Abstract

A survey of a number of the world’s languages reveals that only those languages that have a transitive verb used to express possession (i.e., Have-languages) also have a transitive verb ‘need’. No Be-language lacking a transitive verb for possession has a transitive verb ‘need’. This generalization suggests a Hale and Keyser (1993, 2002)–style incorporation approach, whereby nominal ‘need’ incorporates to an unpronounced verbal HAVE, yielding transitive verbal ‘need’.

Introduction

We begin this article by pointing out a striking generalization regarding the existence of transitive verbal ‘need’ among the world’s languages. Our analysis falls within an approach to macroparametric comparative syntax that aims to account for the presence or absence of various grammatical constructions in the world’s languages by linking them to some other property of the grammar. We show here that the presence of a construction with transitive ‘need’ in a given language depends on the presence of a transitive verb of possession in that same language.

Our investigation was prompted by the following insightful question raised by Isačenko (1974:75) in his discussion of counterparts of have and be within Slavic: “Could it be sheer coincidence that those Slavic languages which have become H[ave]-languages . . . have some modal verbs which are unknown to Russian, a B[e]-language?” In this article we argue, in the spirit of Hale and Keyser 1993, 2002 and Noonan 1993, that this is no coincidence, at least for the transitive modal verb ‘need’. In our analysis, we attempt to account for the crosslinguistic facts at issue in terms of an incorporation approach to verbal ‘need’ that involves a silent HAVE.1

We will use the informal term H-language to refer to a language that has an overt counterpart to English have, in the sense of having an overt verb expressing predicative possession (as in Mary has two brothers, Mary has a new car), such that the possessor has nominative case and the possessee is a direct object (with accusative case and no preposition).2B-language will then refer to a language that lacks transitive ‘have’ and expresses predicative possession using ‘be’.

We will broaden the investigation, relative to Isačenko’s, by taking into account a broader range of languages that extends far beyond Slavic. Strikingly, the sort of generalization Isačenko had in mind for Slavic appears to be tenable crosslinguistically. We state our expanded generalization as follows:

(1) All languages that have a transitive verb corresponding to need are H-languages.

Put another way, B-languages do not have transitive ‘need’. By transitive here, we mean (as above for ‘have’) a verb taking a nominative subject and an accusative direct object (with no preposition).

Many languages are like English in having transitive need (as in They need that book). All of them are H-languages; that is, they all have a productive possessive transitive verb ‘have’. Examples are German, Spanish, Czech, Mapudungun, and Paraguayan Guaraní. In contrast, Russian, Latvian, Sakha, Korean, Hungarian, Irish, Peruvian Quechua, and Hindi are B-languages that, in accordance with (1), lack transitive ‘need’.

It is to be noted that (1) is not a biconditional. Transitive ‘need’ implies the presence in the language of transitive ‘have’, but the converse does not hold. The presence of ‘have’ in a language is not enough to guarantee the existence of transitive ‘need’. French, Farsi, Lithuanian, Ancient Greek, Latin, and Albanian are examples of H-languages that lack transitive ‘need’. In other words, the presence of ‘have’ is a necessary but not sufficient condition for licensing verbal transitive ‘need’ in a given language.

In the following section, we present the relevant data and patterns exhibited by a variety of languages. These data highlight the generalization (1) that we address here: namely, that a language cannot have a transitive verb ‘need’ without having ‘have’. We present our analysis of ‘need’ in section 3 and then conclude with a brief discussion of one apparent counterexample in section 4.

Having and Needing: The Facts

A survey of languages suggests that Isačenko’s (1974) intuition regarding the presence of certain modals such as ‘need’ in H-languages versus B-languages extends far beyond Slavic. In this section, we summarize the patterns attested in a number of different languages.

B-Languages

Russian, Latvian, Hindi, Irish, and Hungarian are examples of B-languages lacking possessive ‘have’. They are also languages that lack transitive ‘need’.3

(2) Possessive ‘be’

  • a.

    graphic

    at me.gen will be new book.nom

    ‘I will have a new book.’

  • b.

    graphic

    me.dat is bicycle.nom ‘I have a bicycle.’

  • c.

    graphic

    Mohan gen-near one book be.3sg

    ‘Mohan has a book.’

  • d.

    graphic

    Mari-dat be-3pl hat-poss.3sg-pl(-nom)

    ‘Mari has hats.’

