conv02_193_little.wav : between "slight" and "little" and between "upwards" and "to" 
 
ANS (initial)
(note: file contains 3 breaks labelled as 2's (inc a 2p), but we have agreed that the labelled 2 after "little" should really be a 1: glottal stop on "upwards" leads to sense of disjuncture, but probably is in service of prominence for pitch accent on "upwards")  
RESPONDENT A #1: Agree

ANS #2: These 2 cases illustrate for us the difficulty of distinguishing between flat phrase tone and no phrase tone. WeÕd like to deal with this type of ambiguity explicitly in the manual revision. 
Can you help us with a formal description of what it is about this example that makes it a 3 or 4 (vs a 3m)? (Is it the pause?)
RESPONDENT A #2: Here, Ô3mÕ means you hear a 3-like juncture but no phrase accent. First, I think the juncture is larger than 3 probably due to the pause. Second, Ôno phrase accentÕ on ÔslightÕ and ÔupwardÕ means that the pitch accent on these words would all belong to the same ip (the best cue of this would be downstep across the boundary). But f0 on the main words (slight, little, upwards, right) seems to suggest a slight pitch reset on ÔlittleÕ, and another reset on Ôto the rightÕ.  I think thatÕs why I heard three groupings (itÕs like a slight)(little upwards)(to the right). 
In sum, to be Ô3mÕ, there should be an f0 cue telling the tones across the boundary belong to the same ip.  

ANS (initial)
"slight": originally labelled as 2
¥"slight": Is there a boundary-related tone? no. 
Is there duration lengthening? yes (also pause after). 
Appropriate for iP? No (doesn't sound like a boundary, esp played in isolation):  
RESPONDENT A #1: I hear the juncture after ÔslightÕ and ÔupwardsÕ the same.  !H* !H- on ÔslightÕ and H* !H- on ÔupwardsÕ. Or !H-L% in both. 
RESPONDENT B #1: We also hear a strong parallelism between Òlike a slightÓ Òlittle upwardsÓ and Òto the rightÓ. These sound like three equally weighted rhythmic chunks of the utterance. So we would label a 3 break after ÒslightÓ and ÒupwardsÓ, and a 4 after ÒrightÓ. There is only a 4 Hz rise on ÒslightÓ, sounds pretty flat. Our labeling is H* !H* H- on Òlike a slightÓ, but we donÕt hear or see a downstep on the phrase accent relative to the !H* on ÒslightÓ. We donÕt hear an accent on ÒlittleÓ, and have simple H* on upwards, followed by H- phrase accent.
RESPONDENT C #1: Yes, those 3 chunks are quite similar. I would go with BI 4 (or 4- on the medial ones) on each of them. ÒitÕs like a slightÓ H* !H* H-L% Òlittle upwardsÓ H* H* H- L%. To add to RESPONDENT A #2Õs description above of 3m, I would say that often 3mÕs (or 4mÕs) are labeled in cases where you hear a large disjuncture but see downstepping across phrases Ð that is, RESPONDENT A #2Õs Òshould be an f0 cue telling the tones across the boundary belong to the same ip. Ó
ANS (initial)
could be 3m, or else whatever 2p becomes
¥too long for a 2, want to call it a 2p--but what will happen to 2p? Will this become a 1p? 1pr and 1ps (vs 3m?): [possible use of dots "." to separate multiple diacritics eg 1pr.ps]: we're not sure whether this prolonged "slight" with following pause should be labelled as a disfluency--because somehow this seems well-formed. 
RESPONDENT A #1:  I also feel this is well-formed. If use 4 (!H- L%), pause is ok.
ANS (initial)
¥ unsure whether "little" is pitch-accented or not, or could be reset for iP  
RESPONDENT A #1: I hear H* pitch accent on ÔlittleÕ.  RESPONDENT B #1: We donÕt, but thereÕs a pitch reset there.
RESPONDENT B #1: If 1pr is for prolongation, what is 1ps?
 
RESPONDENT C #1: I donÕt hear any disfluency here. I wouldnÕt use ÒpÓ.
ANS (initial)
"upwards": originally labelled as 2p
Our question to you: Is this a 3m or a 3 with a flat (H-) phrase accent? 
RESPONDENT A #1: I would choose 3 with a phrase accent, but to represent the same degree of jucture as 'slight', I would label 4 with !H-L%.
 RESPONDENT C #1: same here.
RESPONDENT B #1: We think this is a possible alternative, but just don't hear each of these phrases as a full IP when we listen to the whole file.  But listening to the file just up to "upwards" does allow for the possibility that the speaker ends the utterance there, though.  Tricky.
ANS (initial)
¥very long /s/: lengthened coda increases sense of boundary iP: 
RESPONDENT A #1: agree 
could even be a 4 (4-) with H-L%: continuation flat, tone indicates uncertainty? (But we're inclined to call it a 3 over a 4)  RESPONDENT A #1: Agree
ANS (initial)
¥orig label was 2p, but we don't agree
(note: file also contains "s-": is this a cutoff? prolonged? show to RESPONDENT A, use for c-cutoff, pr-prolongation discussion (c.pr)) 
RESPONDENT A #1:   Since Ôs-Ô is part of a word, we use ÔcÕ (for cutoff) and since itÕs longer than the default ÔsÕ, we would use ÔpÕ (prolongation; we could change the label to ÔprÕ). So, Òs-Ò would be labeled as Ò1c.prÓ
RESPONDENT B #1: we are labeling disfluencies on a separate tier, with the labels ÒREParandumÓ, ÒALTerationÓ and ÒEDIT phraseÓ. In this example of a hesitation/planning disfluency, the REP interval covers ÒitÕs like a s-Ò and the ALT interval is the second ÒitÕs like a slightÓ. There is no EDIT phrase here. The juncture between the REP and ALT is the interruption point. Given that weÕve labeled this on the disfluency tier, we would not indicate a 1 break after the disfluency Òs-Ò, nor would we call this a 1p. ItÕs a lengthened segment and a truncated word, followed by the reset on ÒtheÓ (as in your labeling).
RESPONDENT C #1: just so every break marker has a tag, I would give Òs-Ò a label on the BI tier, even if itÕs already clear from the disfl tier (I am currently labeling disfluencies on an additional tier too). I would vote for Ò1c.prÓ.
RESPONDENT A #1: BTW, I would label H* H-H% on Òto the rightÓ.  
ANS #2: ThanksÑcould you let us know what makes this an H- rather than L-?
RESPONDENT A #2: If itÕs L-, the f0 contour on ÔrightÕ would be falling (H* to L-) and then rising (L-H%). But here, the contour on ÔrightÕ is flat and rising.  Note that f0 is falling from Ôto theÕ to ÔrightÕ, but not during the word ÔrightÕ.   
RESPONDENT B #1: We agree with H* H- H%, since the final rise looks like it comes in two steps, and begins back at the accentual peak. We thought maybe the accent was L+H*, but decided the dip was due to /r/.
RESPONDENT C #1: yes, H* H-H% here.