Glossary of silent letters. Part two.
In a class by itself is the <gh> digraph that evolved from velar fricatives usually spelled <g> or <h> in Old English, and later spelled <g> or yogh or <gh> in Middle English. In a few cases those fricatives became [f] as in laugh; in even fewer they became [g] as in burgh. In most cases, however, the <gh> spelling remained but the sound was completely lost, especially before what became long vowels and diphthongs, as in words like light, eight, sigh, through, though, bough. If you analyze to complex spellings, then the <gh>’s in words like light and eight can be treated as part of the complex <ght> spelling of [t]. But with words like sigh, through, though, bough, I’d be reluctant to treat the <gh> as part of complex vowel spellings. It just seems awkward to call <gh> vowel letters. I’m inclined now to call them silent letters with some diacritic force, since they tend to come after long vowels and diphthongs, and were often responsible for the lengthening and diphthongization of the vowels in earlier centuries. (For more on <gh>, see the earlier questions in this Question and Answer area.)
Here is a partial list of spellings in words containing what in AES I treated as complex consonant spellings but that could in theory be treated as simple spellings plus one or two silent letters: the <pb> of cupboard; the <bp> of subpoena; the <ph> in shepherd; the <ph> in common pronunciations of diphtheria and diphthong; the <ld> in could, should, would, solder, the <bd> in lambda; the <th> in thyme, Thomas; the <bt> in doubt, debt; the <pt> in receipt, pterodactyl, ptomaine, ptarmigan; the <ct> in indict; the <dt> in veldt, the <gh> in ghastly, ghetto, ghoul (which do not descend from the Old English velar fricatives); the <tg> of mortgage; the <ckgu> of one pronunciation of blackguard; the <ch> of chord, echo; the <cch> of saccharin, zucchini; the <qu> of antique, plaque; the <sc> of viscount; the <lv> of calve, halve and the <lf> of calf, half; the <ft> of often, soften; the <tth> in Matthew; the <wh> of whole, whose, whore; the <sc> of scent, scissors; the <ps> of psoriasis, psychology; the <st> of fasten, listen; the <cs> of Tucson; the <dj> of adjoin, adjust; the <dg> of budget, ledger; the <mb> of bomb, lamb; the <lm> in common pronunciations of palm, calm; the <mn> of autumn, hymn; the <gm> of diaphragm, phlegm; the <kn> of knee, knob; the <gn> of gnat, gnome, reign; the infamous <pn> of pneumonia; the <mn> of mnemonic; the <sl> of island, aisle; the <wr> of wrap, wrist; the <rh> of rhetoric, rhubarb; the <rrh> of hemorrhage, diarrhea; the <rps> of corps.
This longish list doesn’t address the common consonant digraphs ending in <h>: <ch>, <sh>, etc. And I’m not sure what would happen if one tried systematically to analyze vowel di- and trigraphs as simple spellings plus silent letters: The <ui> in built, building might be a good candidate, but what about the <ea> of beagle, measles or the <eo> of people, or the <eu> of lieutenant.? It seems to me that rather than multiplying silent letter, it is better to treat examples like those in the preceding paragraph (and the vowel digraphs and trigraphs) as complex spellings and to try to explain the historical processes, usually sound change and simplification, that led to that complexity, remembering always that for good social and technological reasons the spoken language changes faster than the written. The one major case where I analyze to silent letters is to avoid including vowel letters in complex consonant spellings and consonant letters in complex vowel spellings – as with the silent <i> in business and the silent <h> in exhaust.
This is probably a much longer answer than you expected – or wanted – but the issue is surprisingly complicated and subtle, and you have to remember that even retired college professors are used to thinking in fifty-minute units of time.