Feature 5. Height of the subject

found in question(s): 8a, 8b, 8c, 8d, 9a, 9b, 9c, 10a, 10b, 10c, 10d, 11a, 11b, 11c, 11d, 13a, 13b, 13c, 13d, 13e, 13f, 13g, 13h, 13i, 13j, 13k, 13l, 13m, 13n, 13o, 13p, 13q, 13r

Definition and illustration

It is generally assumed in generative grammar that subjects can have at least two positions. They are initially merged within a projection of the verb (according to the so-called VP-internal subject hypothesis, see, e.g., Kitagawa 1986, for an overview McCloskey 1997). In some languages, subjects have to be displaced to a different position, a position usually identified with the specifier of T, i.e., the head expressing tense and agreement. Thus, in English, while the subject originates low, it surfaces before the finite verb, negation and auxiliaries, suggesting it has been displaced. Concrete evidence that the subject has been in a lower position comes from phenomena where the subject can be shown to occupy a lower position on the surface. In English, existential constructions can be used to illustrate that. Thus, while subjects can occur before the finite verb, they can also occur in a lower position when combined with the existential pro-form there:

(1)

The requirement for the subject to occur in the higher position at the surface is referred to as the EPP (Extended Projection Principle, Chomsky 1981). Originally thought to be universal, it has become clear that the EPP does not hold in all languages. In some, movement to Spec,TP may be optional, in others, there may be no such requirement whatsoever (see, e.g., Alexiadou & Anagnostopoulou 1998). Typlogically, this becomes apparent in languages with dominant transitive SVO order that are dominant VS in intransitive constructions such as Polish (Dryer 2013).

The presence/absence of an EPP-requirement and thus the height of the subject has a strong impact on word order variability and was thought to correlate with the VO/OV-distinction (Haider 2010). Concretely, it has been assumed that VO-languages have an EPP-requirement, thus forcing the subject to vacate its base position. OV-languages, on the other hand, have been argued not to have such a requirement. If the subject is obligatorily moved to Spec,TP, there will be less word order variability in the language, assuming that information-structurally neutral reordering (= A-scrambling) affects the VP but not positions above it like TP (displacement to higher positions is usually thought to involve discourse-marked reordering, which is irrelevant to this debate as it is available in both VO- and OV-languages). Thus, in a VO-language, the subject cannot normally follow the verb, except in special constructions like the existential construction above, where the EPP is satisfied by there and not by the subject. Since the subject is externalized, discourse-neutral reordering with the object will not be possible either. In an OV-language, on the other hand, the subject can remain within the VP. Since such languages are usually taken to have A-scrambling, i.e., discourse-neutral reordering, one expects reordering of subject and object to be possible, i.e., OSV orders. In Fanselow (2020), the predictions are somewhat different. There, word order flexibility essentially hinges on the presence of the EPP, which can be present or absent in both VO and OV languages (see also Adam & Hölzl 2024).

Correlations

The questionnaire tests these predictions in several questions.

Question 8 investigates whether S- and A-subjects can occur in postverbal position in a discourse neutral context (= wide-focus context). This is only expected in languages without an EPP, i.e., OV-languages, but not in VO-languages (recall from above). Prediction 1 is based on Haider (2010), while Prediction 2 is based on Fanselow (2020).

• Prediction 1: V-O → *V-S/A
• Prediction 2: S-V → *V-S/A

Question 9 investigates whether S-subjects can occur in postverbal position and follow another argument in experiencer constructions, with unaccusative verbs, and with passivized verbs in a discourse neutral context (= wide-focus context). All verb types are dyadic unaccusatives, i.e., there is an additional argument that is initially projected above the S-subject (an experiencer or a recipient in a passivized ditransitive). This is only expected in languages without an EPP, i.e., OV-languages, but not in VO-languages (recall from above), where the S-argument would move across the verb to the subject position. Prediction 1 is based on Haider (2010), while prediction 2 is based on Fanselow (2020)

• Prediction 1: V-O → *X-V-S_Theme
• Prediction 2: S-V → *X-V-S_Theme

Question 10 investigates whether S-/A-subjects can follow temporal adverbs. It is assumed that those occupy a position between VP and TP. Hence, subjects should only be able to occur after temporal adverbs if there is no EPP-requirement, i.e., in OV-languages, but not in VO-languages:

• Prediction 1: V-O → *Temp-S/A
• Prediction 2: S-V → *Temp-S/A

This question was more explorative in nature such that one expects tendencies in the indicated directions. Clearly, these will not always be strict correlations given that in some VO-languages like English, temporal adverbials can attach to positions above Spec,TP.

Question 11 investigates whether S-/A-subjects can follow TAM-markers. It is assumed that those occupy a position between VP and TP. Prediction 1 is based on Haider (2010), hence, subjects should only be able to occur after such TAM-markers if there is no EPP-requirement, i.e., in OV-languages, but not in VO-languages. Prediction 2 is based on Fanselow (2020), according to which in a language with the EPP, the S/A argument cannot follow material between T and V.

• Prediction 1: V-O → *TAM-S/A
• Prediction 2: S-V → *TAM-S/A

Question 13: This question was essentially used as a background check to make sure we don't miss any potential postverbal subjects. On the other hand we wanted to make sure that the orders in 8 and 9 are truly neutral. To give a concrete example, in Mongolian, the neutral order with an experiencer verb was acc > nom > V, but when the subject was in focus, the preferred order was nom > acc > V.

See also

This feature is closely connected to several others that explore the consequences of the presence/absence of the EPP, i.e.: 7, 12, 13, 17, 18

References