic realizations is accordingly greater in English than in Spanish. In general, the latter language seems to have no problem in adapting English diphthongs when they are similar in both languages. Thus we have email [i’mejl], merchandising [merʧΛņ’dΛjsin], etc. Other schemes that are not so common, and hence relatively uncomfortable, to Spanish articulatory habits, may undergo monophthongization. This seems to be particularly the case with the [əw] diphthong, which does not exist in any Spanish word. That is why, for example, the expression OK (quite frequent in Latin American varieties) is normally pronounced as [’ɔkej], with loss of the first diphthong and maintenance of the second one; esmoquin [es’mɔkin] < smoking, Coca-Cola [kɔkΛ’kɔlΛ], etc. In not-yet-widespread borrowings, it may be maintained or slightly adapted: browser [’bɾɔwseɾ].