Alphabetter starts by showing each Persian word in English letters, then fades that help away and invites you to switch it off. Here is the research behind that choice, and where the evidence is still thin.
Researchers call this orthographic input: the written form a learner sees while learning a word. A large body of work shows that this written form shapes how second-language learners store and pronounce new words. Bassetti (2008) reviews evidence that learners read a new language through the spelling habits of their first one, which leads to pronunciations that a native speaker would not use. A later collection of studies (Bassetti, Escudero & Hayes-Harb, 2015) found the same pattern across many languages and learner groups.
Hayes-Harb, Nicol and Barker (2010) taught English speakers made-up words by sound, with pictures for their meanings. Some groups also saw the words spelled. When a spelling did not follow English spelling habits, learners were thrown off by it when later tested on how the words sounded. Escudero, Hayes-Harb and Mitterer (2008) showed the other side: written forms can also help learners keep two similar sounds apart. Either way, the spelling a learner sees becomes part of what they store, so it matters which spelling that is.
Romanized Persian leans on English habits that do not fit. “Kh” and “gh” stand for sounds English does not have. Three different Persian letters all come out as “s,” and four as “z.” An English reader’s first guess at many of these spellings is wrong.
In the app: once audio is available, the main prompt becomes the sound of the word, not its English-letter spelling.
Ehri’s research on orthographic mapping shows that readers come to know words on sight by tying each letter or letter group to the sound it stands for (Ehri, 2005, 2014). Share’s self-teaching hypothesis (Share, 1995) adds that every time a reader sounds out a word successfully, they learn its spelling a little better. That makes sounding out the engine of reading growth.
Both ideas point the same way. The link a Persian reader needs is between Persian letters and Persian sounds. A Latin-letter spelling adds a middle step (Persian script, then English letters, then sound) that a fluent reader never uses. Hearing a word while looking at its Persian form trains the direct link instead.
In the app: each word question plays the word, and each answer shows the letters right to left, with a play button, so sound and script are seen together.
The idea of scaffolding, from Wood, Bruner and Ross (1976), is that a tutor gives support a learner needs now and then pulls it back as they gain skill. Research on cognitive load later found why the pulling back matters. The expertise reversal effect (Kalyuga, Ayres, Chandler & Sweller, 2003) shows that guidance which helps beginners can hurt learners with more skill, because they spend effort processing help they no longer need. Renkl and Atkinson (2003) found that fading worked examples step by step led to better learning than switching from full help to none in one jump.
In the app: a word’s spelling shows in full the first time, faintly the second, and is hidden behind a “Show spelling” tap after that. Once you have finished two lessons and read six words, the Abc button pulses to suggest turning spellings off entirely.
In a well-known line of studies, Logan, Lively and Pisoni (1991) trained Japanese speakers to hear the English r and l sounds. Training with many different speakers led to learning that carried over to new words and new voices, which training with one speaker did not do as well (Lively, Logan & Pisoni, 1993). This approach is now called high-variability phonetic training.
In the app: every word has a male and a female recording, and the app picks one at random each time.
Persian script usually leaves out short vowels. A word like دست (dast, “hand”) is written with only d, s and t. Studies of Persian readers show that words missing their vowels are read more slowly and with more errors than words whose letters spell out every sound (Baluch & Besner, 1991). Readers fill in the missing vowels from memory of the word.
That means beginners need a way to learn each word’s full sound. A transliteration supplies the vowels, but in English letters. Native audio supplies them as real Persian sounds, which is the memory a reader will actually draw on.
There are few studies on romanization in Persian courses in particular. Most of the evidence above comes from other languages, or from first-language reading. Some research on other scripts suggests a consistent romanization can help early learners, so a blanket rule of “never show it” is not supported either.
That is why Alphabetter fades the spellings rather than removing them, suggests turning them off rather than forcing it, and keeps them one tap away on the Abc button.