AMERINDIA n°25, 2000

Synopsis of a Boruca terminal speaker[1]


University of Toronto

1. Introduction

The phenomenon of language death, as many other aspects of language in this century, has been sufficiently studied and there is a better understanding of its causes and structural consequences, among other things (see Dorian 1989, Sasse 1992a, b). It is widely acknowledged that, aside from certain specific cases, language death is a process that does not befall suddenly but takes some time (see 1. below), during which a process of atrophy occurs, whereby the speakers of the dying language seem to “mess up” structure, meaning, and even sound when trying to speak it. This paper provides an example of a terminal speaker of Boruca, a Chibchan language of Costa Rica, doomed to die at the end of the XXth century (see 2. below), having only six elderly speakers left. The terms terminal speaker and semi-speaker are sometimes used interchangeably in the literature. Following Sasse (1992a), in this study the former term will be used simply to refer to the last speakers of the dying language, and who acquired it as their mother tongue in their early infancy. The latter term will refer to those individuals who show imperfect command of the language because it was not their first and more/most important language in early infancy. Being the death of Boruca of the so-called radical type (see 1.), most of its terminal speakers still show lucid command of the language, in which case the atrophy hypothesis would be “non-applicable”; however, the speaker under scope here does show lacunae that are important analyzing insofar as his performance can shed light on how, depending on the type of language death involved, competence is lost. This paper is organized as follows. Section 2 briefly presents a reference frame for understanding the process of language death, as originally discussed in Quesada (2000); Section 3 provides a glance at the Borucas and introduces the speaker whose performance will be analyzed in Section 4. Finally, Section 5 summarizes the findings of the study.

2. Language contact and language death

Language contact can be defined in psychological terms, “two languages are in contact if they are used alternately by the same person” (Weinreich 1953/1970: 1) or in sociological terms, “Zwei oder mehr Sprachen stehen in Kontakt miteinander, wenn sie in derselben Gruppe gebraucht werden. Dazu ist es nicht notwendig, daß jedes einzelne sprechende Individuum, das zu dieser Gruppe gehört, alle diese Sprachen spricht oder versteht“ (Bechert & Wildgen 1991: 1). Since languages are a correlate of mankind’s history, language contact situations imply historical events. When two or more languages come into contact, they can either keep existing side by side or one (or more) can decease, again, echoing historical events. Language death suggests that kind of contact which is not precisely the nicest among two peoples; it is usually the outcome of confronta­tion between two languages, a local one, which I will call the host language and a foreign one, which I will call the guest language[2]. These rather cynical terms have the virtue of being unmistakable enough as to avoid confusion. The gamut of consequences arising from language contact situations is summarized in (S1).























Modality:         CONVERGENCE




-Code switching  

-Language  Inter






-Linguistic change        


        (adstratum)  (super-




-Foreigner Talk

-Pidgins & Creoles

-Language  Shift



-Language death


  Power relations

                                                                                           (Taken from Quesada 2000).


Following (S1), language contact can be viewed in terms of two basic parameters, intensity and prevailing power relations. The former is chronological in nature -the idea here is that the longer the contact the more intense it is (see Thomason & Kaufman 1988, Ch. 4, on this)-, whereas the latter determines the modality which characterizes the relation­ship between the contacting peoples; these (and their languag­es) may either converge or conflict. The consequences of language contact can be classified considering the modality of the contact. While the consequences on the left can appear under both even and uneven power relations, those on the right are not likely to occur under even power relations; thus linguistic change, as a consequence of high-intensity contact can occur regardless of the evenness of the relations of power (cf. the so called superstratum, adstratum and substratum motivated changes) but language shift cannot; similarly, pidgins and creoles rarely if ever emerge out of even, friendly relations among two groups (cf. Muysken & Smith 1995a: 4; Arends 1995 provides a brief historical background of pidgins and creoles) -in such cases, one speaks of trade jargons (Hock 1991: 522). Language death is the product of intense and uneven power relations, the changes attested in a dying language being in one way or another contact-induced[3]. The opposite, however, is not necessarily the case. It is widely acknowledged (cf. Wurm 1991, for example) that the attitudinal factor can counterbalance the effect of both power and intensity; for instance, if the speakers of the host language cling to their language as a way of ethnic identity; Basque and Hungarian (outside Hungary, e.g. in the Slovak Republic) constitute examples of this situation.

