The Impersonal In Breton

JCeltL, 7 (2002), 1-39 The Impersonal in Breton1 Steve Hewitt UNESCO s.hewitt@unesco.org Abstract Breton verbal syntax is simultaneously VSO and V-2, or more precisely [P = predicate syntagm] PSO/XPSO and [T = tense] T-2. “Bare” presentations begin with a predicate syntagm; “lead-in” presentations with a non-predicate constituent [X = S/O/ADV/CIRC, etc.], which may be either thematic or focused. In “bare” presentation, the negative tense particle ne is sufficient to fill the first position in order to satisfy the T-2 constraint. But in the affirmative, with simple verbs, a dummy auxiliary “do” arises; with auxiliary structures (copula, existential, compound tenses), there is AUX-PRED > PRED-AUX inversion The apersonal conjugation, formally identical with the 3SG, marking tense, but not person or number, is used before expressed nominal subjects, and after initial subjects in the affirmative. The personal conjugation marking tense, person and number represents the inclusion of post-verbal subject pronouns; it is also used after initial subjects in the negative (subject agreement). The impersonal forms in –r and –d constitute a seventh form in the personal conjugation, referring to some indeterminate human subject. In Breton these forms are fully active, may not be used with agentive phrases, and are best translated with French on / English one, even though there is no corresponding pronoun in Breton. Impersonal constructions include the existential, meteorological phenomena, indirect impersonal verbs of the type “it pleases me”, and the impersonal compound passive dañssed e≠veż “es wird getanzt”. With none of these constructions is it possible to reformulate with an initial subject pronoun. A possible analysis is that what appear to be 3SG verb forms may actually be the independently required apersonal conjugation, with no person/number reference, and that these constructions are thus subjectless. 1. 2 Functioning of the Breton verb If there ever was “un système où tout se tient”, it must be Breton, given the difficulty of examining any one aspect of the language 2 Steve Hewitt without touching on a good deal of the rest. So before looking at the various phenomena which may be described as “impersonal” in Breton, it may be useful, in order not to bewilder non-iniatiates, to provide some basic information concerning the verbal system of the language.3 1.1 Tenses Breton has the following six tenses: present, future, preterite, imperfect, “present” conditional (1, potential), and “past” conditional (2, hypothetical): Regular verbs (1, 2, 3 singular and plural, and the “impersonal” form) present -an -omp -ez -et -ont -er imperfect -enn -emp -es -ec’h -e -ent -ed L -it/-oc’h future -in -ffomp L -imp -i -ffet L -ot/-oc’h -o -ffont L -int -ffer (L –or) preterite -is/-jon -jomp -jout -joc’h -as -jont -jod (MBr -at) conditional 1, potential -ffenn -ffemp -ffes -ffec’h -ffe -ffent -ffed conditional 2, hypothetical -jenn -jemp -jes -jec’h -je -jent -jed The preterite is hardly used nowadays in spoken Breton, and is barely understood, apart from the 3SG. The difference between the two conditionals is more or less clear-cut, according to dialect. Some authorities insist that a sequence of tenses must be observed (present – conditional 1; past tenses – conditional 2). There is a clear tendency to use conditional 1 (potential) for simple tenses: (1) ma≠welffenn aneżañ e larffenn deżañ if I.would.see1 him.O e I.would.say1 to.him If I saw him I would tell him. The Impersonal in Breton 3 and conditional 2 (hypothetical, historically a pluperfect) for perfect (compound) tenses: (2) ma miche gweled aneżañ e miche lared deżañ. if I.would.have2 seen him.O e I.would.have2 said to.him. If I had seen him I would have told him. The conditional tenses are also used where French requires a subjunctive. The relatively close correspondence between the Breton and English tenses will allow us to gloss the Breton tenses with simple English tenses (English past = Breton imperfect) rather than with more opaque abbreviations (1 and 2 are used to distinguish the two conditionals). Alone among the Celtic languages, Breton has true “compound” or perfect tenses analogous to those of French or English. According to Hemon (1975, p. 245), they arose towards the beginning of the Middle Breton period (1250-1650); they are formed with the auxiliary “be” (copula) or “have” and the past participle. The choice between “be” and “have” is similar to French; however, “be” is used for changes of state such as “grow”, “cool down”, and for a handful of other verbs such as “last”, “cost”. For reflexive verbs, the choice between “be” and “have” depends on the dialect. There are also experiential “double perfect” tenses similar to those of southern French: (3) bed e meus butuned beked daou=bakad bemdeż been e I.have smoked up.to two pack every.day I used to smoke up to two packs a day. Finally, there is a progressive construction,4 which is formed from the situative forms (where separate forms exist) of the verb “be” and the infinitive of the lexical verb, preceded by the progressive infinitival particle o≠ (é≠ in some areas, including central Treger), which triggers the “mixed” mutation (lenition/provection). The simple/progressive distinction is obligatory and closely resembles the distinction in English (especially late 18th-century English); however, rather than highlighting the “contingency” of a process, is as increasingly the case in English, the Breton progressive appears to lay stress on “control by the subject”. 4 Steve Hewitt 1.2 The verb “be” In the present and imperfect, the verb “be” has separate forms for the copula (auxiliary), the situative verb, and the existential auxiliary (separate form for the present only). There is also a special form in the present (for both the copula and the situative verb) which is used after a subject in the affirmative. Finally, there are separate habitual forms covering the situative, copula, and the existential in the present and the imperfect. There are no separate forms for these various functions in the other tenses, but the syntactic differences regarding the place of the subject (AUX PRED S; V S COMP) are preserved. Verb beżañ “be”, PP bed “been”: Internal functional articulation in the present and imperfect SIT S__PRES.AFF PRES EMAÑ COP so a EO EXIST AFF NEG sob eus SIT IMPERF COP C OA EXIST OA d VEŻ IMPERF.HAB VIJE lower case: single form SMALL CAPS: full paradigm – 1, 2, 3 SG/PL; impersonal form (a) 3SG/PL only in the E; (b) L ez eus; (c) NW EDO, EVEDO; (d) L VEZE PRES.HAB So and eus are unique, invariable forms; the other forms have person/number variants (EMAÑ has 3SG and 3PL only in the E of the Breton-speaking area, which is the historical situation; in the W, and in literary Breton, analogous forms exist for all persons). The imperfect situative forms EDO are current only in the NW (preferably on a base EVEDO); these forms are usual, but not obligatory, in literary Breton. The distribution of EMAÑ and EO obeys syntactic rather than semantic criteria in the Vannetais region (SE), and functions according to yet another, poorly understood, system in the central S region (see Hewitt 1988). 