Constructional change in Germanic

This chapter tracks several of the major changes in English and German word order and accounts for them in terms of constructional change as formulated in Chapter 3. It argues that the changes are relatively simple in constructional terms, although the superficial results are quite dramatic. Topics include clause-initial position, V2, VP-initial and VP-final verb position, the loss of V2 and case marking in English, and verb clusters in Continental West Germanic.

On the syntax of the can’t seem construction in English

/The properties of the English can’t seem construction call for a syntactic resolution of the syntax-semantics mismatch it exhibits. This chapter shows the can’t seem order must be derived from a [seem to [ . . . not can VP ] ] structure. Insights into the derivation come from verb clusters in Germanic OV languages, with complex verb formation and clustering verbs like can and seem playing a central role. Together with infinitival to, dative to, and downward entailing elements, these are instrumental in creating remnant constituents, triggering pied-piping and smuggling a remnant constituent up into the structure, until each element can reach its final landing site. Restrictions fall out from the particular sequence of merge which must hold for convergence, and from the role each element must play. The English derivation in turn sheds light on a potential syntactic resolution of a syntax-phonology mismatch with “displaced” zu in German verbal clusters.

An incremental approach to verb clusters in German

This paper presents an incremental approach to verb clusters in German which radically differs from standard HPSG accounts. While the common assumption is that the verbs in subordinate clauses form clusters and accumulate all their valence requirements on a SUBCAT list, the assumption in this paper is that the arguments in verb final clauses are encapsulated incrementally into syntactic and semantic structures before the verbs are attached. The proposed analysis is in line with psycholinguistic findings. A grammar fragment of German demonstrating an implementation of the analysis is presented.

De verwerving van de groene en rode woordvolgorde in Vlaanderen* : Een descriptieve, methodologische en theoretische aanvulling bij Meyer & Weerman (2016)

The acquisition of [PART+AUX] and [AUX+PART] word order in Flanders. A descriptive, methodological and theoretical addition to <italic/><italic/>Meyer & Weerman (2016)This paper presents new data on the acquisition of verb clusters in Flemish children. The data were collected by means of a sentence repetition task and the results are in line with the development path for verb clusters in Dutch children as proposed by Meyer & Weerman (2016). While Flemish children also show a development from more 2-1 orders in the youngest group to more 1-2 orders in the older group, this development seems to happen more slowly in Flemish children than in Dutch children. In spite of the fact that the results of both the Flemish and the Dutch study refute an analysis that takes the input adult language as the main factor in verb cluster formation in children, the Flemish data suggest that the higher frequency of 2-1 orders in the Flemish context could help to explain why 1-2 orders are acquired more slowly in Flemish children than Dutch children. In addition, this paper also discusses the results of a production test in Flanders that shows a high preference for 2-1 orders until the age of 7, thus questioning the type of linguistic skills that are assessed in a sentence repetition task.

Verb Class Induction with Partial Supervision

Dirichlet-multinomial (D-M) mixtures like latent Dirichlet allocation (LDA) are widely used for both topic modeling and clustering. Prior work on constructing Levin-style semantic verb clusters achieves state-of-the-art results using D-M mixtures for verb sense induction and clustering. We add a bias toward known clusters by explicitly labeling a small number of observations with their correct VerbNet class. We demonstrate that this partial supervision guides the resulting clusters effectively, improving the recovery of both labeled and unlabeled classes by 16%, for a joint 12% absolute improvement in F1 score compared to clustering without supervision. The resulting clusters are also more semantically coherent. Although the technical change is minor, it produces a large effect, with important practical consequences for supervised topic modeling in general.

Een corpus waar alle constructies in gevonden zouden moeten kunnen worden?* : Corpusonderzoek met behulp van automatisch gegenereerde syntactische annotatie

In this contribution, I discuss the use of automatic syntactic annotation in Dutch corpus research, using a case study of five-verb clusters. Large amounts of text can be annotated automatically, but the parser makes mistakes, while correct annotation is very important in linguistic research. How much of a problem is this, and how can we learn about the extent of these parsing mistakes? There are several approaches to evaluating the quality of automatic annotation for specific research questions. I demonstrate these approaches for the case study at hand, which will help us to make claims based on automatically annotated corpus data with greater confidence.

A new and challenging view on verb clustering*

This review article discusses Lotte Dros-Hendriks’ PhD thesis Not another book on Verb Raising (2018). A ground-breaking book, as it provides a simple, principled account for the geographical distribution of the various word orders found in so-called ‘verb’ clusters in the Dutch dialects. It will be shown, however, that the account is based on a number of controversial and insufficiently motivated premises. This article raises a number of potential problems for these premises, which should find a satisfactory solution before we can wholeheartedly adopt the core insights provided by this interesting work.