The immediate consequence of language contact (abstract­ed from its historical setting) is what has been called linguistic interference, originally defined by Weinreich as “rearrangement [emphasis added] of patterns” originating in the transfer of elements of one language (e.g. the host language) into the other (the guest language), but later seen as deviation [emphasis added] from the norm of either language" (Lehiste 1988: 1-2); thus Bechert & Wildgen (1991: 3) prefer the term transference, because interference has a connotation of “defect”. In this paper, interference will be understood as the transfer of structural traits of one language into another as a result of the historical process of language contact; this transfer occurs at the phonological, morphologi­cal, syntactic and lexical levels of a language[4]. By separating linguistic change as an independent phenomenon, which can take place in contact situations, S1 acknowledges the difference between contact-induced change (in convergence situations) and the kinds of changes that a moribund language exhibits obviously due to contact (in conflict situations); Sasse (1992a: 16), for instance, insists on maintaining such a difference, “Language contact phenomena (‘borrowing’ in the broadest sense) involve the transfer of substantial material, of patterns and of category distinctions, they can always be explained as the imitation, in one language, of some linguistic trait of another contact language. In the case of decay, however, we are not dealing with transfer in any sense, but with downright loss leading to a heavy expression deficit”.

Other consequences of language contact are code switching, (“switching back and forth between the coexisting languages, such that portions of a given sentence or utterance are in one language, other parts in another language” (Hock 1991: 479), foreigner talk or simplified use of the language by natives to foreigners; these are said to be short-term results of language contact (Bechert & Wildgen 1991: 4). Another consequence of language contact is language intertwining, “a process which creates new languages which have roughly the following characteristics [...] lexical morphemes from one language and grammatical morphemes from another” (Bakker & Muysken 1995: 42); examples of those languages are Media Lengua (Spanish and Quechua, Ecuador), and Michif (Cree and French, U.S.) Besides the emergence of pidgins and creoles, power relations between two (or more) groups are the ultimate cause of other two consequences brought about by situations of language contact. One of them is language shift, “the change from the habitual use of one language to that of another” (Weinreich 1953/1970: 68); this usually occurs when speakers of a clearly non-ruling group give up their language in order to gain acceptance from the dominant one. At this stage, the language succumb­ing to this kind of pressure is at stake. The other consequence is language death.

Although the origin of language death is always due to contact, not all deaths befall under similar circumstances. Campbell & Muntzel (1989) identify four types of language death: sudden death, when all the speakers of a language are suddenly killed or die; the authors mention the Tasmanian case as an example of this type of death. Radical death, is the case where due to genocide speakers stop using their language as a means for self-defence, as happened in El Salvador after the 1932 massacre; gradual death, is the most commonly attested death, occurring as a process starting with language shift to the guest language until the descendants of speakers of the host language have little or no proficiency in that language; bottom-to-top death, an infrequent type of death whereby the dying language is first lost in the family context but is used in special ritual situations; some American languages are mentioned as examples (Chiapanec, Otomanguean; Southeastern Tzeltal, Mayan). The first two types of death constitute abrupt events while the other two are relatively long-term processes. As a process, language death is characterized by a reduction in input by the older generations, “on which new speakers of the language can draw in order to formulate their own internalized grammar” (Hock 1991: 530). Under language death is meant not only the death of a historical language (Chiapanec; Boruca) but also cases where “an exogenous ethnic group moves to a new society where the dominant language is different from its own and is assimilat­ed. When the group’s language ceases to be spoken by its members we have a case of language death, even though the language may continue to be spoken somewhere else” (Hamers & Blanc 1989: 176). In those cases -see Seliger & Vago (1991) for some case studies- it is more appropriate to speak of variety death (dialect, sociolect) as opposed to death sensu stricto, which is the case where the language dies altogether (the terms guest and host language, then, refer to cases of death sensu stricto.)