1.3 The verb “to have” Among the Celtic languages, only Breton and Cornish have developed a verb “have”. It consists of what are historically proclitic oblique personal pronouns and the existential form (in the present; for the other tenses, the general form is used) of the verb “be”: m-eus [mihi-est, to.me–there.is] “I have”; however, it is no longer perceived as The Impersonal in Breton 5 [to.me–there.is], but rather as a simple, highly irregular verb. As it is derived from the verb “be”, it has separate habitual forms for the present and imperfect. In view of its unique origin, with its completely different morphology from all other verbs, it has no apersonal conjugation for use after a subject in the affirmative or before an expressed nominal subject (see section 2). However, the great mass of “central” dialects along the NE–SW axis have developed personal endings for the plural: hon eus [to.us there.is] (> hon eus-omp) > neusomp, or (> hom eus > hom eus-omp) > meusomp > meump “we have”. The verb “have” is used both as a lexical verb “to possess” and as an auxiliary “have” + PP. As a lexical verb “to possess”, given its origin (to.me–there.is) it is understandably reluctant to allow definite objects; the infinitive for this meaning is kaoud “find”. The infinitive for auxiliary “have” is beżañ, the same as for “be”. The past participle for both the lexical verb and the auxiliary is bed “been”, with the distinction being made by the choice of auxiliary: (4) ...on bed ...I.am been ...I have been (5) ...meus bed ...I.have been ...I have had 1.4 Word order and information structure: presentation types Word order in Breton is often described as being “very free”. In reality, it is above all the choice of the initial constituent which is relatively free, with the order of the remaining constituants depending essentially on that choice. The Celtic languages provide classic examples of VSO typology, which it would appear useful to refine as either TSO or PSO, where T = a constituent bearing a tense-marker: V.T or AUX.T V or AUX.T N, and P = a verbal or nominal predicate phrase, including an auxiliary where applicable: V.T (INF RA.T) / AUX.T PRED (PRED AUX.T); PRED = PP / ADJ / N. VSO order is fairly strict in Irish and Welsh (however, Jones & Thomas 1977 adopt an SVO (more specifically T S PRED O) analysis for Welsh). With regard to Breton, 6 Steve Hewitt the question is controversial; some authors (Timm 1989) assume a basic VSO order, while others (Varin 1979) underscore the relative frequency of SVO. For reasons similar to those of PhillipakiWarburton 1985 for Modern Greek, it seems to me that with no particular context, the truly neutral order of Breton is nevertheless VSO, even though it may be the case that for contextual reasons, SVO is actually more frequent. Another reason for taking VSO as the basic order is that it is the required order in embedded clauses, at least historically and in literary Breton. However, since at least the 18th century, an alternative SVO order is possible in “real, factual” complement clauses such as “I think that John will come”, while only VSO is possible in “virtual, possible, intentional” complement clauses such as “I want John to come”. While it is condemned by purists, it is curious to note that this alternative SVO order is possible only in complement clauses where it is obligatory in formal Arabic (another 5 VSO / SVO language), but impossible in those where it is not allowed in Arabic, contrary to what might be expected if that innovation were due solely to interference from French.6 We thus have a dichotomy between a “bare” presentation: PS(O...) (initial predicate phrase), in which there is no great articulation of the information load, and various “lead-in” presentations: XPS(O...), where X = S / O / OP / CIRC / ADV, etc. In these, the initial X may be either thematic (topical) or rhematic (focused, carrying a major sentence stress); in the latter case, the focusing of a further constituent (by another major sentence stress) is rare, although not prohibited, since it is perceived as a kind of information overload: (6) Information structure XPS(O...): THEME FOCUS FOCUS FOCUS (normal) THEME (initial focus) FOCUS (information overload) This is why it is rare (despite the misleading examples of Breton textbooks) to have an expressed nominal subject following a verb which is preceded by a focused initial object (focus indicated in bold); (7) with a focused initial object and a pronominal subject incorporated in the verb is normal, but the same sentence with an expressed nominal subject (8) is unusual (there is a certain amount of The Impersonal in Breton 7 dialect variation in this regard; the dialects of central Brittany are more open to this type of sentence): (7) krampouzh a=zebront pancakes a they.eat They eat pancakes, it’s pancakes they eat (8) !krampouzh a=zebr ar=vugale pancakes a eatº the children the children eat pancakes 1.5 Predicate structures Under this heading, we find two oppositions: on the one hand, between the simple verb structure, V.T (INF RA.T), and various auxiliary structures, AUX.T PRED (PRED AUX.T); and on the other hand, between the simple verb structure and the double or periphrastic structures. The auxiliary structures include the compound tenses (see section 1.1), the copula, and the existential (see section 1.2). They all share the same normal syntax ...AUX PRED S (no subject in the case of the existential; the entity whose existence is predicated is analysed as being PRED and the existential operator so (AFF) / eus (NEG) as AUX; see section 4.1), and are all subject to AUX/PRED inversion (see section 1.6). (9) Auxiliary structures: AUX.T PRED (PRED AUX.T): compound tenses compound passive copula existential The double or periphrastic structures include the progressive construction (see section 1.1) and constructions with an initial dynamic verb phrase (DVP) + “activity-do”. These double structures link a grammatical (syntactic) verb (“be.SIT” in the case of the progressive; “activity-do” for the other constructions) to a lexical (semantic) dynamic VP, the only instance in which it seems useful to postulate a VP in Breton. Both types operate only with dynamic (nonstative) lexical verbs. 8 Steve Hewitt (10) Double (periphrastic) structures: EMAÑ (“be.SIT”) o DVP: progressive o DVP EMAÑ (“be.SIT”): progressive DVP a OBER “activity-do”: dynamic initial VP 1.6 The T-2 constraint: tense-bearing element in second position Breton has a double heritage: on the one hand, it has inherited a typology from insular Celtic; on the other hand it appears to have become associated, through the medium of Old French, with a continental V-2 Sprachbund (verb in second position; in Breton this applies only to matrix clauses) which appears to have spread out from a Germanic home. Given that the two formulas VSO and V-2 are a priori incompatible, one is tempted to reformulate them as PSO and T-2, for it is only by viewing them in this way that it is possible to explain how Breton manages to comply with both simultaneously! However, this compliance gives rise to two manipulations (not to say “transformations”) with respect to affirmative phrases in “bare” presentation (PSO). In the case of the simple verb structure in “bare” presentation, the negative tense particle ne is sufficient to fill the first slot: VSO (11) ne=zebran ked a=grampouzh [PSO] ne I.eat not of pancakes I do not eat pancakes In the various “lead-in” presentations, the predicate is already in second position: (12) alîes e≠tebran krampouzh [XPSO; X = ADV] often e I.eat pancakes often I eat pancakes (13) krampouzh a=zebran [XPS; X = O] pancakes a I.eat I eat pancakes While the negative tense particle may fill the first slot, it may also, like the affirmative tense particles, come between a constituent in The Impersonal in Breton 9 initial position and a tense-bearing element in second position. In this sense, it is “ambivalent”: (14) alîes ne=zebran ked [XPS; X = ADV] often ne I eat not often I do not eat However, in the affirmative in “bare” presentation, some mechanism is required to get the tense out of initial position, and this is where the “dummy auxiliary” RA “do” comes in. Thus with simple verb structures in “bare” presentation, the affirmative equivalent of the negative (15) is (16): (15) ne=zebran ked a=grampouzh [PSO] ne I.eat not of pancakes I do not eat pancakes (16) dibriñ a ran krampouzh [PSO] eat.INF a I.do pancakes I eat pancakes Similarly, with auxiliary structures (existential, copula, compound tenses), the normal order is ...AUX PRED S... Since it is the auxiliary which carries tense, the trick in order to satisfy the T-2 constraint in “bare” presentation in the affirmative consists in inverting the auxiliary and predicate: 10 Steve Hewitt AUX PRED > PRED AUX inversion in “bare” presentation in the affirmative “Bare” presentation PS AFF: PRED AUX (S) NEG: “Lead-in” presentation XPS ne AUX ked PRED (S) CIRC AUX PRED (S) krampouzh so (L ez eus) pancakes be.EXIST.AFFº there are pancakes n’eus ked a=grampouzh ne be.EXIST.NEGº not of pancakes there are no pancakes neuse so (L ez eus) krampouzh so