    (Szabolcsi 1994:44)

  • e.

    graphic

    be.past several car at Sean art time dem

    ‘Sean had several cars at that time.’

(3) Nontransitive ‘need’

  • a.

    graphic

    me.dat necessary.fem that book.nom.fem

    ‘I need that book.’

  • b.

    graphic

    child.nom need in your help.prep / *your help.acc

    ‘The child needs your help.’

  • c.

    graphic

    me.dat need.3sg fork.gen

    ‘I need a fork.’

  • d.

    graphic

    me-dat water need

    ‘I need water.’

  • e.

    graphic

    Mari-dat need-poss.3sg is hat-pl-onto

    ‘Mari needs hats.’

  • f.

    graphic

    Mari-dat hat-pl need-3pl

    ‘Mari needs hats.’

  • g.

    graphic

    is necessity by.him with the pen

    ‘He needs the pen.’

  • h.

    graphic

    is.deficient bread from.him

    ‘He needs bread.’

In each of these languages, possession is expressed with the verb ‘be’ and the possessee occurs in the nominative case, while the possessor does not. In Latvian (2b) and Hungarian (2d), the possessor is dative, while in Russian (2a), Hindi (2c), and Irish (2e), it occurs as the object of an adposition.4 These two patterns are widely attested among the world’s B-languages (see Stassen 2009 for further examples and discussion). As suggested by (3), none of these languages uses a transitive verb ‘need’. In Russian (3a), Latvian (3c), Hindi (3d), and Hungarian (3ef ), the “needer” occurs in the dative case, while in Irish (3gh) it is preceded by a preposition.

The Russian, Hungarian, and Irish examples in (3) show that there are (at least) two different ways of expressing ‘need’ in various B-languages. Crucially, none of these predicates are transitive. Although the “needer” in (3b) occurs in the nominative case in Russian, the verb does not assign accusative case to its object; that is, it is not a transitive verb in accordance with (1).

H-Languages with Transitive ‘Need’

Among the contemporary Indo-European languages, the majority of (geographically) European languages are H-languages. A number of them have a productive transitive verb ‘need’. Although it seems that the vast majority of currently spoken languages in the world are B-languages, there are H-languages outside of Indo-European (e.g., some Austronesian, Australian, Central and South American, Khoisan, Nilo-Saharan, and creole languages). In this section, we present examples of transitive possessive ‘have’ and ‘need’ from Indo-European Czech, Spanish, and Swedish, as well as from a Native American Tupian language, Paraguayan Guaraní.5

(4) Transitive possessive ‘have’

  • a.

    graphic

    have.3pl new car.acc

    ‘They have a new car.’

  • b.

    graphic

    Cristina have.3sg a car new

    ‘Cristina has a new car.’

  • c.

    graphic

    Anna have.3sg a new car

    ‘Anna has a new car.’

  • d.

    graphic

    I-have one house

    ‘I have a house.’

(5) Transitive ‘need’

  • a.

    graphic

    your children.nom you.acc need.3pl

    ‘Your children need you.’

  • b.

    graphic

    Cristina need.3sg a car new

    ‘Cristina needs a new car.’

  • c.

    graphic

    Anna need.3sg a new car

    ‘Anna needs a new car.’

  • d.

    graphic

    I-need one house

    ‘I need a house.’

H-Languages without Transitive ‘Need’

Thus far, two clear patterns have emerged: (a) B-languages that lack transitive ‘need’ and (b) H-languages that have transitive ‘need’. We now turn to H-languages that lack transitive ‘need’. Examples are given from French, Albanian, and Farsi.6

(6) Possessive ‘have’

  • a.

    graphic

    I have.1sg a car

    ‘I have a car.’

  • b.

    graphic

    I have.1sg.pres a pencil

    ‘I have a pencil.’

    (Stassen 2009:65)

  • c.

    graphic

    Sean book has.3sg

    ‘Sean has a book.’

(7) Nontransitive ‘need’

  • a.

    graphic

    I have.1sg need of a car

    ‘I need a car.’

  • b.

    graphic

    me.dat need.3sg pencil.nom-def

    ‘I need the pencil.’

  • c.

    graphic

    Sean need to/for one book has

    ‘Sean needs a book.’