The last stages of a language are characterized by reduction. Campbell & Muntzel (1989) distinguish between predictable and non predictable changes in the death process. The former entail, at least for phonolo­gy, underdifferentiation of phonological oppositions, maintenance of oppositions common to the host and the guest languages and longer retention of oppositions with a higher functional load. Among the unpredictable changes the authors mention overgeneralization of unmarked features as well as overgeneralization of marked features, development of variability (obligatory rules become optional, do not apply or are substituted), development of irregularity by extremes of regularization and what they call “acts of reception”, which is something like interferences from the guest language. Morphological reduction includes reduction of allomorphy and leveling of paradigms, and syntactic reduction is characterized by “modification of syntactic resources”; as a result, “the rule system of the language undergoes a slow process of atrophy” (Hock 1991: 530). It is important to remark that this “slow process of atrophy” is more readily attested in cases of gradual death, where there are more than one generation of passively competent speakers whose use of the language is every time more reduced. In cases of radical death, both language shift and death occur in one (or two at the most) generation and thus there is no reduction in input for new generations but rather a reduction of output by the shifting generation; thus, it is foreseeable that the terminal speakers’ competence will not be as “degenerated” as in cases of radical death. Similarly, one has to reckon with wide variability in terms of the gradualness/suddenness of the attrition process among individuals; it need not be the case, indeed there is no reason why this should be so, that all speakers exhibit uniformity in decline. This paper illustrates the competence of a terminal speaker of a language undergoing radical death; the remarks made will thus not be made extensive to the remaining terminal speakers.

3. Boruca: the language and its speakers

Boruca (Brunkah ték) was until the first half of this century a relatively healthy language, coexisting with Spanish and spoken in what became the Indian reservation bearing the same name. This area is located in the South-pacific region of Costa Rica and includes the district and reservation of Boruca in the Province of Puntarenas. The population of the district includes some 2500 people, a dozen of who passively understand the language plus six elderly speakers who still use the language in the family environment. Remarkable among them is Doña Paulina Leiva, who in her nineties not only masters the language fluently but especially maintains till today a vigorous Sprachgefühl which has been helpful in the task of recording and describing the language before it passes away. On the other extreme is Don Ernesto González, the focus of the present paper.

Two factors have been pivotal in the death of Boruca; first the universalization of compulsory and free elementary education at the turn of the XXth century in the country, and second, the attitudinal factor of the Borucas, as a result of the former. In terms of the educational system, the older speakers recall how, when the government built schools in the remote areas, the si?kwa (non-Indian) teachers would prohibit the use of Boruca in and out of class, and would go as far as beating them if they spoke the language; two speakers showed the author scars on their heads, caused by hits - with the spine of textbooks - given by their teachers as punishment for speaking Boruca. Prior to the extension of compulsory elementary education there was no systematic repression of the aboriginal languages in Costa Rica. The effect of such repressive practices was a generational surrender to Spanish, a radical case of language shift. Out of fear, the younger generations started using the guest language more and more and relegated Boruca to extremely familiar environments. Don Ernesto’s generation was the one that interrupted what Sasse (1992a: 13) terms the language transmission, that is, “the purposive, directed passing-on of a language from one generation to the next”. Although there are now government-sponsored “language-rescue programs” the fate of Boruca seems to be written; for the language courses offered in elementary school by semi-speakers (the children of the last generation of fluent speakers, who understand the language but do not produce spontaneous speech) and trained linguists do not seem to counter the damage caused in the past. Especially damaging is the general attitude of the Borucas, mainly the younger people, toward the language as unimportant and even shameful; in addition, there is a widespread belief among the young people that the language is unsuitable to adapt to the modern world; a similar attitude in language shift has been reported by Kwachka (1992: 70) for Koyukon (Athabaskan, Alaska).