be.EXIST.AFFº pancakes so there are pancakes bras eo an ti big isº the house the house is big n’eo ked bras an ti ne isº not big the house the house is not big neuse e≠h-eo bras an ti so e isº big the house so the house is big aed eo Yann da=Gemper n’eo ked aed Yann da=Gemper ne isº not gone Yann to Quimper gone isº Yann to Quimper Yann has gone to Quimper Yann has not gone to Quimper neuse e≠h-eo aed Yann da=Gemper so e isº gone Yann to Quimper so Yann has gone to Quimper debred e meus krampouzh ne meus ked debred a=grampouzh neuse e meus debred krampouzh so e I.have eaten pancakes ne I.have not eaten of pancakes eaten e I.have pancakes so I have eaten pancakes I have not eaten any pancakes I have eaten pancakes The only (poorly understood) exceptions to the prohibition of tense in initial position are for the situative verb “be” and the verb “go”, particularly as used for the future of certainty or intention (cf. French je vais faire, English I am going to do; eastern dialects prefer a simple tense of “go”, western dialects the progressive). Thus in (17), (18), and (19), very exceptionally, a tensed verb begins an affirmative sentence: (17) emañ Yann o≠tibriñ krampouzh is.SITº Yann o eat.INF pancakes Yann is eating pancakes (18) e≠h-an da=zibriñ krampouzh [eastern dialects] e I.go to eat.INF pancakes I am going to eat pancakes (19) emon o≠vond da=zibriñ krampouzh [western dialects] I.am.SIT o go.INF to eat.INF pancakes I am going to eat pancakes In view of this exception, there is no need, in the case of the double (periphrastic) structure of the progressive, to get the grammatical The Impersonal in Breton 11 verb emañ “be.SIT” out of initial position in order to satisfy the T-2 constraint. But is it really the grammatical verb that is predicate? In our formula PSO, the symbol P stands for a syntagm which includes two quite different things, namely the lexical predicate (whether verb, noun, or adjective) and the tense-bearing element (whether full verb or auxiliary). The two are conflated in the case of a simple tensed verb ≠tebran / =zebran [I.eat], and adjacent in the auxiliary structures ...meus debred [I.have eaten] or debred e meus [eaten e I.have], but separate by definition in double (periphrastic) structures. So is it the grammatical verb or the lexical verb which best corresponds to the notion of predicate for Breton-speakers? The somewhat disappointing and evasive answer is that it is both at one and the same time. On the one hand, from a pragmatic point of view (information structure), it is clearly (20) which is the most neutral (information content which is typical of “bare” presentation): (20) emañ Yann o≠tibriñ krampouzh is.SITº Yann o eat.INF pancakes Yann is eating pancakes On the other hand, Breton-speakers are so used to the AUX PRED > PRED AUX inversion in the affirmative in “bare” presentation that the affirmative equivalent of the negative (21) is in practice not only (22), but also (23), whose initial VP is obligatorily focused (because of the focus, this type of sentence is very rare with an expressed nominal subject). (21) n’emañ ked o≠tibriñ krampouzh ne he/she.is.SIT not o eat.INF pancakes he/she is not eating pancakes (22) emañ o≠tibriñ krampouzh he/she.is.SITº o eat.INF pancakes he/she is eating pancakes (23) o≠tibriñ krampouzh emañ o eat.INF pancakes he/she.is.SIT he/she is eating pancakes, what he/she is doing is eating pancakes This practical equivalence is reinforced by the fact that situative forms of the verb “be” exist only for the present (and in the NW for 12 Steve Hewitt the imperfect) and that in the absence of a specifically situative form, speakers hesitate to begin a sentence with a tensed verb. What we have here is a classic instance of tension between form (non-situative, prohibited in initial position) and function (situative, allowed in initial position). This hesitation, which can readily be felt in the eastern dialects, where situative forms are lacking for the 1st and 2nd persons in the present, and for all persons in the imperfect, becomes even stronger in other tenses (future, conditional), where there are no situative forms anywhere. Such an initial focused dynamic VP, linked to situative “be” for the progressive in (23), may also be found in conjunction with “activity-do” for simple tenses (25). This is a further double (periphrastic) structure, and must not be taken to be a mere variant of the “bare” presentation simple structure in (24): (24) dibriñ a ran krampouzh [simple structure: infinitive + dummy auxiliary “do”] eat.INF a I.do pancakes I eat pancakes (25) dibriñ krampouzh a ran [double structure: dynamic VP + syntactic verb “activity-do”] eat.INF pancakes a I.do I eat pancakes, what I do is eat pancakes For one thing, the VP must be dynamic, as for the progressive. With a stative verb, only the simple structure is possible: (26) anveżoud a ran Yann [simple structure: infinitive + dummy auxiliary “do”] know.INF a I.do Yann I know Yann (27) *anveżoud Yann a ran [double structure: *stative VP + “activitydo”] know.INF Yann a I.do Furthermore, in the compound tenses, the double-verb (periphrastic) structure equivalent of the simple-verb auxiliary structure (28) is not (29), but (30): The Impersonal in Breton 13 (28) debred e meus krampouzh [simple-verb auxiliary structure: PP of lexical verb] eaten e I.have pancakes I have eaten pancakes (29) *debred krampouzh a meus eaten pancakes a I have (30) dibriñ krampouzh a meus gwraed [double-verb auxiliary structure, PP of “activity-do”] eat.INF pancakes a I.have done I have eaten pancakes, what I have done is eat pancakes 2. The apersonal conjugation: no subject-marking In traditional terminology, Breton is described as having a “personal” conjugation, whose endings express both tense and person and number (T.PN), and an “impersonal”, or more properly, apersonal conjugation, which indicates tense, but contains no reference to the person or number of the subject. The default case is the apersonal conjugation (no subject agreement). The personal conjugation is used in two specific instances: (1) where there is a putative sequence V S.PRON (the personal conjugation thus represents the incorporation of a post-verbal subject pronoun into the tense ending), and (2) with SV order in the negative (subject agreement). A possible explanation for the dissymmetry, with SV order, between the absence of subject agreement in the affirmative and the presence thereof in the negative is suggested in Hewitt 1985. 14 Steve Hewitt Personal and apersonal conjugations (personal forms in bold) Affirmative VS SV xvs Negative dibriñ a ra an=dud eat.INF a doº the people people eat ne=zebr ked an=dud ne eatº not the people people do not eat dibriñ a ran eat.INF a I.do I eat ne=zebran ked ne I.eat not I do not eat an=dud a=zebr the people a eatº people eat an=dud ne=zebront ked the people ne they.eat not people do not eat me a=zebr I a eatº I eat me ne=zebran ked I ne I.eat not I do not eat neuse e≠tebr an=dud so e eatº the people so people eat neuse ne=zebr ked an=dud so ne eatº not the people so people do not eat neuse e≠tebran so e I.eat so I eat neuse ne=zebran ked so ne I.eat not so I do not eat The verb/auxiliary “have” (cf. above, section 1.3) has only a personal conjugation, which is understandable in view of the fact that historically, its person/number markers do not refer to a subject pronoun, but rather to an oblique (dative) participant (cf. Lazard 1994: 142). For all other verbs, the apersonal conjugation always corresponds to the 3SG form of the personal conjugation (both the copula and situative verb “be” has for the present with S V order in the affirmative a special form so; the normal apersonal conjugation form for the present is eo). Thus without any particular context, a sentence like (31) is perfectly ambiguous: The Impersonal in Breton 15 (31) gweled a ra arºc’hizhier en=deñvalijenn (a) see.INF a doº the cats in.the dark cats see in the dark (b) see.INF a he/she.does the cats in.the dark he/she sees the cats in the dark Naturally, this ambiguity may be resolved by converting to S V order: (32) arºc’hizhier a=wel en=deñvalijenn the cats a seeº in.the dark cats see in the