Pieces of the perfect in German and older English
     The perfect in older English looks deceptively similar to that in Modern German. Unlike in Modern
English, the auxiliary alternates between have and be, based largely on the semantics of the main verb
(compare 1a and 1b). However, we have determined in a corpus study that the appearance of be in Middle
and Early Modern English was subject to a series of additional restrictions which have nothing to do with
familiar factors like agentivity and telicity. In particular, it was not used with the perfect in durative, iterative
and past counterfactual contexts (see also Rydén and Brorström, 1987; Kytö, 1997). Interestingly enough,
auxiliary have was used in these contexts, even with verbs like come which otherwise appeared only with be
(as in the counterfactual 2a). This is distinctly unlike German, where such factors play no role in auxiliary
selection, and a verb like kommen `come' can only ever appear with sein `be' (see 2b). Our analysis for these
facts is that perfects with have and be in older English had crucially distinct underlying structures. The have
perfect appears to have had a compound temporal structure similar to that familiar from the Modern English
perfect. In addition to T, which shows up as the finite inflection on the auxiliary, there is what Klein (1992)
calls perfect aspect, denoting that the interval set by T (Klein's Topic Time) is after the temporal interval of
the eventuality expressed by the VP/vP. In the be perfect, however, we will argue that this intermediate level
was missing. It was a simple stative construction of the copula with a resultative participle. Any implication
of anteriority to the Topic Time was derived from the meaning of the resultative, not explicitly encoded in
temporal terms. We will propose that be was restricted from the contexts noted above, because those are just
the ones where an eventive past is required rather than a result state, and will show that this also accounts
for additional peculiarities of the be perfect. This analysis receives further support from a comparison with
the Modern German stative passive in 3a. The stative passive differs from the formally identical be perfect
in 3b not only in voice, but also in tense/aspect. It lacks the layer of perfect aspect in addition to finite T,
just as we claim for the older English be perfect (see Kratzer, 2000). What is especially interesting is that
the past subjunctive form of this stative passive yields a present counterfactual meaning (see 4a), while the
formally identical modern German be perfect yields a past counterfactual (see 4b). The older English be
perfect is again parallel.
     The contrast between 3a and 3b demonstrates the well-known difficulties surrounding the role of the
participle in complex constructions. How do we ensure that auxiliary be + past participle will yield a sta-
tive passive in 3a but a perfect in 3b, without abandoning compositionality or flouting the morphological
evidence to posit two distinct participles? We will argue that we can avoid this dilemma if think of the
auxiliary ist and the participial morphology not as lexical items with fully specified syntactic and seman-
tic features which combine to form the syntactic and semantic structure, but as Vocabulary Items in the
Distributed Morphology sense, which are underspecified and inserted into a structure that is already fully
determined syntactically and semantically. The syntactic structures underlying and 3a 3b are indeed dis-
tinct, such that the latter has an additional Aspect head, yielding the difference in the temporal semantics
by normal compositional means. However, as there is no Vocabulary Item specified for this more highly
determined environment, the same underspecified auxiliary and participial morphology is inserted in both
structures Embick (see 2004, for related ideas on participial morphology). We will show that this approach
can be extended to the older English data as well, and to dealing with cross-linguistic variation in the per-
fect and related constructions more generally. It implies that the `perfects' of familiar languages are not
representatives of a single, monolithic category, but rather complexes built out of a number of simpler com-
ponents. Languages differ in which components they use, how they combine them, and how they spell them
out.
  (1)    a. as   ha þreo weren ifolen onslepe. . .
            when they three were fallen asleep. . .
             `When the three of them had fallen asleep. . . ' (CMANCRIW-2,II.272.440)
        b.   . . . huanne hi heþ wel yuoZte
             . . . when he has wel fought
             `. . . when he has fought well' (CMAYENBI,252.2315)
  (2)   a.   And if þow hadest come betyme, he hade yhade þe maistre
             and if you had        come timely he had had the master
             `And if you had come in time, he would have prevailed.' (CMBRUT3,227.4102)
        b.   Wenn du gekommen wärest/*hättest. . .
             if       you come       were/*had
             `If you had come. . . '
  (3)   a.   Das Buch ist geöffnet.
             the book is opened
             `The book is opened.'
        b.   Das Buch ist angekommen.
             the book is arrive-ptc
             `The book has arrived.'
  (4)   a.   Wenn das Buch geöffnet wäre. . .
             if       the book opened were. . .
             `If the book were opened. . . '
        b.   Wenn das Buch angekommen wäre. . .
             if       the book arrived        were. . .
             `If the book had arrived.'


References
Embick, David. 2004. On the structure of resultative participles in English. Linguistic Inquiry 35:355­392.

Klein, Wolfgang. 1992. The present perfect puzzle. Language 68:525­551.

Kratzer, Angelika. 2000. Building statives. In Proceedings of the 26th annual meeting of the Berkeley
  Linguistics Society, 385­99.

Kytö, Merja. 1997. Be/have + past participle: the choice of the auxiliary with intransitives from Late Middle
  to Modern English. In English in transition: corpus-based studies in linguistic variation and genre styles,
  ed. Matti Rissanen, Merja Kytö, and Kirsi Heikkonen, 19­85. Berlin: Mouton de Gruyter.

Rydén, Mats, and Sverker Brorström. 1987. The Be/Have variation with intransitives in English. Stockholm:
  Almqvist & Wiksell International.