The Boruca sense of cultural identity is their craftsmanship, not the language. The people living in the reservation feature their handicrafts as a specialty second to none of the neighboring groups in the region (e.g. Térrabas, Cabécar or even Guaymís); they pride themselves on it, but when it comes to language their attitude is that of irreverence, something proper of old people. Craftsmanship, on the contrary, is a valuable tool to survive in the government-promoted marketing of Indian heritage. Perhaps, the reason for this lies in the immediate tangibility of the profits created by the business. This mercantilistic attitude of the Borucas has been featured by historians and voyagers (cf. Fernández 1886); the Borucas are known for their early alliance with the Spaniards and betrayal of other the Indian groups of the country in exchange for good treatment. The neighboring groups refer to them as vividores (‘hustlers’), and attribute the relatively better social condition of the Borucas to this aspect.

In fifty years, the number of fluent speakers decreased drastically to the numbers given above[5]. This means that the death of Boruca constitutes an example of radical death. As mentioned above, this is important because the type of atrophy which is characteristic of the language death process varies depending on the type of death; thus, in cases of sudden and radical death the languages involved do not “decay” in a way similar to the most common cases of gradual death. Aspects of language death mentioned by Campbell & Muntzel (1989), above, such as underdifferentiation of phonologi­cal oppositions, longer retention of oppositions with a higher functional load, overgeneraliz­ation of unmarked features as well as overgeneralization of marked features, and development of variability are absent in the speech of both Don Ernesto and the competent speakers; presumably, such aspects require more than two shifting generations, which is not the case of Boruca. Actually, such phenomena are typical of what have been called rusty speakers, roughly people “who mostly develop from former fluent speakers who were on their way to becoming full speakers, but never reached that stage of competence due to the lack of regular communication in the language” (Sasse 1992b: 61). The last six speakers of Boruca are all terminal, not rusty speakers, having used the language in a wide variety of situations (in their infancy and, as they grew up, with terminal speakers who passed away before them, their spouses in the first place), but who due to repression relegated usage to the intimate environment. This also explains why the majority of the few speakers left still show good command of the language; this is also the reason why Don Ernesto’s case of accelerated attrition is intriguing.

Typologically, Boruca is a nominative-accusative SOV language (strictly SOV in discourse-initial and isolated sentences and OVS elsewhere), with postpositions, a possessor + possessed and noun + modifier order. Verbal categories are expressed by suffixes, with the following categories being grammaticalized: past/non-past~future tense; perfective/imperfective and pluperfect aspect; habitual and terminative aspect is expressed by separate morphemes accompanying those previously listed; these have been termed “suffixes of the second series” (Constenla and Maroto 1979) because their presence is not obligatory as is that of the perfective/imperfective/pluperfect set. Progressive and ingressive aspect is expressed by means of periphrasis. In addition, there is deontic, desiderative, and imperative modality, expressed by bound morphology. Nominal categories include the plural, expressed by the particle róhk, which also marks verbal plurality. The plural marker is used only when the plural pronouns appear in subject function (the other two functions of these pronouns, besides subject and object, are obliques and possessors). Participants are not cross-referenced on the verb, their role being made explicit by word order and by a set of emphasis and focus markers, whose use is a bit complex (see Quesada in press for details). The complex sentence is expressed asyndetically, save for purposive clauses, which are introduced by the marker chá, which is a grammaticalized form of a verb meaning ‘want’, and causative clauses, which are marked by the particle úge?, ‘because’. Temporal and locative clauses are introduced by adverbs in the respective functions. Boruca has two tones (high, orthographically represented as / over a vowel, and low, unmarked). Sketches of Boruca grammar are found in Constenla & Maroto (1979), Quesada Pacheco (1995); aspects of its morphosyntax are discussed in Quesada & Quesada Pacheco (1995), Quesada (1996), Quesada (in press). The first reference contains stories narrated by deceased speakers of the language, including Don Ernesto.

