This book is a paperback edition of the hardcover monograph pub-lished by Oxford University Press in 2005, a much-revised version of Los's doctoral dissertation Infinitival Complementation in Old and Middle English. Modifications with respect to the first edition are limited to the updating of some references and the correction of some typos and stylistic oversights.

Los's study is basically an in-depth investigation of Old English (OE) infinitival structures. Its results convincingly defy two central assumptions in previous studies on the history of the English infinitives: (i) the wider distribution of the to-infinitive in present-day English is the product of its encroachment upon the domain of the bare infinitive; and (ii) the to-infinitive was originally a noun that underwent recategorization as a verb in the Middle English (ME) period.

Los's rejection of (i) is based on a careful examination of the relevant OE data. Her main argument is that to- and bare infinitives systematically differ in their syntactic distribution, and thus never com-pete with each other. Los concurs with the traditional view that the primary function of OE to-infinitives is that of purpose adjuncts, but refutes the idea that bare infinitives could be purposive toomost of the examples given by Callaway (Reference Callaway1913) are simply the product of slavish translations of Latin originals. The only exception is constituted by bare infinitives with motion verbs and verbs of rest (cuman ‘come’, faran ‘go’, licgan ‘lie’, standan ‘stand, etc.), where, according to Los, the interpretation at stake is not “purpose,” but rather “imperfect, progressive aspect.” However, the purposive function was not restricted to to-infinitives, given that they overlap with (non-spatial) to-PPs and subjunctive clauses with œt.Footnote ¹ From Los's perspective, this suggests that the OE to-infinitive encroaches upon the domains of to-PPs and subjunctive œt-clauses, against the traditional view.

Evidence from Los's main hypothesis comes from the verbs of persuading and urging (biddan ‘ask’; cierran ‘persuade’; lœran ‘advise, teach’; tyhtan ‘induce’, ‘urge’; undereodan ‘subject, force’; etc.), in the context of which the bare infinitive is also completely excluded, whereas both to-PPs and subjunctive œt-clauses are allowed. Los contends that these examples are the product of a syntactic reanalysis of a primitive purpose adjunct as a goal argument, as shown by the possibilities of extraction out of the to-infinitive. Los suggests that this process may have been triggered by “the fact that verbs of persuading and urging require a controlled subject” (p. 67).

Los's analysis of the structures in which the to-infinitive occurs with intention verbs also supports her main argument, despite some complexi-ties. One is related to the existence of a relatively important number of verbs, namely seventeen, attested with both the bare and the to-infinitives; among them are: earnian ‘deserve, strive’; fon ‘attempt’; giernan ‘desire’; leornian ‘learn’; secan ‘seek’; wulnian ‘desire’; etc. Another has to do with the syntactic function of the to-infinitive in these examples, which is clearly that of a theme argument, not a goal. Los accounts for these facts by claiming that the spreading of the to-infinitive to this particular syntactic environment was triggered by three different but parallel developments: (a) the reanalysis of to-infinitives and sub-junctive œt-clauses as goal arguments of the small subset of intention verbs known as the conative verbs, which served as a model for the rest of the verbs in the class;Footnote ² (b) another process of reanalysis through which to-infinitival adjuncts after nominal objects were reinterpreted as the theme-argument of the matrix verb (Bock Reference Bock1931); and (c) the analogy with the subjunctive œt-clause, already a theme-argument with verbs of this kind.

The role played by the subjunctive œt-clause in the development of the to-infinitive is also prominent in Los's treatment of the theme argu-ment of verbs of commanding and permitting (bebeodan ‘order’; beodan ‘order’; don ‘make’; hatan ‘command’; sellan ‘grant’; tiian ‘permit’; etc.). These verbs typically have either a three-place frame as in 1 or a two-place frame as in 2, which might suggest that there is competition between bare and to-infinitives in OE.

1. (1)

2. (2)

However, Los convincingly shows that the two structures above must be kept distinct, and that the emergence of the to-infinitive in 1 can be attributed to its overlapping with subjunctive œt-clauses.

As indicated above, the second traditional issue Los's study addresses is whether the to-infinitive was originally a noun that under-went recategorization as a verb in the ME period. Her arguments against this are solid, since, as she shows, to-infinitives exhibit an uncon-troversial verbal behavior. They are consistently incompatible with determiners, and their internal arguments never show up in the genitive case. Moreover, different pieces of evidence reveal that to cannot be treated as a preposition in Old English, contrary to what is usually claimed: (i) to must appear immediately adjacent to the infinitive, whereas some material may intervene between a preposition and its nominal complement; (ii) prepositions may be deleted in the second conjunct of coordinate structures, while to must be necessarily present; (iii) PPs may precede as well as follow the matrix verb, but to-infinitives obligatorily follow it.

