edit 2024-baap

README.md

@@ -8,3 +8,6 @@ The file `code/01_analysis.qmd` includes code of the analysis of the paper *Intr
 
 The data used in `code/01_analysis.qmd` are available through the R package <https://github.com/stefanocoretta/coretta2018itaegg>.
 
+## Talks
+
+- [BAAP 2024](comm/talks/2024-baap/)

comm/talks/2024-baap/index.html

@@ -394,9 +394,92 @@ <h2>Hypotheses</h2>
 <p>(a) Full mediation.</p>
 <p>(b) No mediation.</p>
 <p>(c) Partial mediation.</p>
+</section>
+<section id="methods" class="slide level2">
+<h2>Methods</h2>
+<div class="box-note">
+<ul>
+<li><p>Pre-existing data: <a href="https://osf.io/xdgfz/" class="uri">https://osf.io/xdgfz/</a>.</p></li>
+<li><p>19 speakers of Northwestern Italian (Verbano-Cusio-Ossola).</p></li>
+<li><p>CVCo words (mostly nonce, some real), in 5 frame sentences.</p>
+<ul>
+<li><p>C = /p, t, k/, all permutations.</p></li>
+<li><p>V = /a, e, i, ɔ, u/.</p></li>
+</ul></li>
+</ul>
+</div>
+<div class="fragment">
+<div class="box-tip">
+<ul>
+<li><p>Vowel duration: from appearance of F2/3 to disappearance of F2/3.</p></li>
+<li><p>F1/F2 measurements from <a href="https://github.com/santiagobarreda/FastTrack">FastTrack</a> (Barreda 2021).</p></li>
+</ul>
+</div>
+</div>
+<div class="fragment">
+<div class="box-note">
+<ul>
+<li><p>Causal Inference (McElreath 2019).</p></li>
+<li><p>Bayesian models fitted with <a href="https://paul-buerkner.github.io/brms/">brms</a> (Bürkner 2017, 2018, 2021) in R (R Core Team 2024).</p></li>
+</ul>
+</div>
+</div>
+</section>
+<section id="hypotheses-revisited" class="slide level2">
+<h2>Hypotheses (revisited)</h2>
+<div class="box-warning">
+<p>In a model with vowel and F1 as predictors:</p>
+<ul>
+<li><p>(a) Full mediation</p>
+<ul>
+<li>The effect of vowel should be (practically) 0.</li>
+</ul></li>
+<li><p>(b) No mediation</p>
+<ul>
+<li>The effect of F1 should be (practically) 0.</li>
+</ul></li>
+<li><p>(c) Partial mediation</p>
+<ul>
+<li>The effect of both vowel and F1 should not be (practically) 0.</li>
+</ul></li>
+</ul>
+</div>
+</section>
+<section id="results" class="slide level2">
+<h2>Results</h2>
+
+<img data-src="images/bms1-pred-plot-ms-hz.png" class="quarto-figure quarto-figure-center r-stretch"></section>
+<section id="interpretation" class="slide level2">
+<h2>Interpretation</h2>
+<div class="box-note">
+<ul>
+<li><p>There is a <strong>robust effect of F1</strong> and the effect is <strong>non-linear</strong>.</p>
+<ul>
+<li>Negative up to 400 Hz, positive from 450 Hz.</li>
+</ul></li>
+<li><p>There is a <strong>robust effect of vowel</strong>.</p>
+<ul>
+<li>F1 alone is not sufficient to explain vowel duration.</li>
+</ul></li>
+</ul>
+</div>
+<div class="fragment">
+<div class="box-tip">
+<p><strong>The results are compatible with the (c) Partial Mediation hypothesis.</strong></p>
+</div>
+</div>
+<div class="fragment">
+<div class="box-note">
+<p>This means that:</p>
+<ul>
+<li><p>Each vowel must have a duration target.</p></li>
+<li><p>Distance to target explains duration within each vowel category.</p></li>
+</ul>
+</div>
 <div class="quarto-auto-generated-content">
 <div class="footer footer-default">
 
+</div>
 </div>
 </div>
 </section>

comm/talks/2024-baap/index.qmd

@@ -77,3 +77,81 @@ Process (or rule) scattering is the scenario in which the same process coexists
 \(c\) Partial mediation.
 
 ## Methods
+
+::: box-note
+-   Pre-existing data: <https://osf.io/xdgfz/>.
+
+-   19 speakers of Northwestern Italian (Verbano-Cusio-Ossola).
+
+-   CVCo words (mostly nonce, some real), in 5 frame sentences.
+
+    -   C = /p, t, k/, all permutations.
+
+    -   V = /a, e, i, ɔ, u/.
+:::
+
+. . .
+
+::: box-tip
+-   Vowel duration: from appearance of F2/3 to disappearance of F2/3.
+
+-   F1/F2 measurements from [FastTrack](https://github.com/santiagobarreda/FastTrack) (Barreda 2021).
+:::
+
+. . .
+
+::: box-note
+-   Causal Inference (McElreath 2019).
+
+-   Bayesian models fitted with [brms](https://paul-buerkner.github.io/brms/) (Bürkner 2017, 2018, 2021) in R (R Core Team 2024).
+:::
+
+## Hypotheses (revisited)
+
+::: box-warning
+In a model with vowel and F1 as predictors:
+
+-   \(a\) Full mediation
+
+    -   The effect of vowel should be (practically) 0.
+
+-   \(b\) No mediation
+
+    -   The effect of F1 should be (practically) 0.
+
+-   \(c\) Partial mediation
+
+    -   The effect of both vowel and F1 should not be (practically) 0.
+:::
+
+## Results
+
+![](images/bms1-pred-plot-ms-hz.png){fig-align="center"}
+
+## Interpretation
+
+::: box-note
+-   There is a **robust effect of F1** and the effect is **non-linear**.
+
+    -   Negative up to 400 Hz, positive from 450 Hz.
+
+-   There is a **robust effect of vowel**.
+
+    -   F1 alone is not sufficient to explain vowel duration.
+:::
+
+. . .
+
+::: box-tip
+**The results are compatible with the (c) Partial Mediation hypothesis.**
+:::
+
+. . .
+
+::: box-note
+This means that:
+
+-   Each vowel must have a duration target.
+
+-   Distance to target explains duration within each vowel category.
+:::