As these examples show, the existence of possessive ‘have’ in a language is not a sufficient condition for licensing a transitive verb ‘need’.

Before we move on, let us briefly take stock. Table 1 summarizes the patterns considered thus far. We include a number of languages here that were not discussed above, but for which we have gathered the relevant data.8

Table 1.

Distribution of transitive ‘need’

H-languagesB-languages
Languages with transitive ‘need’ Czech, Slovak, Polish, Slovenian, Croatian, Serbian (dialects), Belorussian, English, German, Yiddish, Luxemburgish, Dutch, Swedish, Norwegian, Icelandic, Spanish, Catalan, Basque, Paraguayan Guaraní, Purépecha (Tarascan), Mapudungun *** 
Languages without transitive ‘need’ Bulgarian, Serbian (standard), Lithuanian, French, Italian, Bellinzonese, Portuguese, Romanian, Farsi, Armenian, Albanian, Latin, Ancient Greek Russian, Latvian, Sakha, Bhojpuri, Bengali, Hindi, Marathi, Irish, Welsh, Scots Gaelic, Georgian, Hungarian, Turkish, Korean, Peruvian Quechua (Cuzco, Cajamarca, Huallaga), Bolivian Quechua, Yucatec Maya, Tamil, Mohawk, Amharic 
H-languagesB-languages
Languages with transitive ‘need’ Czech, Slovak, Polish, Slovenian, Croatian, Serbian (dialects), Belorussian, English, German, Yiddish, Luxemburgish, Dutch, Swedish, Norwegian, Icelandic, Spanish, Catalan, Basque, Paraguayan Guaraní, Purépecha (Tarascan), Mapudungun *** 
Languages without transitive ‘need’ Bulgarian, Serbian (standard), Lithuanian, French, Italian, Bellinzonese, Portuguese, Romanian, Farsi, Armenian, Albanian, Latin, Ancient Greek Russian, Latvian, Sakha, Bhojpuri, Bengali, Hindi, Marathi, Irish, Welsh, Scots Gaelic, Georgian, Hungarian, Turkish, Korean, Peruvian Quechua (Cuzco, Cajamarca, Huallaga), Bolivian Quechua, Yucatec Maya, Tamil, Mohawk, Amharic 

Four logical possibilities present themselves in terms of the distribution of ‘have’ and ‘need’. Yet, as the table makes clear, one of these possibilities is not attested. We have not found a B-language that also has transitive ‘need’. This is the primary generalization that any crosslinguistic account of ‘need’ must come to grips with, and we bear it in mind in our analysis below.

Accounting for ‘Need’

In light of the generalization just mentioned, our analysis of transitive verbal ‘need’ will crucially rely upon the presence of ‘have’ in a given language. We adopt a Hale and Keyser (1993, 2002)–style incorporation approach to transitive verbal ‘need’, whereby nominal ‘need’ raises and incorporates into unpronounced verbal HAVE.9 The derivation we have in mind is roughly schematized in (8).10

(8)

graphic

Incorporation of nominal ‘need’ into verbal HAVE results in the appearance of a transitive verb ‘need’ inheriting the accusative-case-licensing properties of HAVE.11 Strictly speaking, though, ‘need’ is never a verb; that is, it is never exhaustively dominated by the node V.

An examination of the morphology of transitive ‘need’ crosslinguistically lends further support to our Hale and Keyser–style incorporation analysis. Such an analysis might lead one to expect that nominal ‘need’ and transitive verbal ‘need’ will share (root) morphology in languages that have transitive ‘need’; that is, English should not be anomalous in this respect. Table 2 confirms the expectation that nominal and verbal ‘need’ share morphology in other H-languages.12

Table 2.

The form of ‘need’

Nominal ‘need’Transitive ‘need’
English need need 
Swedish behov behoöva 
Norwegian behov behøve 
Czech potřeba potrřebovat 
Polish potrzeba potrzebować 
Slovenian potreba potrebovati 
Spanish necesidad necesitar 
Catalan necessitat necessitar 
Nominal ‘need’Transitive ‘need’
English need need 
Swedish behov behoöva 
Norwegian behov behøve 
Czech potřeba potrřebovat 
Polish potrzeba potrzebować 
Slovenian potreba potrebovati 
Spanish necesidad necesitar 
Catalan necessitat necessitar 

As far as Case is concerned, we can say that in the absence of incorporation, as in (9)–(10), need requires Case from have and in, respectively; a new car receives Case from of.13

(9) ?They have need of a new car.

