Syllabic affiliation of glides in chicano
Last update: 2023-12-01
Paso 1 (hecho): grabar gente silabificando.
Paso 2: Poner las grabaciones en línea para que otros raters den su juicio (i.e., ¿esta persona ha dicho la ka pi aisto? - nativos (de donde?) - cuantos participantes? - cuantas opciones? (mirar doc a ver si hay tendencias) - que se nos olvida?
Paso 3: Reclutar más participantes para que hagan la tarea de silabificación en línea y las opciones serían las respuestas más comunes del piloto (paso 1) que han sido confirmadas por raters (paso 2). - ven texto - escogen entre dos opciones - nativos (de donde?) (dos grupos?) - cuantos participantes?
Analyses
Syllabification task
The following plots show results from the syllabification task. Four syllable nonse words containing a CGVGC sequence in the penultimate syllable were presented on a screen. Participants read the words out loud, syllable by syllable. The plots show the overall distribution of the responses, which have been classified in the following manner:
- tripthong: the critical sequence was produced in a single syllable
- vowel + dipthong: the critical sequence was produced as two syllables: CV + VGC (labeled as ‘hiatus’), e.g. “lakapiaisto” [la.ka.pi.’ai̯s.to]
- simplification: a segment was elided from the critical sequence, simplifying the critical syllable. The most common strategy was to drop the pre-vocalic glide, creating a sequence like [la.ka.pai̯s.to] instead of [la.ka.’pjai̯s.to]), but the post-vocalic glide was also dropped occasionally.
Figure 1 plots the proportion of responses across all data (collapsing across the pre-vocalic glide and onset). We observe that tripthongs were produced in approximately 45% of the targets. A production containing a hiatus made up roughly 30% of the data, followed by a simplification of some sort (~25% of the time).
Figure 1: Responses in syllabification task.
If we combine hiatus and simplification (Figure 2), we see that they
make up approximately 60% of the data.
Figure 2: Responses in syllabification task combining ‘hiatus/simplification’.
Overall, the task provides evidence supporting the hypothesis that pre-vocalic glides can be part of the onset in this variety of Spanish.
To consider:
- What factors can explain the variation in response types?
- Possibilities:
- Frontness of glide. Is there a difference between [j] and [w]?
- Preceding consonant. The pre-vocalic glides in this task are preceded by a stop (voiced and voiceless) or [f]. Is there is a difference in response strategies as a function of preceding consonant?
- Figure 3 suggests that frontness and preceding consonant are not critical in determining response type.
Figure 3: Responses in syllabification task as a function of glide, preceding consonant, and voicing.
Statistical analysis:
- Multinomial mixed effects regression
- Criterion: response (tripthong/hiatus/simplification)
- Predictors: glide ([j]/[w]), consonant ([b, d, g, p, t, k, f])
- Random effects:
(1 + glide + consonant | participant)(1 + glide + consonant | item)
- Another possibility is to combine hiatus/simplification and use standard logistic regression.
Carrier sentence task
The following plots illustrate the results from the carrier sentence task. These analyses test the hypothesis that pre-vocalic glides will be disallowed if preceded by a palatal consonant. To test this, we measure duration, F1, and intensity of the pre-vocalic glide in two environments: after a palatal consonant and after any other consonant. If pre-vocalic glides are blocked after palatals, we expect to observe differences in overall duration. Figure 3 plots the duration data.
Figure 3: Duration as a function of preceding consonant.
We see a longer duration for glide segments following a palatal consonant (contrary to the hypothesis). There are two possible explanations (that I can think of). 1) The participants are able to produce a glide in this context (at least in some cases) because the preceding consonant is not palatal (i.e., ‘ch’ is surfacing as [ʃ] or the fact that it is alveo-palatal is different enough to allow a glide afterwards). 2) The participants are not producing a glide, but rather are lengthening the onset.
We can examine the plausibility of (1) by looking at the formant trajectory over the course of the segment. If the participants are able to produce a glide in some contexts but not others we should see differences in F1 as a function of the preceding consonant.
Again, the second possibility is that participants are lengthening the palatal consonant (not producing a glide) because of the fact that they cannot produce both. In other words, they might be trying to produce something, but because the target is illicit, they resort to lengthening the onset. If this is indeed what is happening, then we would expect the intensity of the lengthened segment to be lower from that of a glide. Why? Because more consonant-like productions have lower intensity than more vowel-like productions. Thus we analyze the time course of intensity in order to see how the two environments differ as they near the following [a].
