Table 2.

CoST-P iterations

#	Changes	Testing procedure
1	This initial version included five broad recommendations and 20 sub-recommendations structured in list format	Further research regarding unique connected speech characteristics
2	Additional recommendations added; this version included six broad recommendations, 30 sub-recommendations, and 15 additional recommendations; of particular note was the consideration of linguistically/contextually appropriate speech alterations (e.g., reduced forms such as “dunno”) and the addition of examples to explain recommendations	Areas for further change determined by the research team (through consensus)
3	Table formatting and contingencies for sub-recommendations	Suitability of each structure determined by consensus
4	Decision to transcribe words as connected where they are naturally joined in connected speech; spaces to be used to indicate appropriate inter-word pause, and pause markers to be used to indicate inappropriate segmentation	Transcription and analysis assessed using test file
	Decision to transcribe dysfluencies in all tiers (to eliminate impact on relational phoneme analysis) and to note the number of dysfluencies in each utterance (e.g., “$ $”)	Transcription and analysis assessed using test file
	Decision to transcribe stereotyped speech or sing-song-style speech and include a note below indicating the speech type	Transcription and analysis assessed using test file
5	Decision to use audio (rather than video) recordings to eliminate effects of auditory illusions (e.g., McGurk-like effects) which may occur due to inadvertent audio/video misalignment or visual dominance in information processing [43]	Audio was isolated from files; transcription and analysis assessed using test file
	Decision for unintelligible phonemes or utterances to be transcribed with use of consistent codes; for example, if the manner of consonants is audible, but the consonant place is unidentifiable, transcriber should use generic marker	Transcription and analysis assessed using test file
6	Decision to use diacritics to transcribe allophonic variation only where it is linguistically inappropriate (e.g., /æ/ in /mæn/ does not require a hyper-nasal diacritic as it is appropriate for the vowel preceding a nasal consonant to be nasalised)	Decision made by author consensus; transcription and analysis assessed using test file
7	Decision to transcribe primary and secondary stress markers in both the children’s target productions and actual productions	Transcription and analysis assessed using test file
8	Recommendations to ensure an error is accounted for only once in computer-based analysis	Assessed with use of a test file; analyses run in Phon and compared to manual analysis
9	CoST-P finalised and implemented for all test transcriptions; protocol fidelity conducted on this final version of the CoST-P