(10) They are in need of a new car.

On the other hand, for (11), in which need has incorporated to HAVE, we will follow Baker (1988) and say that incorporated need does not need Case, so that HAVE is available to assign Case to a new car, obviating the need for of.

(11) They need a new car.

The structure shown in (8) underlies the derivation of transitive ‘need’ in H-languages such as English, German, Icelandic, Czech, Basque, and Purépecha. The lack of transitive ‘need’ in B-languages (e.g., Russian, Hindi, Korean, Georgian, Celtic languages) follows straightforwardly from an analysis that posits HAVE as a necessary component in the derivation of transitive verbal ‘need’. Put another way, B-languages lack transitive ‘need’ precisely because they lack transitive HAVE.

Our proposal clearly recalls the proposal made by Noonan (1993), who takes the absence of stative transitive verbs in Irish such as ‘know’, ‘respect’, ‘fear’, and ‘love’ to be traceable back to the absence in Irish of transitive ‘have’. She argues for an incorporation approach to these predicates (in the style of Hale and Keyser) whereby an underlying noun incorporates into verbal HAVE (in those languages that have transitive verbal ‘know’, ‘respect’, ‘fear’, and/or ‘love’). Our proposal for transitive ‘need’ is simultaneously narrower and broader than Noonan’s proposal for Irish. It is narrower in that we have said nothing about any (derived) verb other than ‘need’ (which Noonan does not discuss explicitly). It is broader in that her proposal is limited to Irish, whereas ours is formulated in terms of a universal crosslinguistic generalization.

The absence of transitive ‘need’ in Irish is accounted for both by Noonan’s proposal and by ours. The absence of transitive ‘need’ in other B-languages is accounted for only by ours. The absence of transitive ‘know’, ‘respect’, ‘fear’, and ‘love’ in Irish is accounted for only by hers.

The question is whether our proposal for ‘need’ can be generalized to other verbs, and whether Noonan’s for Irish can be generalized to other languages. At first glance, it looks as if Noonan’s caution in confining her proposal to Irish is well-taken, in that Russian, for example, while lacking transitive ‘need’, as previously discussed, does have stative transitives such as xotet’ ‘to want’, ljubit’ ‘to love’, and znat’ ‘to know’ (among others), suggesting that ‘have’ is in fact not crosslinguistically crucial to the underlying representation of ‘want’, ‘love’, and ‘know’ (as opposed to ‘need’). We leave further analysis of these predicates for future research.

An Apparent Counterexample

The analysis presented here appears to account for an important aspect of the distribution of transitive ‘need’ crosslinguistically: namely, it accounts for the lack of transitive ‘need’ in Blanguages, since transitive ‘need’ requires a derivation involving incorporation of nominal ‘need’ to silent HAVE, in the spirit of Hale and Keyser’s work.

One may ask whether this analysis can truly be taken to hold universally. It might be the case that the generalization noted here is not a universal phenomenon, and that a number of other B-languages have a different strategy that would allow them to have transitive ‘need’ despite lacking ‘have’. One language that initially seems to pose a challenge to our generalization is Finnish.

Finnish is canonically considered to be a B-language, since it lacks a morphologically distinct transitive verb equivalent to have for possession and instead uses the verb ‘be’ that is found in locative and existential sentences. It also appears to have transitive ‘need’.14

(12)

  • Transitive ‘need’ in Finnish

  • Minä tarvitse-n sinu-t.

  • I.nom need-1sg you-acc

  • ‘I need you.’

However, a close look at the Case marking on the possessee of possessive ‘be’-sentences reveals a striking difference with respect to the group of B-languages mentioned above.

(13)

  • Possessive ‘be’ in Finnish

  • Minu-lla on *hän /häne-t.

  • I-adess be.3sg *him.nom / he-acc

  • ‘I have him.’

  • (Pylkkänen 1998:4)

As (13) shows, the possessee is accusative in Finnish. This state of affairs contrasts sharply with the nominative case found on possessees in other B-languages (recall the examples in (2)). It appears that the ‘be’ of possession in Finnish is an accusative case assigner. This behavior contrasts with ‘be’ in Finnish locative, existential, and predicational sentences; here, only nominative case is grammatical.