4. Don Ernesto’s performance in Boruca

The focus of the present paper is the terminal speaker, Don Ernesto González (henceforth DE), who in his early nineties has little or no chance to practice the language because he has little contact with the remaining old speakers; his older – monolingual - children, with whom he lives, feel ashamed of the language and do not hide that feeling; they even seem hostile toward linguists interested in any information he might provide. As it seems to be the rule with the old Borucas, he was also physically punished in school whenever he dared speak the language. It should be noted that despite his age DE enjoys a mentally health condition, being able to recall stories, legends and anecdotes in Spanish. His wife was fluent in Boruca, but, being members of the shifting generation (cf. Quesada Pacheco 1995: 13), they raised their children in Spanish. After he became a widower in 1984, his regular use of Boruca ceased, a “marked reduction of language function due to narrowing of social communicative situations [already scarce in DE’s life, DQ] and narrowing of functional range” (Craig 1992: 18) ensued. Prior to that, DE used to be a competent speaker of the language, as can be seen from the various stories narrated by him, and which are contained in Constenla & Maroto (1979). None of those stories catches the reader’s eye in any way because of its grammar. Between that time and now, DE’s performance has decayed considerably. The remainder of this section intends to provide a glimpse of his current performance in Boruca.

Since the death of Boruca is not gradual but radical and since we are dealing here with the shifting generation[6], the competence of the remaining speakers, though jeopardized by the use of the language almost exclusively in the intimate context -and even here performance is not necessarily constant-, has remained latent and phonetic/phonological interferences are rare. Indeed, it is not the case that all terminal speakers must show under or overdifferentiation of phonological oppositions, Elmendorf (1981: 40) reports no reduction or replacing language influence in the last speaker of Wappo (Yukian, California), a language which, interestingly, suffered a radical death as Boruca.

(S2) illustrates a merged inventory of the consonants of the two languages in contact. The underlined segments are those exclusively present in Spanish, while the double-underlined ones represent those exclusively present in Boruca.

(S2) Phoneme inventory of Spanish/Boruca

                                                           Ld       Bb     D       A       P       V       G

                                    vl.                              p      t                                  k      ?


                            vd.                             b      d                      g                                                         vl.                                                           ts      č



            Fricatives   (vl.)               f                              s       š              h

            Nasals                                           m        n              ñ      ŋ

            Vibrant (simple)                                                  r

                        (multiple)                                                  rr

            Glides                                                                                 y      w

There is not so much disparity between the two systems phonologically speaking. Boruca has a five-vowel system similar to that of Spanish and the consonant systems of both languages differ in the absence of /p/ (except for loan words), /f/, and /rr/ (multiple vibrant), plus the presence of the glottal stop, /?/, the voiceless alveolar affricate /ts/, and the phonological opposition /š/ vs /č/ in the former. In addition, there are the two tones (high and low), absent in the host language.

DE’s speech shows only one symptom of phonetic interference, namely the Spanish rule of voiced stop lenition in intervocalic position, both in elicited sentences (1) and (2) and in “fluent" discourse (3), (4):

(1)       daba-krá  [da.ßa.krá]


            ‘(he) came’

(2)       doyéng-ra  ba-ng  se?sat!  [do.žéŋ.ra.ßaŋ.se$.sat]

                sit-pres   2sg-foc   quiet

            ‘be quiet!’

(3)       ya    ta  ing    daba-krá españole  róhk  [ja.tiŋ.da.ßa.kráño.le.róh]

                that in  3-foc come-perf Spaniard pl  

            ‘there came the Spaniards’

(4)       ihchí tebek  so?  ka... uge?  di?   teng-íra   [ih.čí.te.ß$.ka…$.di$.teŋ.í.ra]

                and   snake  big  in... caus  1pl   cure-imp

            ‘and because of the old snake we got well’

In (5) the 1pl  pronoun di?, is repeated seven times, while trying to remember the story line in Boruca (an attempt that failed; see (10), below); in the sequence, the glottal element is first lost, feeding the application of the Spanish rule:

(5)       [di? di? di? di di Di Di]

DE’s tones is Boruca are stable. He is able to both produce and identify pairs such as kup (‘seed’ vs kúp (‘egg’).