As for the exact category of OE to-infinitives, Los adopts Chomsky's (Reference Chomsky1981) Government and Binding (GB) as a framework, and classifies them as Complementizer Phrases (CPs), that is, full-fledged clauses. The reasons for this are theoretical rather than empirical: in infinitival structures of the kind discussed by Los, GB postulates a null subject PRO, whose licensing conditions are restricted to the CP-domain. This, however, ties in nicely with Los's crucial hypothesis that the to-infinitive is the non-finite counterpart of the subjunctive œt-clause:

In the last decade, however, several studies have suggested that the categorial status of the infinitival clause is irrelevant for the licensing of PRO (Chomsky & Lasnik Reference Chomsky, Lasnik, Jacobs, von Stechow, Wolfgang and Venneman1993, Bošković Reference Bošković1997, Martin Reference Martin2001, etc.). This would seriously question the need to invoke categorial selection to explain the distribution of the different infinitival structures attested in a given language.

The empirical evidence Los provides for this particular claim is not robust enough either. Her first piece of evidence is the existence of to-infinitival relatives in Old Engish. Los's reasoning is that since relative movement of the silent operator targets the specifier of C, these structures must be CPs. However, this argument is undermined by studies such as Law Reference Law1991 and Bošković Reference Bošković1997, which indicate that zero null-operator relatives are Inflectional Phrases (IPs). Los relies also on the fact that to-infinitives as well as œt-clauses seem to undergo extra-position in Old English. However, as she herself observes, the extraposed position does not make them opaque for object scrambling, that is, the displacement of the complement of the infinitive to a position within the matrix clause. Given that the same structure is found in present-day West Germanic languages, where scrambling is strictly clause-bound, Los wonders whether the link between CP and PRO is necessary at all, but does not dispense with it.Footnote ³

With respect to the exact nature of OE to, Los follows Pullum (Reference Pullum1982) and classifies it as a non-finite modal. However, she departs from Pullum's analysis in treating to as an instance of tense (t), being consistent with her theoretical framework. Given, as we saw, that to and the infinitival form are necessarily adjacent, Los assigns to the morpho-logical status of a bound morpheme (an affix), and concludes that it is connected with T by means of what Chomsky (Reference Chomsky1995 and subsequent work) labels Covert Movement, that is, movement without morpho-phonological reflexes. Los contends that the kind of T encoded in to is, in all relevant respects, the one corresponding to subjunctive forms, namely Noonan's (Reference Noonan and Shopen1985) Determined Time Reference (DTR), which fits her main thesis about the encroachment of the to-infinitive upon œt-subjunctive clauses. Note in this respect that adopting something like Bošković's (Reference Bošković1997) unspecified tense for irrealis complementson which, according to him, the licensing of PRO is dependentwould have had some advantages for Los's analysis: the controversial data about the CP status of OE to-infinitives would no longer be relevant, and a more uniform theoretical framework for the licensing of both PRO and to would be provided.

The final chapters of Los's monograph are devoted to discussing some of the major changes that affect to-infinitives in Middle English. These changes are basically what Los treats as the morphological degrammaticalization of to, which allows it to appear as a free morpheme and undergo overt moment to T, on the one hand, and the emergence of Exceptional Case Marking (ECM) structures with verbs of thinking and declaring, on the other. Los's basic argument is that this latter construction, promoted by the loss of the Verb Second Contraint (V2), is not part of the core grammar of English, and must be strictly distinguished from the ECM found with verbs such as want or prefer, which is the product of the reanalysis of object control structures after verbs of commanding and permitting.Footnote ⁴ Los closes her monograph with a discussion of the causes for the disappearance of the OE indefinite pronoun man in the fifteenth century, which she links to the demise of V2 as well as to the rise of arbitrary PRO for the expression of contentless subjects.

The main achievement in Los's study is, in my view, that she offers a principled account of data that have been traditionally considered as chaotic. She shows that the problems that derive from the use of limited corpora and the absence of native speaker judgments may be overcome by a careful examination of the data as well as a rigorous application of procedures that are current in synchronic linguistics. Furthermore, her stimulating conclusions confirm the idea that grammars are intricate systems where changes in one domain may provoke apparently unrelated changes in another, different domain. The new edition of The Rise of the to-Infinitive must be welcomed.