(14)

  • a.

    Locative ‘be’ in Finnish

    Hän / *Häne-t on huonee-ssa.

    he.nom / *him-acc be.3sg room-iness

    ‘He is in the room.’

  • b.

    Existential ‘be’ in Finnish

    Huonee-ssa on vieras.

    room-iness be.3sg guest.nom

    ‘There is a guest in the room.’

  • c.

    Predicational ‘be’ in Finnish

    Hän / *Häne-t on vanha.

    he.nom / *him-acc be.3sg old.nom

    ‘He is old.’

Taken together, these facts suggest that Finnish is neither a canonical B-language nor a canonical H-language. It differs from B-languages in assigning accusative case in possessive sentences, much like H-languages. Yet, like B-languages, and unlike H-languages, it lacks a morphologically distinct verb corresponding to have. The existence of such languages might suggest that the generalization stated in (1) should be rephrased as in (15).

(15) All languages that have a transitive verb corresponding to need are languages that have an accusative-case-assigning verb of possession.

If this is the correct generalization, then Finnish has nominal need incorporating into an accusative-case-assigning BE.

Alternatively, as proposed by Pylkkänen (1998), Finnish is not a B-language at all; it is an H-language that involves incorporation of a silent preposition into ‘be’, just as in H-languages, accounting for its ability to assign accusative case. Yet, for some reason, Finnish lacks a morphologically distinct verb. If this is the correct analysis, then the generalization in (1) still holds.15 The extent to which this generalization directly reflects a property of the human language faculty remains to be determined.16

Concluding Remarks

The crosslinguistic distribution of transitive verbal ‘need’ presented here supports a particularly strong form of Hale and Keyser’s (1993, 2002) proposal; that is, it suggests that a light V + N incorporation analysis is the only way the language faculty has to express verbal ‘need’. More specifically, there can be no primitive (transitive) verbal ‘need’. This raises interesting questions regarding the acquisition of transitive ‘need’ by children. For instance, how do children know that transitive ‘need’ involves lexical decomposition? How is transitive ‘need’ acquired?17 We are suggesting that Universal Grammar only makes one structure available for transitive ‘need’, such that children do not have to learn that ‘need’ involves lexical decomposition; they already know it. Our proposal therefore predicts that children will only acquire transitive ‘need’ after (or around the same time as) they have acquired transitive ‘have’.

A recent study on the acquisition of have and need by 10 English-speaking children suggests that this prediction is borne out. Sugisaki (2009) analyzed 10 longitudinal corpora from the CHILDES database (MacWhinney 2000), which provide a total sample of more than 236,000 lines of child speech. The results of this study revealed that 4 children acquired transitive have significantly earlier than transitive need and that 6 children acquired these predicates at the same age. Crucially, no child acquired transitive need significantly earlier than transitive have. In contrast, this same study shows that the acquisition of transitive want does not follow a similar path. Within this same group of children, the statistical results show that 6 children acquired transitive want significantly earlier than transitive have. These results lend further support to the analysis of transitive ‘need’ presented here, as well as to the analysis of transitive ‘want’ presented in Harves 2008.18 Harves argues that ‘want’ does not rely on a silent counterpart of ‘have’ crosslinguistically, unlike ‘need’, given that a number of B-languages indeed have transitive ‘want’ while lacking transitive ‘need’. In other words, the prediction is that the acquisition of transitive ‘want’ should not depend on the acquisition of transitive ‘have’, in contrast to the acquisition of ‘need’. It remains to be seen whether further acquisition studies of these predicates in other languages will yield the same results.

Notes

1 Full capitals for a word or morpheme will be used to indicate a silent counterpart.

2 In languages that have no overt case marking, we assume that a verb of possession is transitive if neither of its arguments is introduced by a preposition.

3 We thank Rajesh Bhatt for help with the Hindi examples, Anna Szabolcsi for Hungarian, and Benedict Williams for Irish.

4 We note that Russian does have a transitive verb imet’ ‘to have, to possess, to own’. However, it is not used productively to express possession colloquially; rather, it occurs primarily with abstract nouns, for example, in the phrases imet’ pravo ‘to have the right’, imet’ značenie ‘to have significance’, and imet’ v vidu ‘to have in mind’. When it does occur with concrete nouns, the focus is on ownership, as in Ja imeju dom ‘I have a house/I am a homeowner’. This verb is inappropriate in broader ‘have’-contexts since it cannot be used to say, ‘I have a daughter’ #Ja imeju doč’. This seems to disqualify (a silent counterpart of) it from being the target of ‘need’-incorporation, for reasons that will need to be made clear.

