I launched the Speech Sound Pics (SSP) Approach in Australia around a decade ago.
I can't launch it here in the UK because it is not 'synthetic phonics' and I include activities to develop GSF (an important bridging process) etc I want the children to use other cues eg vocabulary knowledge, context, to figure out words with unfamiliar GPCs. I have included an example below. See Speech Sound Detective - this prompts the children to figure out the word and then 'track back' ie so what is the new grapheme?
Share identifies this as an important skill
Contextual information plays a secondary role in self-teaching in that it compensates for partial or incomplete decoding resulting from poor decoding skills (e.g., poor readers) or word property (e.g., irregularly spelled words in English).
The Self-Teaching Hypothesis, first developed by Jorm and Share (1983) and elaborated by Share (1995), is a leading theoretical framework in conceptualizing how unassisted orthographic learning occurs among children, a process called self-teaching. Children could not acquire all new written words via direct instruction due to the vast number of new words developing readers encounter in printed texts Direct instruction involves explicit, direct teaching. The Self-Teaching Hypothesis argues that children implicitly acquire new words in their independent reading, and the key underlying mechanism to make this happen is phonological recoding (i.e., print-sound translation)
Phonological recoding helps children’s orthographic learning over their course of development; the occurrence of self-teaching via phonological recoding depends on the presence of unfamiliar words in the text. Relatedly, self-teaching not only supports children’s acquisition of individual written words (e.g., friend), but also helps them acquire general spelling patterns within the writing system (e.g., the spelling pattern of -end), which is termed lexicalization in Share (1995). Conceptually, lexicalisation (i.e., learning general spelling patterns) in self-teaching may involve a special type of statistical learning—implicitly extracting rule-based regularities in a writing system (Nation & Castles, 2017; Share, 1995). Two simulation studies that modelled self-teaching in English suggest that the model can acquire 25,000 words via self-teaching starting with a small number of grapheme-phoneme correspondences (Ziegler et al., 2014) and that context benefits self-teaching when phonological recoding is partial for irregular words (Pritchard et al.2018).
In Australia I wanted children to practice their Code Level graphemes within books (so more realistic) and so used to go through the phonics books available, and I would check all graphemes and organise them for teachers.
This shows Green and Purple Code Level Readers
(there are over 500 books in the guide)