On exceptional stress assignment in Latvian: the case of prefixes

ABSTRACT

In this article, we examine some previously understudied exceptions to the generalization that Latvian assigns stress to the left-most syllable in a prosodic word, specifically those that involve prefixation. We will show that these apparent exceptions in stress assignment follow from the internal structural properties of the word and are a result of attaching the prefix outside the domain where stress is assigned, which is up to the first functional head inside the hierarchy. Our treatment combines the syntactic structure of a neoconstructionist approach to word formation with an optimality theory formalization at the phonological level.

KEYWORDS:

• Stress
• prosodic word
• prefix
• negation
• degree
• quantification

Acknowledgments

We are grateful to Olga Urek and Daira Vēvere for comments on an earlier draft and help with some Latvian data, and to Norway Grants for partially funding this research in the LAMBA project (no. NFI/R/2014/053).

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Throughout, we present examples in Latvian orthography with word stress marked with the respective IPA symbol, a high stroke preceding the stressed syllable: ˈ. Additional phrasal stress is indicated by boldface and further stress by additional underlining. Where present, dots within words indicate syllable boundaries and hyphens show morpheme boundaries.

2. As will become clear in the analysis, we assume a Neo-Constructionist model where the same set of principles and primitives builds both ‘morphological words’ and phrases. Therefore, we distance ourselves from proposals like Williams (2007), where the domain of word formation involves a distinct module from syntax. In this sense, both instances where stress is attracted by a prefix and cases where it is not are cases where one builds complex constituents in syntax, and our contention is that the minimal difference is whether they are contained within the first cycle or not.

3. Interestingly, as the analysis that we will propose predicts, the same unusual stress pattern emerges with other words derived from the same interrogative elements: cf. jeb-’kurš ‘whoever,’ jeb-’kad ‘whenever.’ We are grateful to an anonymous reviewer for this observation.

4. Here, we take the term ‘universal quantifier’ to refer to a semantic object, which can be instantiated in different grammatical categories, and which translates semantically as a logical constant (”) interpreted as “for all members of the class X defined in a context.” The class X is its logical restrictor, and the combination of quantifier and restrictor applies to another expression. Thus, in Ikviens dzied “everybody sings,” the expression translates as “for all members of the class of humans, it is the case that they sing.”

5. A different situation is illustrated by (i), where, despite the difference in marking, both forms display exceptional stress.

(i)  a. paˈtiesi  ‘really, indeed’

   b. paˈtiešām ‘really, indeed’

While the form in -ām is accounted for by the analysis proposed in (33), something additional has to be said about (ia), given that here the case marking cannot be behind the difference. Our proposal is that here the functional layer that is responsible for the prefix being more external relates to the meaning of this form. Coming from an archaic form meaning ‘truth,’ the adverb in (ia) has developed a meaning as a focus emphatic. marker, as in (ii).

(ii) jūra patiesi garoja kā pirts sestdienā

  sea truly smoke.pst like sauna Saturday

  ‘The sea was indeed smoking like a sauna on Saturday’

Here, the meaning of the root is not the literal one but has been turned into a grammatical function marker associated to focus. For this reason, we expect that in the internal structure of the prefixed word, the base contains functional layers that mark focus. The prefix would be attached above these focus layers, which would explain why in this case, but not (32b), it does not attract stress.

6. In optimality theory (Prince and Smolensky 1993), stratal or parallelist, universal violable constraints are ranked in a hierarchy on a language-specific basis. A Generator function proposes a set of output candidate forms for every input submitted to the system. The evaluator function compares these candidates and registers violations of constraints. The form with the least severe violations wins the competition and is selected as the output form.

7. In recent literature, Edgemost constraints are formalized as alignment or anchoring constraints (McCarthy and Prince 1993, 1999). Alignment constraints have undergone some evolution from gradient to categorical (McCarthy 2003; Hyde 2012; Martínez-Paricio and Kager 2015). The difference is immaterial to our concerns here and our analysis works under either conception of alignment. We thus use the constraint schema we expect readers to be more familiar with.

8. OT tableaux are organized as follows: The cell in the top left shows the input, usually given between forward slashes, indicating its lexical/phonemic nature. In this tableau, we use vertical slashes to indicate that this is not the underlying representation, but rather the output chosen at an earlier level of evaluation. To the right, we see the relevant constraints each heading a column. Uninterrupted lines between these columns indicate that the constraints to the left and right are ranked with regard to each other, i.e. the violations, represented by asterisks, incurred on a constraint to the left are more important than those on the right. If an interrupted line separates two constraints, they are not ranked with regard to each other and violations to either side of the dashed line count equally. Below the input is the list of output candidates. Constraint violations incurred by each candidate are given in the same line as the candidate in the column under the constraint violated. The hand symbol points at the winning candidate that is chosen as the output form. All other candidates are discarded. Exclamation marks next to asterisks mark fatal constraint violations. At this point, a candidate loses against another candidate and is excluded from comparison.

9. Finally, Eckert, Bukevičiūtė, and Hinze (1994, 279) mention cases in which stress is found on the third syllable, such as pamaˈzītiņām ‘gradually.’ While these forms deserve further investigation, we suspect that they will be amenable to the analysis proposed here for monosyllabic exceptional prefixes.

Additional information

Funding

This work was supported by the Latvijas Universitate; Norway Grants [LAMBA]; Universitetet i Tromsø.

Notes on contributors

Antonio Fábregas

Antonio Fábregas is full professor of Hispanic Linguistics at the Department of Language and Culture, University of Tromsø – The Arctic University of Norway. His research concentrates on the syntactic analysis of morphological structures.

Martin Krämer

Martin Krämer is professor of General Linguistics at the Department of Language and Culture, University of Tromsø – The Arctic University of Norway. His research focuses on phonology.

Anna Vulāne

Anna Vulāne is full professor of Synchronic Linguistics at the Faculty of Education, Psychology and Art and Leading Researcher at the Latvian Language Institute, University of Latvia. Her research concentrates on word-formation and morphology.