One possibility suggested by a reviewer is that ‘need’-incorporation may perhaps be incompatible with imet’ if (a) imet’ is used to express possession of concrete objects that is in some way more permanent, rather than transitory, and (b) transitive ‘need’ is used for transitory possession, capable of expressing a stage-level property of subjects. This leads us to expect that sentences such as #Ja imeju pivo ‘I have a beer’ are unacceptable, which is in fact the case.

Pushing this suggestion further, we note that the following two sentences appear to have different interpretations, depending on whether the sentence involves transitive imet’ or the standard u + GEN construction normally used for possession in Russian (we thank Sonia Kasyanenko for discussion of these examples):

(i)

  • a.

    Včera Ivan imel mašinu.

    yesterday Ivan.nom had car.acc

    ‘Yesterday Ivan was a car owner.’

  • b.

    Včera u Ivana byla mašina.

    yesterday at Ivan.gen was car.nom

    ‘Yesterday Ivan had a car/had access to a car/was a car owner.’

The translations of these sentences suggest that while the u + gen construction can indicate either transitory or permanent possession (i.e., a stage-level or individual-level property of the possessor), transitive imet’ appears to be incompatible with transitory possession. We set aside further investigation of the differences between these two constructions for future research.

5 We thank Anna-Lena Wiklund for help with the Swedish examples and Judith Tonhauser for examples from Paraguayan Guaraní.

6 We thank Dalina Kallulli, Albinot Shaqiri, and Ruth Brillman for help with Albanian, and Sean Corner for the Farsi facts.

7 Note that French also makes use of an impersonal construction with the verb falloir (and a dative “needer”) to express ‘need’. For example:

(i)

  • a.

    Il me faut ces livres.

    it me.dat needs these books

    ‘I need these books.’

  • b.

    Il me les faut.

    it me.dat them needs

    ‘I need them.’

We omit further discussion of this construction here.

8 In addition to our informants/consultants mentioned above, we are grateful to the following for help with our data: David Adger (Scots Gaelic), Mark Baker (Amharic, Mapudungun, Mohawk, Sakha), Eric Besson (Ancient Greek), Ruth Brillman (Yiddish and Basque), Andrea Cattaneo (Bellinzonese), Oana Săvescu Ciucivara (Romanian), Beatriz Fernández (Basque), Carlos Guedes (Luxemburgish and Portuguese), Tricia Irwin and Hyejin Nah (Korean), Janet Klein (Turkish), Neil Myler (Peruvian and Bolivian Quechua), Léa Nash (Georgian), Nikolay Nikolov (Bulgarian), Mira Seo (Latin), Violeta Vázquez-Rojas (Purépecha), Judith Tonhauser (Paraguayan Guaraní and Yucatec Maya), Gunnar Hrafn Hrafnbjargarson, Halldór Sigurðsson, and Jim Wood (Icelandic).

9 This may be supported by Kayne’s (2007) argument that the double behavior of English need, as a lexical verb and as a modal, is inherited from the double behavior of English possessive have as a lexical verb and as an auxiliary, to which need incorporates.

10 We abstract away here from a decompositional approach to ‘have’ as in Freeze 1992 and Kayne 1993, whereby ‘have’ involves incorporation of a preposition into verbal ‘be’. As far as we can tell, the decompositional approach to ‘have’ is compatible with our analysis. The derivation would simply contain an additional instance of head movement, whereby ‘need’ would be the spell-out of [‘need’ + P + BE].

11 The proposal that transitive ‘need’ involves unpronounced HAVE recalls in part Larson, Den Dikken, and Ludlow 1997, Schwarz 2006, and Lechner 2007. However, these analyses have a different focus from ours in that they propose a structure whereby there is a HAVE embedded within a complement to verbal ‘need’, whereas we are focusing on what is found above nominal ‘need’. That is, we are proposing that nominal ‘need’ is the complement of a higher HAVE, a proposal based on our examination of the crosslinguistic facts. If both proposals are correct for some languages, then in those languages a sentence like ‘John needs a new car’ must contain two instances of HAVE, corresponding to the two overt instances of have in John has a (pressing) need to have a new car.

