The DFG-funded Emmy Noether Junior Research Group `Grammatical Strength in Prosodic Morphology: Typology and Theory' (starting date: May 2019) is hosted by the Department of Linguistics at Leipzig University and led by Eva Zimmermann.
Members
Research group leader: Eva Zimmermann
PhD students: Razieh Shojaei (since March 2020), Ekaterina Medvedeva (since February 2021)
Former student assistants: Eleonore Laubenstein, Matthew Cummins, Bernardas Jurevicius
Former intern: Ekaterina Medvedeva
Output: Presentations, Papers
What this Research group is about
The speech sounds of languages have gradient phonetic properties. The final sound of the English word ‘bad’, for example, has arguably different acoustic properties in contexts like ‘bad times’ and ‘bad guys’ for most speakers: It is partially devoiced in the former context but not in the latter (Zsiga, 2013, 49). The phonological representation that encodes speakers’ grammatical knowledge, however, abstracts away from this particular gradient difference and the same phonemic representation of a voiced obstruent /d/ is standardly assumed in both contexts. This neutralization is a consequence of the fundamental assumption that linguistic representations are categorical: An element is present or not and either has a certain property or not (Chomsky and Halle, 1968). Some phonological phenomena, however, apparently challenge such a view. For example, so-called latent segments attested in many languages only surface if their realization does not create additional marked structure whereas other segments surface faithfully in all contexts – the latent segments thus seem ‘weaker’ than other segments and not fully present in the phonology (Zoll, 1996; Zimmermann, 2018d).
This research group investigates the hypothesis that phonological elements indeed have degrees of strength and that grammatical computation is sensitive to these gradient differences. The different degrees of strength for latent and non-latent consonants in our example thus directly in- fluences the interaction of phonological processes but not necessarily their phonetic realization: Latent segments and non-latent segments behave differently with respect to segment deletion but are acoustically identical. This new perspective that challenges the view of categorical linguistic representations has recently been put forward for single cases studies with different theoretical implementations (Smolensky and Goldrick, 2016; Rosen, 2016; Zimmermann, 2018a). This research group will now develop a general theory and formal implementation of phonological strength that is based on a large-scale typological study.
P1: Competition in Lexical Accent Systems One prediction that follows from assuming phonological strength is competition between elements of different strength. An empirical area where this is immediately relevant are lexical accent systems; an area that has not received a lot of systematic theoretical investigation. In competition-based accent systems, the position of the accent is not phonologically predictable but relies on the lexical identity of the involved morphemes that compete for being accented.
P2: Under- and Overapplication in Tonal Morphology Whereas in accent competition, the phonological strength of two morphemes interact, phonological strength also predicts interactions of general phonological processes of a language with the strength of phonological elements. More concretely, the two strength-related phenomena of lexical under- and overapplication and lexical cooperation predict phenomena that can be summarized as lexical idiosyncrasies: In principle identical phonological elements that are part of different morphemes behave differently. Phonological strength attributes this different behaviour to a representational contrast in the strength of the relevant elements. An empirical domain that is ideal for investigating the predictions of phonological strength for lexical idiosyncrasies is the tonal morphology of Otomanguean languages.
P3: Morphologically Distinct Foot-Templates The interaction of strength of phonological el- ements with the general phonological processes of a language also predicts different degrees of markedness within larger prosodic units. An area where this prediction allows a new theo- retical account to otherwise challenging data are morphologically distinct templates within one language. Templates are requirements about the prosodic shape of a word and play an important role in the productive morphology of many languages.
Members
Research group leader: Eva Zimmermann
PhD students: Razieh Shojaei (since March 2020), Ekaterina Medvedeva (since February 2021)
Former student assistants: Eleonore Laubenstein, Matthew Cummins, Bernardas Jurevicius
Former intern: Ekaterina Medvedeva
Output: Presentations, Papers
- Eva presented an invited talk on Gradient Symbolic Representations and the typology of phonological exceptions in the phonology colloquium at Goethe University Frankfurt on December 2nd, 2020 (slides).