The morphology seems to be equally stable in DE’s passive competence. When elicited, he can produce perfective/imperfective (-íra vs. -kra, respectively) pairs with relative ease. The same is true to other complex areas of Boruca morphosyntax such as the focus, specificity and emphasis markers, which perform the function of a non-existent voice system. His syntax follows the rules of word order: SOV in discourse-initial (and isolated sentences) and OVS elsewhere, with the corresponding subject pronoun set in each case (but see below).

DE’s main trouble with the language is both vocabulary and connected speech, which he simply cannot sustain. Concerning vocabulary, DE confuses opposite pairs like tru? vs. uht (‘fall’ vs ‘climb’); he also confuses members of the same lexical field, as in tru?, which for him means both ‘fall’ and ‘go down’, clearly forgetting the corresponding word for the latter, beht, which he confused with uht (‘climb’). Thus when asked to give the equivalent to ‘I came down the tree’ he produced (6), which exhibits perfect syntax but wrong lexico-semantics:

(6)       at    ki             tru?-krá           krang ka

            1sg subj     fall-perf      tree in

            ‘I fell from the tree’

Similarly, when asked about the meaning of at ki behtkrá (‘I came down’), he provided the Spanish equivalent me subí (‘I went up’). On occasions he would simply be unable to give the words asked, but would make clear that “there does exist an equivalent”.

Examples of other mismatches when elicited are (elicitations were in Spanish):

 (7)      asked                      A snake bit a man

            provided   bi?  kita  e?tse  tebek ang      ba tu?-kra

                                   at   night one   snake foc    2sg bite-perf

            means         ‘At night a snake bit you’

Again, the phrase structure is grammatical but the entries do not correspond to what he was asked. In addition to the missing adverb in the elicitation, he uses the subject focus marker ang for no apparent reason; this form is strictly discourse-bound and the examples were contextless elicitations. (7) already illustrates DE’s main problem: correct syntax but defective discourse.

(8)       asked                      That man a snake bit (That man was bit by a snake) (Y-movement)

            provided   chi?   abih   ki     tebek  ki     tu?-kra            and

                                   dem  man   spec snake  def  bite-perf

                                   chi?   abih  ki     e?tse tebek   ang     i           tu?-kra

                                   dem  man  spec one  snake  foc            3sg    bite-perf

            means         ‘That man bit the snake’ and ‘That man, a snake bit him’

Interestingly, he supplied two alternatives. Whether this is a sign of doubtfulness is unclear; what is important is that he “calqued” the order of appearance of the entities named in Spanish regardless of the grammatical relations involved; thus Spanish OSV is reproduced as SOV, but ‘that man’ is sentence-initial and ‘the snake’ is preverbal; notice that the eliciting sentence contains the indefinite article, while he produced a definite NP ('the snake'). The second alternative he volunteered is correct assuming there was a pause between chi? abih ki and the rest of the sentence, with i in a resumptive function (otherwise, this would be a highly marked, indeed ungrammatical, construction in Boruca). Nevertheless, the presence of ang -the subject focus marker (cf. Quesada 1996)- is at odds with the communicative structure of the sentence, where the object is topicalized through fronting. Again this is a syntax-discourse mismatch. An isolated case of language attrition that resembles the performance of a semi-speaker is (9):

(9)       asked                      I saw a snake

            provided   *at   ki            e?tse  tebek    at     tu?-kra

                                   1sg   subj   one    snake    1sg   bite-perf

            means         * ‘I a snake you bit’ (‘I bit a snake’ (?)  + ‘A snake bit you’ (?))