dark (33) hennezh/honnezh a=wel arºc’hizhier en=deñvalijenn that.one.M/that.one.F a seeº the cats in.the dark he/she sees the cats in the dark However, if arºc’hizhier in (32) is focused (initial subjects may be focused or thematic; fronted objects are always focused unless there is also a resumptive pronoun in the normal object position), the sentence may once again be ambiguous: (34) arºc’hizhier a=wel en=deñvalijenn (a) the cats a seeº in.the dark cats see in the dark, it’s cats that see in the dark (b) the cats a he/she.sees in.the dark it’s the cats he/she sees in the dark 3. The impersonal forms in -r and -d: indeterminate subject All the Celtic languages have “impersonal” verb forms in -r and -d. In English-language Celtic studies, the term “autonomous” is also used. These forms are said to be related to the “passive” conjugation of Latin (-itur), but there is only one form per tense (at least in the modern languages; Pierre-Yves Lambert (1998a: 304; 1998b: 843) gives 3SG and 3PL forms for Old Irish, Old Welsh, and Old Breton). Thus, they cannot perform the twin functions of the majhūl “unknown, passive” vowelling in Arabic: impersonal with an indeterminate human subject (36c), and personal with a passive subject < object (36a, b; 38): 16 Steve Hewitt (35) qatala (Arabic) (a) killedº S S killed (b) he.killed he killed (36) qutila (Arabic) (a) was.killedº S S was killed (b) he.was.killed he was killed (c) was.killedº “it was killed”, there was killing, people were killed (37) qatalū (Arabic) they.killed they killed (38) qutilū (Arabic) they.were.killed they were killed In Breton, (see the conjugation table for regular verbs in section 1.1), the forms in –ed are now moribund except in Leon (NW), which is precisely the region where the 2PL of the present is -it rather than -et. King provides a concise description of the force of these forms in Welsh: “Although sometimes listed as ‘passives’, these two forms are properly referred to as autonomous or impersonal, since they are not strictly speaking passive in sense (note that they can be formed for all verbs, including intransitives like come and go that have no passive). They convey the idea of the general action of the verb without specifying who or what is doing it. English has no exact equivalent of these, and must resort to paraphrases with one if a close translation is sought: (non-past) siaredir one speaks/will speak; (past) siaredwyd one spoke. But in practice the English passives is/will be ...-ed and was/has been ...-ed are usually the closest natural equivalent: Siaredir Cymraeg fan hyn Welsh (is) spoken here.” (King 1993: 220) The Impersonal in Breton 17 It should be added that in Welsh, these forms are emphatically part of the literary register, and apart from a handful of set phrases, are rarely heard in spontaneous conversation. In Breton, on the other hand, the impersonal forms for the present and future at least (the future impersonal form may also occasionally serve for the conditional) are very much alive and widely used. Rather than the passive, the closest equivalent is with French on (English one). Le Roux (1957: 273-4) gives several examples of impersonal forms in Middle Breton used in conjunction with explicit agents, but notes (p. 274) that “cet emploi déjà relativement rare en moyen-breton, et probablement surtout littéraire [probably an attempt to render French passives with an explicit agent], disparaît en breton moderne”. Welsh allows an impersonal form plus an agentive phrase: (39) Pregethir Dydd Sul gan y Parch. Elwyn Davies (Welsh) one.will.preach/there.will.be.preaching Sunday with the Rev. Elwyn Davies Sunday sermon by the Rev. Elwyn Davies Breton does not: (40) Prezeg a raffer dissul Preach a one.will.do Sunday One/somebody will preach on Sunday (41) *...gant an Tad Erwan Lagadeg ...with the Father Erwan Lagadeg ...by Father Erwan Lagadeg In order to express (39), Breton has to use the impersonal compound passive (see below, section 4.4): (42) Prezeged e≠vo dissul gant an Tad Erwan Lagadeg Preached e will.beº Sunday with the Father Erwan Lagadeg Sunday sermon by Father Erwan Lagadeg For Irish Gaelic, Russell gives: (43) bristear an fhuinneog (Irish, Russell 1995: 101) one.breaks/is.broken the window the window is broken/somebody breaks the window 18 Steve Hewitt (44) bristear í (Irish, Russell 1995: 101) one.breaks/is.broken her.O it is broken/somebody breaks it In (44), í is an object pronoun rather than a subject pronoun (sí); the construction would therefore seem to be active. Russell also gives (45), with an agentive phrase; apart from the fact that the order is wrong, this would appear to be rather doubtful, and not to occur in spontaneous speech.7 (45) bristear í liom (Irish, Russell 1995, p. 101; more likely bristear liom í) one.breaks/is.broken her.O with.me it is broken by me It is in Welsh that these forms appear to be least active and most passive.8 As we have already seen, the presence of an agentive phrase poses no problem for Welsh. A further indication of passiveness is the following: in Welsh, indefinite objects (without any article) generally undergo soft mutation following (even at a distance) a tensed verb form: (46) gwelaf =dŷ mawr acw (Welsh) I.see house big over.there I see a big house over there However, the fact that this does not happen following an impersonal form suggests that tŷ has the syntactic status of subject: (47) gwelir tŷ mawr acw (Welsh) one.sees/is.seen house big over.there one can see a big house over there, a big house can be seen over there All the same, it may be that the reluctance to render these Welsh impersonal forms with English one may have more to do with the relatively low frequency of that indefinite pronoun in English, and that functional equivalence between the Breton impersonal forms and French on may be boosted by the extremely high currency of on in French. Whether we like it or not, the Celtic languages find themselves today in a state of symbiosis with their respective The Impersonal in Breton 19 “metropolitan” languages, which now provide an inescapable model of linguistic reference. In any case, the impersonal forms are felt to be fully active in Breton, since a non-countable object takes the partitive a “of” in the negative, just as with personal forms; similarly, impersonal forms can take the same object pronouns derived historically from a “of” as the personal forms do: Personal forms Impersonal form dibriñ a ran krampouzh eat.INF a I.do pancakes I eat pancakes dibriñ a rer krampouzh eat.INF a one.does pancakes one eats pancakes ne=zebran ked a=grampouzh ne I.eat not of pancakes I do not eat pancakes ne=zebrer ked a=grampouzh ne one.eats not of pancakes one does not eat pancakes gweled a ran ahanout see.INF a I.do you.O [of.you] I see you gweled a rer ahanout see.INF a one.does you.O [of.you] one sees you ne=welan ked ahanout ne I.see not you.O [of.you] I do not see you ne=weler ked ahanout ne one.sees not you.O [of.you] one does not see you However, the Breton impersonal form differs from the French on in two ways. On the one hand, it does not usually replace the 1PL as in colloquial French: (48) chez nous on mange beaucoup de crêpes at us one eats much of pancakes at our place we eat a lot of pancakes In Breton, the 1PL is usual: (49) =du-mañ e≠tebromp ur bern krampouzh side-this e we.eat a heap pancakes at our place we eat a lot of pancakes The impersonal form suggests difficulty in identifying or a reluctance to specify who eats the pancakes: 20 Steve Hewitt (50) =du-mañ e≠tebrer ur bern krampouzh side-this e one.eats a heap pancakes at our place one eats a lot of pancakes At the most, the Breton impersonal form may be used for suggestions commonly expressed with on in French (on y va?): (51) mond a raffer? go.INF a one.will.do? shall we go? But here, the 1PL is equally current: (52) mond a raffomp? go.INF a we.will.do? shall we go? Secondly, and far more importantly, there is no pronoun corresponding to the impersonal form. So while a VSO order such as (53) may be converted into SVO (54), there is no possible SVO for (55): (53) dibriñ a ran krampouzh eat.INF a I.do pancakes I eat pancakes (54) me a=zebr krampouzh I