- Razieh presented her GSR account of Japanese pitch accent in the Leipzig phonology reading group (Japanese Lexical Accent) on 26.10.2020.
- Eva presented an invited talk on Phonological exceptions result from gradient constraint violations: An argument for Gradient Symbolic Representations at the Atelier de Phonologie, Paris, on October 7, 2020 (slides)
- Katja presented her GSR analysis of Russian lexical accent in the phonology reading group at Leipzig on July 13, 2020.
- Eva presented a poster `Gradient activity results in gradient markedness: A representational account of phonological exceptions' (long slides/short slides) at virtual GLOW (April 01-03, 2020).
- Eva presented the invited talk `Gradient Symbolic Representations and the Typology of Phonological Exceptions' (slides) at MIT on February 28th, 2020.
What this Research group is about
The speech sounds of languages have gradient phonetic properties. The final sound of the English word ‘bad’, for example, has arguably different acoustic properties in contexts like ‘bad times’ and ‘bad guys’ for most speakers: It is partially devoiced in the former context but not in the latter (Zsiga, 2013, 49). The phonological representation that encodes speakers’ grammatical knowledge, however, abstracts away from this particular gradient difference and the same phonemic representation of a voiced obstruent /d/ is standardly assumed in both contexts. This neutralization is a consequence of the fundamental assumption that linguistic representations are categorical: An element is present or not and either has a certain property or not (Chomsky and Halle, 1968). Some phonological phenomena, however, apparently challenge such a view. For example, so-called latent segments attested in many languages only surface if their realization does not create additional marked structure whereas other segments surface faithfully in all contexts – the latent segments thus seem ‘weaker’ than other segments and not fully present in the phonology (Zoll, 1996; Zimmermann, 2018d).
This research group investigates the hypothesis that phonological elements indeed have degrees of strength and that grammatical computation is sensitive to these gradient differences. The different degrees of strength for latent and non-latent consonants in our example thus directly in- fluences the interaction of phonological processes but not necessarily their phonetic realization: Latent segments and non-latent segments behave differently with respect to segment deletion but are acoustically identical. This new perspective that challenges the view of categorical linguistic representations has recently been put forward for single cases studies with different theoretical implementations (Smolensky and Goldrick, 2016; Rosen, 2016; Zimmermann, 2018a). This research group will now develop a general theory and formal implementation of phonological strength that is based on a large-scale typological study.
P1: Competition in Lexical Accent Systems One prediction that follows from assuming phonological strength is competition between elements of different strength. An empirical area where this is immediately relevant are lexical accent systems; an area that has not received a lot of systematic theoretical investigation. In competition-based accent systems, the position of the accent is not phonologically predictable but relies on the lexical identity of the involved morphemes that compete for being accented.
P2: Under- and Overapplication in Tonal Morphology Whereas in accent competition, the phonological strength of two morphemes interact, phonological strength also predicts interactions of general phonological processes of a language with the strength of phonological elements. More concretely, the two strength-related phenomena of lexical under- and overapplication and lexical cooperation predict phenomena that can be summarized as lexical idiosyncrasies: In principle identical phonological elements that are part of different morphemes behave differently. Phonological strength attributes this different behaviour to a representational contrast in the strength of the relevant elements. An empirical domain that is ideal for investigating the predictions of phonological strength for lexical idiosyncrasies is the tonal morphology of Otomanguean languages.
P3: Morphologically Distinct Foot-Templates The interaction of strength of phonological el- ements with the general phonological processes of a language also predicts different degrees of markedness within larger prosodic units. An area where this prediction allows a new theo- retical account to otherwise challenging data are morphologically distinct templates within one language. Templates are requirements about the prosodic shape of a word and play an important role in the productive morphology of many languages.