Speech Units Workshop

17-19 April 2023

University of Zurich [ˈt͡sʏrɪ]

Language endangerment is a topic of concern and the interest in language documentation has grown (Himmelmann 1998, 2006; Thieberger 2020). With the technological advent language documentation has extended to include more speech data. However, the number of instrumental studies dealing with the acoustic characteristics of under-documented languages remains relatively small. This workshop offers a forum for the discussion of the experimental study of Speech Units and aims at providing training for those interested in engaging with the topic.

Instrumental studies in intonational phonology have shown that languages vary in how they phonetically encode units of different sizes, such as the prosodic word, the accentual phrase, or the intermediate phrase (Jun 2006, Jun 2014). Despite great advances contributing new insights into the field of prosodic typology, studies dealing quantitively with more diverse languages remain under-represented. There is a need for more data and studies from more diverse languages to better understand typological variability and e.g., why some languages mark units on metrically strong heads, while others prioritise edges of larger units and whether this has an impact on language segmentation processes or first language acquisition.

From an acquisition perspective, studies on several Western-European languages have shown that children can package and process their input into larger units such as clauses at around 6 months of age (Nazzi et al. 2000 for English, Schmitz et al 2003 for German, Johnson & Seidl 2008 for Dutch). Later, children are able to segment the speech stream into smaller speech units such as words, using the salience of utterance edges (Seidl & Johnson 2006) and the rhythm, i.e., syllable vs foot, and phonotactic properties, such as vowel harmony (Nihan Ketrez 2014), of their target language. This order of segmentation from larger clauses to smaller units, and the types of cues identified remain to be investigated in more typologically diverse languages.

The whole-word phonology approach (Vihman & Croft 2007) is an emergentist approach to early phonological development, according to which children do not first learn the individual elements composing words, such as morphemes or affixes etc., but instead start by trying to produce whole words. To date, most studies were carried out on speech production of European languages (but see Khattab and Al-Tamimi 2013 for Arabic, and Ota 2013 for Japanese), and data is still lacking on other languages.