a eatº pancakes I eat pancakes (55) dibriñ a rer krampouzh eat.INF a one.does pancakes One eats pancakes There is a kind of indefinite pronoun an nen, derived from an den “the person”, which may be used with the apersonal conjugation / 3SG: (56) dibriñ a ra an nen krampouzh pa neveż nawn eat.INF a doº “the.man” pancakes when he.has.HAB hunger one eats pancakes when one is hungry However, it is more customary to use the impersonal form: The Impersonal in Breton 21 (57) dibriñ a rer krampouzh pa=veż nawn eat.INF a one.does pancakes when be(.EXIST).HAB hunger one eats pancakes when one is hungry Note that there can be no impersonal form for the verb “have”, given that all its forms consist of existential forms (where available) of the verb “be” plus proclitic oblique pronouns; since the impersonal form corresponds to no pronoun which might provide such an affix, Breton is forced to fall back on existential “be” (there is) in order to render one has. 4. Impersonal constructions: absence of subject Breton is a pro-drop language; unlike English, French or German, it has no need of “dummy subjects” along the lines of it, there; il; es. Let us recall (section 2) that the apersonal conjugation coincides with the 3SG of the personal conjugation. Therefore, the only way of telling whether a form which looks like 3SG actually has a referent is by converting a PS string to SP, as in (53) and (54) above. For none of the following Breton examples is it possible to find a conversion with an initial subject pronoun. One is therefore led to conclude that we have an apersonal conjugation, and that there is no formal subject. 22 Steve Hewitt Impersonal constructions il est difficile de dire it is difficult to say dîaes eo da lâred difficult beº to say il faut y aller it is necessary to go red eo mond necessary beº go.INF il y a des pancakes there are pancakes krampouzh so pancakes be.EXIST.AFFº il fait chaud it is hot tomm eo hot beº il me semble it seems to me kaoud a ra din find.INF a doº to.me es wird getanzt il est dansé, there is dancing, people are dancing dañssed e≠veż danced e be.HABº 4.1 The existential In section 1.5, we described the existential as being an auxiliary structure in which the existential operator functions as an auxiliary, and the indefinite entity whose existence is predicated functions as predicate. In order to satisfy the T-2 constraint, this auxiliary structure undergoes AUX PRED > PRED AUX inversion in “bare” presentation in the affirmative: (58) n’eus ked a=grampouzh ne be.EXIST.NEGº not of pancakes there are no pancakes (59) krampouzh so (L ez eus) pancakes be.EXIST.AFFº there are pancakes Recall (section 2) that with SV order in the negative (but not the affirmative), there is subject-verb agreement (personal conjugation). Therefore, if the existential entity were subject, we would expect it in initial position to trigger agreement in the tense-bearing element, but that does not happen – the form eus in (60) is not 3PL; a further The Impersonal in Breton 23 indication that the existential entity is not subject is the use of the negative partitive (58). (60) krampouzh n’eus ked! pancakes ne be.EXIST.NEGº not pancakes, there are none! This is not to say that an indefinite entity, once its existence has been predicated by the existential construction, cannot then become the subject of a VP complement: (61) tud so o≠tibriñ krampouzh people be.EXIST.AFFº o eat.INF pancakes people are eating pancakes / there are people eating pancakes The form so in (61) is analysed as being existential rather than the special subject-initial present apersonal conjugation of “be” on the grounds of what happens in the negative: (62) n’eus ked a=dud o≠tibriñ krampouzh ne be.EXIST.NEGº not of people o eat.INF pancakes people are not eating pancakes / there are no people eating pancakes Apart from the verb “be”, there are a number of “presentative” verbs which serve to introduce an indefinite entity, which may in turn become the subject of a VP complemement: (63) dond a ra touristed da=weled an ilis come.INF a doº tourists to see the church tourists come to see the church Again, the initially non-subject nature of the indefinite entity is suggested by the negative partitive: (64) ne deu ked a=douristed da=weled an ilis ne comeº not of tourists to see the church tourists do not come to see the church / no tourists come to see the church 4.2 Meteorological phenomena Meteorological phenomena pose a special problem with regard to actancy: while in pro-drop languages such as Breton it is not possible 24 Steve Hewitt to find a subject pronoun with which to present meteorological sentences in SPO order, one can nevertheless imagine the existence of a specific referent as subject: (65) braw eo (an amser) fine beº (the weather) it is fine / the weather is fine (66) c’hwezhañ a ra (an awel) blow.INF a doº (the wind) it’s blowing / the wind is blowing In (67), glaw is not subject, but rather object, as demonstrated by the negative partitive in (68): (67) glaw a ra rain a doº it’s raining (68) ne ra ked a=c’hlaw ne doº not of rain it’s not raining Many meteorological phenomena are rendered by nouns plus “do” (69), but there are also a number of meteorological denominative verbs, such as rewiñ “to frost (rew)” or skornañ “to freeze” (skorn “ice”). (69) erc’h a ra snow a doº it’s snowing 4.3 Indirect impersonal verbs Breton has a fair number of indirect impersonal verbal expressions using the prepositions da “to” or gant “with”: (70) kaoud a ra din find.INF a doº to.me I find, I think, it seems to me (71) ne=gav ked din ne findº not to.me I don’t find, I don’t think, it doesn’t seem to me The Impersonal in Breton 25 The pronoun representing the participant in question may be fronted as a topic in “lead-in” presentation (72), but that does not mean that it is the subject, as shown by the lack of agreement (73): (72) me a=gav din I a findº to.me I find, I think, it seems to me (73) me ne=gav ked din I ne findº not to.me I don’t find, I don’t think, it doesn’t seem to me A typical example of an indirect impersonal construction from Jules Gros: (74) Me a=vez =welloc’h ganin insultiñ an=dud ewid mond d’ur pred eureud: muioc’h a=blijadur am-bez! I a be.HABº better with.me insult the people than go to’a meal wedding: more of pleasure I.have.HAB! I prefer insulting people to going to a wedding feast: it gives me greater pleasure! (Gros 3: 291) Typical indirect impersonal verbs (always involuntary phenomena, no control by patient) kaoud a ra din findº to.me soñjal a ra din ffelloud a ra din digoueżoud a ra din tomm eo din red eo din mad eo din gwelloc’h eo din/ganin sevel a ra din/ganin tapoud a ra ganin thinkº to.me wantº to.me happenº to.me hot isº to.me necessary isº to.me good isº to.me better isº to.me /with.me riseº to.me / with.me pick.upº/grab.holdº with.me I find, I think, it seems to me I think, it seems to me I want I happen to..., it happens that I I am hot I must, I have to I am willing, I am happy to I’d rather, I prefer I get a hard-on I’m in luck 26 Steve Hewitt 4.4 The impersonal compound passive The impersonal compound passive is an auxiliary structure consisting of the auxiliary “be” (always habitual in the past and the imperfect) and the past participle: (75) dañssed e≠veż danced e be.HABº “it is danced”, there is dancing, people dance (76) ne=veż ked dañssed ne be.HABº not danced “it is not danced”, there isn’t dancing, people don’t dance The following example was heard from someone who was horrified at a spate of murders in Paris: (77) dre amañ ne=veż ked laz’hed re c’hwazh by here ne be.HABº not killed too.much yet around here there’s not too much killing yet, they don’t kill too much yet hier herum wird noch nicht zu viel getötet (77) is a real passive, unlike (78), an active using the impersonal form: (78) dre amañ ne laz’her ked re c’hwazh by here ne one.kills not too.much yet around here one does not kill too much yet The impersonal compound passive may be applied to fairly complex double (periphrastic) structures (see sections 1.5 and 1.6) whose dynamic VP includes an embedded clause: (79) diskar traeoù so mad en o sav c’hwazh a=veż gwraed pull.down things beº [post-subject