One reviewer suggests that perhaps silent embedded HAVE is all that is necessary to account for the generalization presented here. However, despite the strength of the syntactic and semantic arguments (dating back at least to McCawley 1974) suggesting that intensional transitives such as ‘need’ and ‘want’ involve silent embedded clausal complements, there are a number of arguments against taking this silent clausal complement to necessarily contain a silent embedded HAVE (as opposed to BE) in all languages.

First, as shown in Harves 2008, there is evidence (from temporal adverbial modification) for silent embedded clausal structure even in B-languages such as Russian that lack transitive ‘need’ entirely. This is one of McCawley’s (1974) original arguments in favor of silent embedded clausal structure.

(i) Ivanu byli nužny den’gi do sobranija.

Ivan.dat were.3pl necessary.adj.pl money.nom.pl before meeting

‘Ivan needed some money before the meeting.’

a. There was a time before the meeting at which Ivan needed some money.

b. Ivan’s need was to have some money before the meeting.

Since Russian lacks transitive ‘have’, the silent embedded predicate modified by do sobranija in reading (b) must be something other than HAVE. Second, not all paraphrases of need are best paraphrased with HAVE; for example, John needs a kiss is best paraphrased as John needs to get/be given a kiss and not #John needs to have a kiss (pointed out by Harley (2004) for similar sentences with want). Third, positing a silent embedded HAVE for all intensional transitives, as in Larson, Den Dikken, and Ludlow 1997, will not capture the fact that while transitive ‘need’ does not occur in any B-language, transitive ‘want’ occurs in some (e.g., in Russian and Hindi). For further details, see Harves 2008.

12 We also note here that a number of languages that lack transitive ‘need’ exhibit shared root morphology between nominal ‘need’ and intransitive ‘need’, such as Russian (N nužda/V nuždat’sja), Bulgarian (N nužda/V nuždaja se), and Italian (N bisogno/V bisognare), among others. This may suggest that nominal ‘need’ may incorporate to BE (or some other light verb) in these languages.

13 Case from ‘of ’ is also relevant to H-languages that lack transitive ‘need’, such as French, Romanian, and Bulgarian.

(i)

  • a.

    graphic

    I have.1sg need of a car

    ‘I need a car.’

  • b.

    graphic

    I.nom have.1sg need of a book

    ‘I need a book.’

  • c.

    graphic

    Nikolay.nom has.3sg need of help

    ‘Nikolay needs help.’

Why so many H-languages lack transitive ‘need’ is a question that we leave for future research.

14 We thank Liina Pylkkänen for examples from and discussion of Finnish.

15 One must still account for the lack of nominative Case on possessors in this language as well. We leave this as a question for future research.

16 Preliminary evidence suggests that Hebrew is similar to Finnish. The verb yeS ‘be’ used in possessive sentences also occurs with accusative Case on the possessee (see Boneh 2002 for examples and discussion). Hebrew also appears to have transitive ‘need’. We omit further discussion of Hebrew here, however, since forms of ‘need’ in this language (e.g., carix [benoni] and hictarex [past tense]) exhibit a number of morphosyntactic irregularities compared with other members of their binyanim. We leave an analysis of Hebrew ‘need’ to future research and thank Daniel Harbour and Elisha Nuchi for bringing these facts to our attention.

17 We thank a reviewer for raising this question.

18 One reviewer asks whether an input-based account might explain these acquisition facts as well—in other words, perhaps it is the frequency of have, need, and want in the input that is relevant here. Sugisaki (2009) considers such an account, noting that indeed, transitive have is far more frequent in the input than transitive need. However, transitive want is also far less frequent than have in a number of cases; yet a number of the children acquired want significantly earlier than transitive have. In addition, since several children acquired transitive have and need at the same time, despite large discrepancies in these verbs’ frequency in the input, Sugisaki concludes that frequency cannot be the crucial factor determining the acquisition of these predicates.

