A spelling pronunciation is the pronunciation of a word according to its spelling when this differs from a longstanding standard or traditional pronunciation. Words that are spelled with letters that were never pronounced or that were not pronounced for many generations or even hundreds of years may sometimes be affected by this. it tends to effect words people tend to encounter first in writing but which many people know the most. sometimes a spelling pronunciation actually eventually becomes the dominant way to say the word. (for example the "D" in "admiral" in english). as hebrew is the only dead language to ever be successfully revived, it could easily contain many spelling pronunciations. (The written forms of dead languages are almost always better preserved then their spoken counterparts). I could conceive of those circumstances making a language infected with spelling pronunciations; I just don't know if the language actually contains them, and am curious. I was wondering if there are specific ones known as well as if those exist in the language and how prevalent they are.
One phenomenon that fits well in the definition of "spelling pronunciation" is the reflexes of tsere in Modern Hebrew. There are two graphical versions of the vocalization sign tsere: without yod (סֵ) and with ("full" tsere: סֵי). Historically, it is the same vowel (the exact pronunciation varies according to tradition), but Modern Hebrew developed a loose tendency for tsere to be pronounced as /e/ in words that are spelled without yod and as /ey/ in words that are traditionally spelled with a "full" tsere. A systematic application of this tendency led to minimal pairs in the third-person plural possessive ending for singular and plural nouns: morenu מורֵנו (our teacher) ≠ moreynu מורֵינו (our teachers), again, a distinction that should have been purely graphical. It is documented e.g. in S. Cohen (2016) Person Markers in Spoken Spontaneous Israeli Hebrew: A Systematic Description and Analysis, p. 45. Pronunciation of the masculine plural ending of nouns in construct as /ey/ is also very systematic: morey מורֵי (teachers [of]) and this also created minimal pairs, e.g.: bne בנֵה (build!) ≠ bney בנֵי (children [of]).
Another set of phenomena concerns vocalization pointing. The system of Tiberian pointing that is also used for vocalization of modern Hebrew in pointed texts today was originally devised for Tiberian pronunciation of Hebrew and became the dominant pointing system among the different Jewish communities in Medieval times. None of the European Medieval pronunciation traditions or that of Modern Hebrew actually correspond to the Tiberian pronunciation, so the pointing system was open to interpretation which in turn influenced pronunciation.
One example of this is the pronunciation of hataf patah (סֲ) and vocalic shwa (סְ). From the writing of Tiberian Masoretes (see a summary and history of the tradition e.g. in G. Khan (2020) The Tiberian Pronunciation Tradition of Biblical Hebrew), it is clear that they intended the hataf patah sign on gutturals to be a graphical guide to remind the reader not to reduce the corresponding vowel. For them, shva-na was to be pronounced in an unmarked case the same way as hataf patah, i.e. with as a short [a]. I don't know at what point this graphical distinction made its way into pronunciation of Medieval Hebrew traditions, but it was inherited in Modern Hebrew as follows: shva-na (when pronounced) is pronounced as /e/ (originally, [ə]) while hataf patah is pronounced as /a/ (originally understood as an extra-short [ă]) which created distinctions that for the Tiberian Masoretes did not exist: adaber אֲדבר (I will speak) vs tedaber תְדבר (you will speak).
Going beyond Modern Hebrew: The vowels of Ashkenazic Hebrew gradually developed from a Sephardi-like five-vowel system to a "Tiberian"-like seven vowel system. For example, kamats (סָ) was interpreted (likely based on works of David Kimhi and other Medieval grammarians) as denoting a long [a:] vowel. This led to realignment of the reading tradition facilitated by pointed texts. (Next, perhaps under the influence of a similar process in Upper German dalects, [a:] became rounded and approached the quality actually intended by the Tiberian Masoretes: /ɔ/.) See A. Beider (2015) Origins of Yiddish Dialects.
For completeness sake, one might also mention that there are pronunciation habits of Modern Hebrew speakers that do not fit exactly in "spelling pronunciation" but are certainly influenced by the ambiguity of the unpointed spelling:
- Modern Hebrew speakers very frequently mis-pronounce loanwords and non-Semitic names, but also Semitic-origin words. Just a couple of recent examples from personal encounters (all to be considered incorrect): pruton < proton פרוטון, sotskever < sutskever סוצקבר, khulon < kholon חולון.
- Similar to these are the many cases of paradigmatic levelling which is also facilitated by the spelling. In Modern Hebrew they are considered incorrect but some are very widespread: likfots < likpots לקפוץ, pitaron < pitron פתרון, shlavim < shlabim שלבים, shlav < shalav שלב, pa'amim < pe'amim פעמים.