AFFº form] good in their standing yet a be.HABº done things that are still standing sound are pulled down, “they” pull down things that are still standing sound The impersonal compound passive may take an indefinite object; note the negative partitive in (81): The Impersonal in Breton 27 (80) debred e≠veż krampouzh eaten e be.HABº pancakes on mange des crêpes pancakes are eaten (81) ne=veż ked debred a=grampouzh ne be.HABº not eaten of pancakes on ne mange pas de crêpes pancakes are not eaten, no pancakes are eaten One must not confuse the impersonal compound passive with the personal compound passive: a definite patient acts like a subject: (82) debred e≠veż arºc’hrampouzh eaten e be.HABº the pancakes on mange les crêpes the pancakes are eaten (83) debred e≠vent eaten a they.be.HAB on les mange they are eaten If the regular forms of “be” are used instead of the habitual forms, the effect is a perfect resultative (84); compare the difference in Dutch between (85) and (86): (84) debred eo arºc’hrampouzh eaten isº the pancakes on a mangé les crêpes, les crêpes sont mangées the pancakes have been eaten (85) de pannekoeken worden gegeten (Dutch) the pancakes become eaten the pancakes are eaten (86) de pannekoeken zijn gegeten (Dutch) the pancakes are eaten the pancakes have been eaten Nikolaz Davalan, a doctoral student in Breton, has drawn attention to an interesting example of interference from French among neo-Breton (French-mother-tongue) pupils in the all-Breton Diwan 28 Steve Hewitt schools and French-Breton bilingual schools in Brittany.9 The impersonal compound passive in (87) is so strongly identified with the active French construction using on that these children regularly add object pronouns (88), whereas the normal Breton would be the personal compound passive in (83): (87) debred e≠veż eaten a be.HABº there is eating, people eat, “they” eat on mange (88) *debred e≠veż anê eaten a be.HABº them.O they are eaten on les mange Finally, given that there is a personal compound passive (82), (83), it is only logical to be able to use the impersonal form in that construction, as in the following example describing the charms of a tropical country: (89) ma ne=ver ked debred gant ar ffubu, e≠ver laz’hed gant an=dommder if ne one.is not eaten with the mosquitoes, e one.is killed with the heat if you’re not eaten alive by the mosquitoes, you’re killed by the heat 5. Conclusion We began by making a rapid presentation of the verbal system of Breton, which achieves the considerable feat of remaining faithful to its Celtic VSO heritage (which we reanalyse as [P = predicate syntagm] PSO), while at the same time obeying a northern European V-2 constraint (relayed through Old French, but largely residual in Modern French), which ought more properly to be renamed T-2, where T represents the tense-bearing element (verb or auxiliary). We distinguish three basic predicate structures: the simple verb structure, the auxiliary structure, and the double or periphrastic structure combining a dynamic non-tensed lexical VP and a tensed syntactic verb, either “situative-to be” ou “activity-do”. The auxiliary structure comprises the compound tenses (AUX + PP), the copula (COP + The Impersonal in Breton 29 adjectival or nominal predicate), and the existential (existential operator + indefinite nominal predicate) because these constructions all have an identical syntax; I suspect that such a grouping might well be extended to many other languages, and that one should not automatically suppose that the copula and the existential operator have the same syntactic status as the lexical verb. We also distinguish between a “bare” PSO presentation and a “lead-in” XP... presentation, in which X may be any major constitutent except the tensed predicate syntagm, X being either thematic or focused. In the “bare” presentation, the T-2 constraint has the effect of transforming the negative utterances ne=zebran ked [ne I.eat not] “I do not eat” and n’eo ked bras an ti [ne beº not big the house] “the house is not big” into the corresponding affirmative utterances dibriñ a ran [eat.INF a I.do] “I eat” (“dummy” auxiliary creation) and bras eo an ti [big beº the house] “the house is big” (AUX PRED > PRED AUX inversion). We then distinguished three different acceptations of the impersonal, first the impersonal, or rather apersonal conjugation (no subject-marking), then, within the personal conjugation (which shows subject-marking), the impersonal form (indeterminate, unspecified subject), and finally various impersonal constructions (no subject). The apersonal conjugation involves a marking of tense, but not of person or number; there is thus a single form per tense, which is identical with the 3SG (the verb/auxiliary “have” has no apersonal conjugation, and the verb/auxiliary “be” has in the non-habitual present a special form so which is used after a preceding subject in the affirmative). This apersonal conjugation is used after any subject (nominal or pronominal) in the affirmative, and before all expressed nominal subjects. With a preverbal subject, it is the apersonal conjugation which is used in the affirmative: me a=zebr [I a eatº] “I eat”, but the personal conjugation in the negative: me ne=zebran ked [I ne I.eat not] “I do not eat”. The personal conjugation appears to represent the inclusion in the tense-marking of a postverbal pronominal subject: dibriñ a ra an=dud [eat.INF a doº the people] “the people eat”, dibriñ a reont [eat.INF a they.do] “they eat”. Given that the single form of the apersonal conjugation is identical with the 3SG form, this may give rise to ambiguities wherever a nominal entity could equally well be subject or object of the verb in question. The impersonal forms in -er and -ed, typical of the Celtic languages, constitute a seventh form in the personal conjugation 30 Steve Hewitt paradigm. They refer to a putative human subject whose identity one either does not wish to or is unable to specify. The meaning is thus very close to that of the French on “one”, the main difference being that there is no pronoun which corresponds to the impersonal or autonomous form, for which an SV presentation is thus excluded. While this construction is fully active in Breton, the cognate forms in Irish, and especially in Welsh, may also convey a passive sense for transitive verbs, as witnessed by the possibility of using them in conjunction with agentive phrases of the type “by X”, which may not be used with the impersonal form in Breton. Impersonal constructions, which we see as having no subject, include the existential and related presentative verbs, meteorological phenomena, indirect impersonal verbs of the type “it pleases me to...”, and the compound impersonal passive, such as “!it is danced”, “il est dansé”, “es wird getanzt”. Given that Breton does not require a subject to be expressed (it has no “dummy” subjects on the lines of English it, there, French il, German es), and that the apersonal conjugation (absence of subject-marking) coincides with the 3SG of the personal conjugation, it is not always clear a priori how to interpret the verb forms used in these impersonal constructions. What is beyond doubt, however, is that a non-subject-marking analysis (the apersonal conjugation) is otherwise unavoidable for Breton. This being the case, I see no reason compelling reason to assume that the verb forms used in these impersonal constructions actually involve a 3SG. Furthermore, I think that Breton may well point to a similar analysis for impersonal constructions in other languages which do not require an explicit subject, even if a specifically non-actancial marking is not independently required for those languages. Notes 1. Sincere thanks go to Nancy Stenson for providing a copy of her article (1989) and for extensive discussion by e-mail of the impersonal form in Irish, and to Elizabeth Pyatt for providing a copy of her unpublished paper (1995). 