References

References
Baker
,
Mark
.
1988
.
Incorporation: A theory of grammatical function changing
.
Chicago
:
University of Chicago Press
.
Boneh
,
Nora
.
2002
.
The grammar of the possession link: Modern Hebrew yeS constructions
. In
Proceedings of ConSOLE 9
, ed. by
Marjo
van Koppen
,
Erica
Thrift
,
Jan
van der Torre
, and
Malte
Zimmerman
,
1
13
.
Leiden
:
University of Leiden, Holland Institute of Generative Linguistics (HIL)
.
Freeze
,
Ray
.
1992
.
Existentials and other locatives
.
Language
68
:
553
595
.
Hale
,
Ken
, and
Samuel Jay
Keyser
.
1993
.
On argument structure and the lexical expression of syntactic relations
. In
The view from Building 20: Essays in linguistics in honor of Sylvain Bromberger
, ed. by
Ken
Hale
and
Samuel Jay
Keyser
,
53
109
.
Cambridge, MA
:
MIT Press
.
Hale
,
Ken
, and
Samuel Jay
Keyser
.
2002
.
Prolegomenon to a theory of argument structure
.
Cambridge, MA
:
MIT Press
.
Harley
,
Heidi
.
2004
.
Wanting, having, and getting: A note on Fodor and Lepore 1998
.
Linguistic Inquiry
35
:
255
267
.
Harves
,
Stephanie
.
2008
.
Intensional transitives and silent HAVE: Distinguishing between want and need
. In
WCCFL 27: Proceedings of the 27th West Coast Conference on Formal Linguistics
, ed. by
Natasha
Abner
and
Jason
Bishop
,
211
219
.
Somerville, MA
:
Cascadilla Press
.
Isačenko
,
Alexander V
.
1974
.
On ‘HAVE’ and ‘BE’ languages (A typological sketch)
. In
Slavic forum: Essays in linguistics and literature
, ed. by
Michael
Flier
,
43
77
.
The Hague
:
Mouton
.
Kayne
,
Richard S
.
1993
.
Toward a modular theory of auxiliary selection
.
Studia Linguistica
47
:
3
31
.
Kayne
,
Richard S
.
2007
.
Some English (and Romance) auxiliaries
. Paper presented at
XVII Colloquium on Generative Grammar
,
Girona
.
Larson
,
Richard S
.,
Marcel
den Dikken
, and
Peter
Ludlow
.
1997
.
Intensional transitive verbs and abstract clausal complementation
.
Ms., SUNY at Stony Brook and Vrije Universiteit Amsterdam
.
Lechner
,
Winfried
.
2007
.
Interpretive effects of head movement. Ms., University of Tübingen
. Available at http://ling.auf.net/lingBuzz.
MacWhinney
,
Brian
.
2000
.
The CHILDES Project: Tools for analyzing talk
.
Mahwah, NJ
:
Lawrence Erlbaum
.
McCawley
,
James
.
1974
.
On identifying the remains of deceased clauses
.
Language Research
9
:
73
85
.
Noonan
,
Máire
.
1993
.
Statives, perfectives and accusativity: The importance of being HAVE
. In
Proceedings of the Eleventh West Coast Conference on Formal Linguistics
, ed. by
Jonathan
Mead
,
354
370
.
Stanford, CA
:
CSLI Publications
.
Pylkkänen
,
Liina
.
1998
.
The ‘be’ of possession in Finnish
.
Ms., MIT, Cambridge, MA
.
Schwarz
,
Florian
.
2006
.
On needing propositions and looking for properties
. In
Proceedings of Semantics and Linguistic Theory XVI
, ed. by
Masayuki
Gibson
and
Jonathan
Howell
,
259
276
.
Ithaca, NY
:
Cornell University, CLC Publications
.
Stassen
,
Leon
.
2009
.
Predicative possession
.
New York
:
Oxford University Press
.
Sugisaki
,
Koji
.
2009
.
On children’s NEEDs
. In
The proceedings of the Seventh GLOW in Asia
, ed. by
Rajat
Mohanty
and
Mythili
Menon
,
225
236
.
Hyderabad
:
EFL University Press
.
Szabolcsi
,
Anna
.
1994
.
The noun phrase
. In
The syntactic structure of Hungarian
, ed. by
Ferenc
Kiefer
and
Katalin É
.
Kiss
,
179
274
.
Syntax and Semantics 27
.
San Diego, CA
:
Academic Press
.