Some of my clips don’t have too much speech. Is that a mistake? No, that’s ok---some recordings have a lot of speech while others only have a few seconds of speech. The percentage of speech will be recorded as data for each clip. If a clip doesn’t have any speech mark “no speech” under “Classify clip” and move on.
What speaker do I select if there isn’t any speech in the clip? Just ignore speaker selection. You only need to select “No speech” under “Classify clip” and you can move to the next clip.
What do I mark if the speech is too faint to make out speakers or languages? Mark “Unsure”
Does music count as “Media”? Yes
I made a mistake! Can I change my annotation? Yes, look in the terminal and note the clip number. Then, go into the responses.csv and edit that annotation. Be sure to use the same text (caps when necessary, don’t add spaces)
What if I hear a researcher talking in the clip? Mark “Researcher present” under “Classify clip” and move on
One of my clips is less than 30 seconds. What’s wrong? This is probably an “end clip” - the remainder after the rest of the recording had been divided into 30 second chunks. Just classify as best you can.
How do I classify coughing, sneezing, etc. if there is no other speech in the clip? “No speech”
The app froze! Just kill it in the terminal and run again. Your responses have been saved!
How do I classify vocalizations from the target child? Particularly younger children sometimes say syllables like “ba ba” which are not one of the target languages but are certainly speech. Check the "Is the child vocalizing" box. If it's an older child (not the target child) select "unsure" for language.
Does live music count as “Media?” Yes.
How do we classify children’s word play? (i.e. we can’t understand the language but the child is definitely producing words, rhymes, etc.) Mark “Unsure” under Language and the appropriate speakers.
What is the line between an adult and a child? Rely on acoustic information: if it sounds like the adolescent has gone through puberty, they’re an adult And it’s ok to say ‘unsure’!
How are we classifying child vocalizations? Crying, laughing, vowels, consonants. Any type of crying counts: moaning, wailing, etc. Coughs and sneezes are excluded - these are vegetative sounds are not vocalizations
Do we select ‘no speech’ if the target child is the only one vocalizing? No, child vocalizations still count as speech.
What do we mean by ODS and CDS? CDS: All cues available to determine if speech was directed to child or other children. If it could be directed to the child, it’s CDS (even if it’s also directed to the other children, even if it’s directed to the entire family)
What do we do when there’s part of the recording that we cannot understand, and would mark as unsure, but another part that we are sure of? Code for the parts that we can understand and ignore the things we cannot hear/understand. Obviously if you’re unsure about everything in the clip, then mark it as “unsure.” Unsure comes at the end when we don’t think we can decide between any of the other categories
What do we do if there are two different media sources, one has language and one doesn’t have language? Just classify the media source that has language and ignore the one without.
How do we classify the language in clips where there are nonsense words but the words are clearly in the phonology of one of the languages? If the nonsense words are clearly in one phonology or the other, mark it as that language If the nonsense words are more ambiguous wrt phonology, say that the language is unsure
How do we classify speech that is expressed with words from one language in the phonology of the other (basically accented speech)? Prioritize words over phonology - so pick the language that the words are expressed in, not the phonology (e.g. “oh my god” expressed with a Spanish accent would be marked as “English”)
What do we put down for speakers in clips where the only vocal activity comes from media? Don’t classify any speakers, and you don’t need to make a selection in the “Classify Clip” menu - just classify the media language and move on
How do I classify an interaction where I can’t tell who one of the speakers is? Default to assume that the unknown speaker is an adult and classify accordingly (by selecting multiple adults, classifying the language, etc.) Note that this is different from just selecting speech to someone unknown because annotating this way helps us clarify that there were two speakers in the clip