The ill-formedness of (9) might probably be the result of a clash between syntax and world-knowledge due to faulty lexical choice. DE seems aware of the strict preverbal placement of objects in Boruca and it might have been the case that he realized that humans normally do not bite snakes; hence the object pronoun at. And the reason seems to be the wrong correspondence between the expected word isht  (‘see’) and that provided, tu? (‘bite’).

The discourse level plays an important role in Boruca; rules of anaphora and reference in general, as well as the focus system are dependent on the status of the referents in a given discourse as either given or new, as in some Salishan languages (see Beck in press). It is here where DE’s command of the language reveals its advanced state of attrition. The following short text, for which he needed some minutes to concentrate and which, as will become evident below, he could barely bring to a close, shows the performance of a terminal speaker of a shifting generation, whose language is undergoing a radical death[7].

(10)    a. At   tsasúh  róhk ki  at    yené... Drake  ta e?tse... sukia   ang   kawí?-ira ye?é.

1poss  old     pl  def 1sg to soon... Drake  in a... medicine man rel  live-imp there.

b. Ya   ta  i-ng            daba-krá...   españole róhk... ihchi-ng... ya   ta  i-ng

   That in  3-foc       come-perf... Spaniard  pl... and-foc... that in  3-foc

c. raht-krá    rohk...   ch’ing  warísh-ira     we?é // Pero... néngkra   ka...

   leave-perf  pl...  to.foc  come up-imp  here //        But...   road        in...

d. di? róhk      di?                 // di?                [...]  di?        tsasúh róhk ki   i?    ai?-kra       uge?...

   1     pl         1pl.poss// [...] 1poss old    pl            def 3sg kill-perf    caus...

e. e?tse resh-íng // e?tse resh-íng // yu?-kra...        ch'-i-ng   ya-íra... apresarlo a San José

   one    only-foc // one only-foc // take-perf... to-3-foc  go-imp... to arrest him in S.J.

f. o quién sabe dónde... pero que sí lo llevaron amarrado // el el el... el... le

   or who knows where... but they did take him all bound up // the the the... the... him

g. sí, el e... e [...] e ha... i?      u?se?kra... verdá...  el capitán, o sea ell // vino hasta aquí

   yes, the uh [...]uh ha... 3poss boss...    you see... his captain, that is, the //came here

h. a...  ch-ing        i?      i?...      i?       wi?-ira      róhk... estee...  di?... di? di? di? di? di? di?

    to    to-3sg.foc 3sg 3sg...  3sg    carry-imp  pl...     uhh...         1pl.poss

i. [...] diiii?           mmm [que nom no no puedo...]  di?         wa? rohk        o   di? //

   [...] 1pl.poss mmm [I can’t]                                   1poss    child pl    yes or 1pl.poss //

j. [sí tiene] [...]                     í aí?-íra do?sh                                // y lo entregaron allá //

   [there’s a way to say it] [...] 3sg kill-imp like  // and they handed him over there //

k. Shi-krá              róhk y los...    los   estee [...] los marineros...  ya-kra   i       yu?  i...

 come back-perf pl  and the... the   uuh   [...] the sailors...        go-perf 3sg take 3sg...

l. i  kwing  í...    Ahí fue a dejarlos...                     ya   ta i-ng    ya-krá   i     kwing í //

 3sg put   to... There's where he left them...         that in 3-foc go-perf  3sg put  to//

m. y de ahí no más...             pero sí, lo terminaron...                ¿entiende? [...]

   and from there it was just... but they did, they terminated him... understand? [...]

Free approximate translation:

‘Among my elders in Drake there lived a medicine man. There came the Spaniards and then they left to come up here. But because on the way our elders killed him... only one took and left [Spanish] to arrest him [?] in San José or who knows where, but they did take him [?] all bound up. The the [Boruca] his boss [Spanish] the captain, that is... came up here [to Boruca] to take them [?]... uh... Our... [I can’t]... our children yes or our [Spanish] [there is a way]... [Boruca] to kill him [?] too [Spanish] and they handed him over there. [Boruca] They/he came back [Spanish] and the sailors [Boruca] went to take him... to put him there. [Spanish] There’s where he [?] left them [?]; [Boruca] there he [?] went to put him [?]. [Spanish] And that was it... but they did terminate him... do you understand?

