|Abstract:||The palatal lateral is a rare sound in the world's languages; a review of the literature reveals just 23 languages that currently possess the palatal lateral. Similarly, only 15 (or 3.33%) of the languages in the UCLA Phonological Segment Inventory Database (UPSID) (Maddieson and Precoda, 1991) can claim to currently possess the palatal lateral. While UPSID reports that an additional five languages (Basque, Guarani, Iate, Spanish, Turkish) possess the palatal lateral, these languages have either lost the palatal lateral or were included erroneously. Understanding the production and perception of rare speech sounds is important for understanding the distribution of speech sounds cross-linguistically, especially with regards to the establishment of a single phonetic alphabet (i.e. the International Phonetic Alphabet (IPA)) that can be used to describe and transcribe the languages of the world (Ladefoged and Everett, 1996). An investigation of rare speech sounds can also reveal important findings regarding the physical limitations of the vocal tract and human auditory system.
Given that the palatal lateral is a rare speech sound, a complete description of the articulation, acoustics, and perception of this sound does not currently exist. Accounts of the palatal lateral vary with regards to terminology; the palatal lateral has also been referred to as a so-called "phonemically" palatalized lateral (Zilyns'kyj, 1979), a laminal post-alveolar lateral (Ladefoged and Maddieson, 1996), and an alveolopalatal lateral (Recasens, 2013). Furthermore, current literature also does not distinguish between the palatal lateral and a palatalized lateral. The lack of agreement in literature regarding terminology can present problems when attempting to assess whether a palatal lateral in one language is similar to a palatal lateral in another language. This dissertation provides a comprehensive description of the palatal lateral, as a means of initiating cross-linguistic comparisons of the palatal lateral as well as understanding the difference between a palatal and palatalized lateral.
A two-part study of the articulation and acoustics of the palatal lateral in Brazilian Portuguese (BP) was undertaken in this dissertation. Articulatory data was collected using electromagenetic articulography (EMA) from 10 female native speakers of BP from São Paulo state in Brazil, which permitted the simultaneous collection of acoustic information. Study 1 investigated the articulation of the palatal lateral through a battery of measures and compares the palatal lateral against the palatalized lateral approximant, alveolar lateral approximant, palatal approximant, palatal nasal, palatalized nasal, and alveolar nasal. Study 2 analyzes the acoustics of the palatal lateral in comparison to the palatalized lateral approximant, alveolar lateral approximant, and palatal approximant.
A third study was included in the appendix. This study incorporates a phone identification task to understand the role of acoustic saliency in the rareness of the palatal lateral, i.e. compared to other palatal sounds, is the palatal lateral more likely to be misidentified and if so, as which sounds? This task also investigates whether there is a perceived difference between the palatal and palatalized lateral that may not be captured by Study 1 and 2, in addition to whether native speakers of BP are better at distinguishing the two sounds than non-native speakers (here, native speakers of American English). The palatal lateral was compared to the palatalized lateral, palatal approximant, alveolar lateral approximant, palatal nasal, palatalized nasal, alveolar nasal, voiced alveolar stop, and voiced palatalized alveolar stop. 25 (11 male, 14 female) natives speakers of BP and 20 (11 male, 9 female) native speakers of American English with no extensive exposure to BP participated in this study.
Results from Study 1 show that the palatal lateral is articulated laminally with a high front tongue body and concave anterior tongue shape that gradually becomes straighter as the phone progresses. Acoustic results in Study 2 indicate a median F1, F2, and F3 of 367 Hz, 1954 Hz, and 3035 Hz respectively for female speakers of BP. Statistical analysis reveals little or no evidence of significant difference between the palatal lateral and palatalized lateral with regards to the shape of the tongue body, duration of the phone, or formant frequencies.
The perception study included in the appendix finds that while both native and non-native speakers of BP distinguish between the palatal lateral and palatalized lateral at chance level, native speakers of BP perform better than the non-native speakers at correctly identifying the palatal and palatalized nasal. This study also finds that of all the sounds included in this task, the palatal and palatalized lateral are the most likely to be misidentified as the palatal approximant for both participant groups, with the addition of -3 dB of speech-shaped noise greatly increasing the rate of confusion. However, the palatalized lateral is inaccurately identified as a palatal approximant at a confusion rate nearly double or more than the palatal lateral.
This dissertation reveals that the palatal and palatalized lateral are essentially the same sound in BP. Furthermore, there is no evidence that indicates that the palatal or palatalized lateral are composed of two separate phones, i.e. an alveolar lateral approximant followed by a palatal approximant. Findings from the perception study support the proposal that yeísmo (i.e. the merger of the palatal lateral in favor of the palatal approximant (Colantoni, 2001; Hualde et al., 2005)) occurs because lateral sounds are less robust against added noise than nasal sounds. I argue here that this contributes directly to the rareness of the palatal lateral.