2. This section reflects a descriptive framework initially proposed in Hewitt 1988. The Impersonal in Breton 31 3. The Breton examples reflect a normalized form of the Treger dialect (NE of the Breton-speaking area). I use my own “etymological” orthography, a further elaboration of the etrerannyezhel “interdialectal” orthography created in 1974 and used in Favereau 1992 and 1997, but more effective than the latter in predicting dialect reflexes (see Hewitt 1987). The most common orthographies are first (at least 85 %) the peurunvan “completely unified” orthography, also known as the ZH orthography, launched in 1941 under Nazi auspices and often associated with a nationalist political stance; and a distant second, the skolveurieg “university” orthography, created in 1955 and promoted not very successfully by the French authorities in an attempt to divide and counter overtly nationalist aspirations. As a result of French educational policy, functional literacy in Breton among native speakers (the ability to write a simple personal letter) is well under 1 %. The presence of initial consonant mutations is indicated by = for the “soft” mutation (lenition), ≠ for the “mixed” mutation (lenition / provection), and º for the “aspirate” mutation (spirantization + lenition); d does not normally lenite to z in Treger either in tensed verbs or in lenited adjectives (contrary to popular belief among language activists, it does so more regularly in other contexts, e.g. da=zorn “your hand”) – the d > z mutation is shown here in order to bring the examples into line with majority usage. The tense particles a= (direct affirmative, used after subjects, direct objects, and infinitives), e≠ (indirect affirmative, used after other constituents, such as prepositional phrases, adverbials, and past participles), and ne= (negative, more usually na= in Treger; before vowels, n’eo ked would be better written for Treger as na n-eo ked), as well as the progressive infinitival particle o≠ (é≠), are not usually pronounced at normal tempos; however, the mutations they trigger remain, regardless of whether the particle is heard or not. In the large group of “central” dialects (including Treger) lying along a NE-SW axis, the “indirect affirmative particle e≠ is moribund, and is generally replaced (on the evidence of mutations) by the “direct” particle a=. This is not reflected in this article in order to conform to standard literary usage, which is based on the highly divergent “peripheral” dialects of Leon (NW) and Gwened (SE). In central Treger, as in a number of other areas, the progressive particle, when heard, is é≠ rather than o≠; again, this is not reflected here. 4. For more details, see Hewitt 1986. 32 Steve Hewitt 5. More precisely V-initial / SVO. With verb-initial clauses in Arabic, there is a low, but significant proportion of VOS and other orders which violate VSO. The principle that accounts for the order of constituents in all Arabic verb-initial clauses, including VSO, appears to be pragmatic rather than syntactic: VGN (verb-given-new); see Hewitt (forthcoming). 6. See Hewitt 1985: 228-33. 7. The Irish situation seems particularly difficult to nail down precisely. The impersonal/autonomous form is found with agentives (“by someone”) in modern officialese, no doubt prompted by the passive English translation of the form: Cuireadh an tuarascáil parlaiminte le chéile ag Astrid Thors MEP, ball de phobal na Sualainnise san Fhionlainn [was.put the report parliament.GEN with companion [=together] at [=by] Astrid Thors MEP, member of people the Swedish.language.GEN in.the Finland] “The parliamentary report was put together by Astrid Thors MEP, member of the Swedishspeaking people in Finland” (Nolan 2001: 61). Nancy Stenson (1989 and personal communication) indicates that impersonal/autonomous forms do not normally occur with agentives in native-like Modern Irish, although agentives were possible in earlier periods (e.g. Keating), and there is sporadic evidence of agentives in certain modern dialects (Donegal: James McCloskey, personal communication to Stenson). John P. Henry (1906: 19-24) gives a number of examples of impersonals with human agent phrases, but notes (p. 19) that the “construction is quite obsolete in [Munster and South Connaught], but we have abundant examples of it in the old literature, and it is still a living form in the counties of Mayo, Sligo, Roscommon, Leitrim and the whole of Ulster, but only in the mouths of the old speakers. It is becoming more and more rare among the young generation”. William Gillies (1993: 187) gives for Scottish Gaelic rinneadh an t-òran le Iain [was.done the song with John] “the song was made by John”. There thus seems to be a split between Southern Irish (Munster and Conamara), which does not now allow agentives, and Northern Irish (Mayo and Donegal) and Scottish Gaelic, which do allow them (or did so until quite recently). Elizabeth Pyatt (1995) adduces syntactic evidence suggesting that the autonomous forms were truly passive in Old Irish, that by 11th-century Early Middle Irish, “the autonomous verbal form still has many passive characteristics, but there is a noticeable increase in the impersonal The Impersonal in Breton 33 usage.” (pp. 20-21), and that in Modern Irish, the construction is fully impersonal and active. Mícheál Ó Siadhail (1980: 163) has the following passage (impersonal forms shown in bold) about an accident involving a car following a lorry with a loose load: “Thit an bairille anuas i mullach an chairr. D’imigh sí ó smacht. Crochadh den bhóthar í agus buaileadh faoin gclaí i. Caitheadh Tomás Mór amach i mullach a chinn. … Tháinig tumálaí an leoraí slán as. Níor gortaíodh ar chor ar bith é.” (The barrel fell down on top of the car. It [the car] got out of control. It was lifted off the road and was flung against the stone wall. Tomás Mór was thrown out head first. … The lorry driver came out of it unscathed. He wasn’t hurt at all.) None of these impersonal forms imply human agency, and for that reason would be impossible in Breton; furthermore, the Breton preterite tense is moribund, and the ostensible preterite impersonal form -jod is completely unknown, except to literati. It is interesting to note that with two exceptions, all of Henry’s examples, both historical and contemporary, of agentives involve the past tense impersonal –adh. There may be a connection between this and the fact that the examples in Ó Siadhail (1980) of impersonal forms in which human agency is not implied are also in the past tense: if human agency is not necessarily implied, even today, by the past tense form –adh, there may have been less reason, historically, not to allow agentive phrases to be used with that form. 8. For a stimulating discussion of the syntactic status of the impersonal forms in Welsh, see Fife (1992). Elizabeth Pyatt (1995: 26) agrees with James Fife that the construction is basically impersonal: “Based on the lack of agent phrases, the presence of intransitive autonomous verbs and the position of patient/theme NPs in auxiliary constructions, one can straightforwardly conclude that Middle Welsh autonomous verbs are active voice with proarb [indeterminate human] subjects and that patient/theme NPs are always surface direct objects. … the situation did not remain so clear-cut in Literary Modern Welsh. In that language, it appears that the autonomous verb developed a secondary passive usage.” Alexander Falileyev (personal communication, and 2002) is of the view that, contrary to what appears to be the case in Old Irish, the Old Welsh forms were impersonal active rather than truly passive. It would be desirable to trace the history of the usage of these cognate impersonal/autonomous forms in Irish and Scottish Gaelic, 34 Steve Hewitt Welsh and Breton. Elizabeth Pyatt made a good start with her 1995 draft, which was unfortunately never completed or published. The key questions to be asked for each language and period, and for each tense of the impersonal/autonomous form, are whether the patient has subject or object characteristics (case, agreement, position), whether or not human agency is implied by the form, and whether or not agentive (as distinct from instrumental) NPs are possible with it. Whatever the case at the various stages of Irish or Welsh, it is clear that the Modern (and probably Middle) Breton forms are fully active and impersonal, imply an indeterminate human subject, and do not allow human agent phrases. 9. Personal communication and Davalan 1997: 115. Abbreviations and symbols º (doº) * ! ? = ≠ º a= ADJ ADV AFF AUX CIRC COMP COP DVP DYN e≠ EXIST apersonal conjugation: marking of tense, but not person/number incorrect, prohibited utterance unusual, surprising utterance, but not completely prohibited questionable utterance, of doubtful reliability soft mutation (lenition) mixed mutation (lenition/provection) spirant mutation (spirantization+lenition) direct affirmative tense particle (after subject, object, infinitive) adjective adverb affirmative auxiliary circumstant complement copula dynamic verb phrase dynamic indirect affirmative tense particle (after prepositional phrase; adverbial; past participle; normally replaced by a= in Treger and other central dialects) existential The Impersonal in Breton 35 F GEN HAB IMPERF INF L M MBr N NE ne= NEG NW O o≠ OP P PL PN PP PRED PRES PRET PRON RA S SE SG SIT SW T V X feminine genitive habitual imperfect infinitive Léon (NW), literary masculine Middle Breton noun north-east negative tense particle (na= in Treger) negative north-west object progressive infinitival particle (é≠ in central Treger and other areas) oblique participant, object of preposition predicate syntagm: V.T / AUX.T PRED (PRED AUX.T) plural person and number past participle predicate (verb, adjective, noun) present preterite pronoun “auxiliary”-do subject south-east singular situative south-west tense: V.T(.PN), AUX.T(.PN) verb any non-predicate initial element 36 Steve Hewitt References and general bibliography Awbery, Gwenllian M. (1976) The Syntax of Welsh: A Transformational Study of the Passive, Cambridge University Press. Blevins, James P. (draft, 2001) Passives and impersonals, forthcoming in the Journal of Linguistics. http://www.rceal.cam.ac.uk/staff/papers/pas.pdf Ball, Martin J. & Fife, James (eds) (1993) The Celtic Languages, Routledge Language Family Descriptions, London & New York: Routledge. Davalan, Nikolaz (1997) Interférences linguistiques chez des enfants scolarisés en breton, pp. 97-118 in Francis Favereau (ed.) Le Bilinguisme précoce en Bretagne, en pays celtiques et en Europe atlantique. Actes du Colloque international de Plésidy (Côtesd’Armor), octobre 1997. Klask 5, Presses Universitaires de Rennes, 1999. Denez, Per (1983) An dibersonel [The Impersonal], Hor Yezh [Our Language] (Lesneven, 29) 151, 5-29. Evans, D. Simon (1989) A Grammar of Middle Welsh, Dublin Institute for Advanced Studies. Falc’hun, François (1981) Perspectives nouvelles sur l’histoire de la langue bretonne, Paris: Union générale d’éditions. Revised and expanded version of his doctoral thesis (Rennes, 1951) originally published under the title Histoire de la langue bretonne d’après la géographie linguistique, Paris: Presses Universitaires de France, 1963. Falileyev, Alexander (2002) Drevnevallijskij jazyk [Old Welsh], St Petersburg: Nauka. Favereau, Francis (1992) Dictionnaire du breton contemporain, Morlaix (29): Skol Vreizh. –– (1997) Grammaire du breton contemporain, Morlaix (29): Skol Vreizh. Feuillet, Jack (1998) Actance et valence dans les langues d’Europe, “Empirical Approaches to Language Typology, Eurotyp 20-2”, Berlin & New York: Mouton de Gruyter. Fife, James (1985) The impersonal verbs in Welsh, Bulletin of the Board of Celtic Studies 23, 92-126. The Impersonal in Breton 37 –– (1992) Autonomy and the Welsh impersonal verb, Journal of Celtic Linguistics 1, 61-100. Fleuriot, Léon (1964) Le vieux-breton. Eléments d’une grammaire, Paris: Klincksieck. Gillies, William (1993) Scottish Gaelic, in Ball & Fife (1993: 145-227). Gros, Jules, Le trésor du breton parlé (Eléments de stylistique trégorroise): 1. Le langage figuré, 2nd revised and enlarged edition, St.Brieuc (22): Les Presses bretonnes, 1970. 2. Dictionnaire breton-français des expressions figurées, St.Brieuc (22): Les Presses bretonnes, 1970. 3. Le style populaire, Lannion (22): Barr Heol, 1976. Hemon, Roparz (1975) A Historical Morphology and Syntax of Breton, Dublin Institute for Advanced Studies. Henry, John P. (1906) A Hand-book of Modern Irish: Part IV, Dublin: The Gaelic League/M.H.Gill. Hewitt, Steve (1985) Quelques ressemblances structurales entre le breton et l’arabe. Conséquence d’une typologie ordinale commune?, La Bretagne Linguistique 1, 223-62. –– (1986) Le progressif en breton à la lumière du progressif anglais, La Bretagne Linguistique 2, 132-48. –– (1987) Réflexions et propositions sur l’orthographe du breton, La Bretagne Linguistique 3, 41-54. –– (1988) Un cadre pour la description de la syntaxe verbale du breton, La Bretagne Linguistique 4, 203-11. –– (forthcoming) L’arabe : VSO ou VDN (verbe-sujet-objet ou verbedonné-nouveau)?, in Anaïd Donabédian & Dan Xu (eds) Cahiers de Linguistique de l’INALCO 3/2000, L’Ordre des mots. Jackson, Kenneth H. (1967) A Historical Phonology of Breton, Dublin Institute for Advanced Studies. Jones, M. & Thomas, A.R. (1977) The Welsh Language. Studies in its Syntax and Semantics, Cardiff: University of Wales Press. King, Gareth (1993) Modern Welsh. A Comprehensive Grammar, Routledge Grammars, London & New York: Routledge. Lambert, Pierre-Yves (1998a) L’impersonnel, in Feuillet (1998: 295-345). –– (1998b) L’actance dans les langues celtiques, in Feuillet (1998: 811-47). 38 Steve Hewitt Lazard, Gilbert (1994) L’actance, Paris: Presses Universitaires de France. La Bretagne Linguistique: Travaux du Groupe de recherche sur l’économie linguistique de la Bretagne (GRELB), Centre de recherche bretonne et celtique (CRBC), Université de Bretagne Occidentale, B.P. 814, 29285 BREST Cédex. Le Roux, Pierre (1924-1953) Atlas Linguistique de la BasseBretagne, 6 fascicles containing 100 maps each, Rennes-Paris (reprinted, Éditions Armoricaines, Brest, 1977; currently available from Brud Nevez, 6 rue Beaumarchais, 29200 Brest). Le Roux, Pierre (1957) Le Verbe breton (Morphologie, syntaxe), Rennes: Librairie Plihon / Paris: Librairie Champion. Lewis, Henry & Piette, J.R.F (1966) Llawlyfr Llydaweg Canol [Handbook of Middle Breton], Caerdydd [Cardiff]: Gwasg Prigysgol Cymru [University of Wales Press] (revised and corrected edition). MacAulay, Donald (ed.) (1992) The Celtic Languages, Cambridge Language Surveys, Cambridge University Press. Müller, Nicole (2000) Agents in Early Welsh and Early Irish, Oxford University Press. Nolan, Brian (2001) Passive voice constructions in Modern Irish, ITB Journal 3, 51-78, Institute of Technology, Blanchardstown, Dublin. http://www.itb.ie/research/journal.html Noonan, Michael (1994) A tale of two passives in Irish, pp. 275-311 in Paul Hopper & Barbara Fox (eds), Voice: Form and Function, Amsterdam/Philadelphia, John Benjamins. http://www.uwm.edu/~noonan/2Irish.passives.pdf –– (2001) Subjectless clauses in Irish, Proceedings of the International Symposium on Non-nominative Subjects, Tokyo, 1721 December 2001. http://www.uwm.edu/~noonan/Irish.newpaper.pdf Ó Siadhail, Mícheál (1980) Learning Irish, Dublin Institute for Advanced Studies. –– (1989) Modern Irish. Grammatical Structure and Dialectal Variation, Cambridge University Press. Phillipaki-Warburton, Irene (1985) Word order in Modern Greek, Transactions of the Philological Society, 113-43. The Impersonal in Breton 39 Pyatt, Elizabeth (1995) The historical usage of the autonomous (impersonal passive) form in Celtic languages, unpublished manuscript, Harvard University. Russell, Paul (1995) An Introduction to the Celtic Languages, Longman Linguistics Library, London & New York: Longman. Stenson, Nancy (1989) Irish autonomous impersonals, Natural Language and Linguistic Theory 7, 373-406. Stephens, Janig (1993) Breton, in Ball & Fife (1993: 349-409). Ternes, Elmar (1992) The Breton language, in MacAulay (1992: 371-452). Timm, Lenora (1989) Word order in 20th-century Breton, Natural Language and Linguistic Theory 3, 361-78. Trépos, Pierre, [1968] Grammaire bretonne, Rennes: Simon (reprinted Ouest France, Rennes, 1980; new edition Brud Nevez, Brest, 1994). Varin, Amy (1979) VSO and SVO structure in Breton, Archivum Linguisticum 10, 83-101. Willis, David (1998) Syntactic Change in Welsh: A Study of the Loss of Verb-Second, Oxford University Press.