Several aspects are worth noting here; first and foremost is DE’s inability to tell the story without resorting to Spanish; this is an indicator of lack of both vocabulary and discourse grammar competence to link the story line, a fact acknowledged by DE himself on line i. [no puedo, 'I can't']. Second, the long pauses and the use of Spanish even for murmuring [este, e... e...] are also symptomatic of vanishing competence. Third, the lack of textual coherence - that is, the fact that although the sequence of events might be orderly and clear in DE’s memory, in the actual text production there is no clear arrangement nor natural flow of events - reveals how poor speech production has become and, more important, it shows that code-switching, though apparently an alternative to solve communicative pressure, does not necessarily amount to textual clarity but merely plays a lacunae-filling function[8].

The aspects mentioned in the previous paragraph indicate that the textual level can be among the first aspects of competence that can be touched by language death (at least by the radical type); in turn, this implies that syntactic rules in isolation are active in the mind of the speaker, but when it comes to higher levels of organization and structuring, the ability to handle them is simply not there. This is why DE keeps phrase structure intact, has no problem in the use of emphasis and topicality markers ang and ki, and has good command of both morphology and phonology (save for the interferences mentioned above); but his sentences in connected speech are “chopped”, or, better, his text is a mosaic of, independent, isolable, sentences. When it comes to reference and anaphora, he fails to establish the connections properly. In fact, as early as line d. it is practically impossible to identify both participants and situations. Lines d., e. and f. seemingly make reference to a cause-effect situation whereby someone killed someone and, as a result, was arrested and taken to San José. However, neither this author nor Doña Paulina Leiva (see below) could, with complete certainty, identify the participants involved. An aspect that might be playing a role in all this is the fact that the Boruca plural marker róhk can be disposed of whenever identification of participants is not threatened. The third person pronoun i? can thus be both singular and plural. Knowing in his mind who/what he is talking about, DE simply deletes the plural marker in places where fluent speakers would retain it and makes use of it at will and not depending on his - not necessarily the hearer’s - linguistic (textual) needs. As a result, in this crucial part of the story it becomes impossible to ascertain who did what to whom[9]. This becomes even more critical when one realizes that the rest of the story apparently revolves around this third person participant: on line f. this participant is taken hand-bound; on line j. apparently another participant is killed (that is why there is do?sh (‘too’)) and is handed over. On line k. a plural participant comes back, it could be the elders mentioned at the onset (line a.) or the Spaniards (introduced on line b.), who are apparently the same as the sailors mentioned on the same line (k.) To add more confusion, it is a third person singular agent who leaves a plural patient referent in the Spanish interference of line l.: Ahí fue, (go-3sg.perf) a dejarlos (‘to leave them’). But then again, on the next line (m.) it is a third person plural agent who “terminates” a third person singular patient: lo (3sg.obj) terminaron (terminate-3pl.perf). At this point there is complete referential chaos in the narration.

DE’s text was shown to linguistically lucid Doña Paulina Leiva for inspection. Although she suggested a couple of lexical changes, her main comment was that the text was incomprehensible because “DE does not know what he is talking about and does not speak the language as it should be spoken”. This simply means that there being no textual coherence, it is impossible to follow the story line. Asked whether DE was able to speak the language in the past, she said he did; she attributed his present inability to his wife’s death who “did speak properly” (ella sí lo hablaba clarito). Concerning the text in (10), she suggested a couple of changes, “although the story will still remain unclear”, namely on line a. instead of at ká yené it should be at ka yeng-krá (1sg to teach-perf; ‘told me’); this error seems to fall within DE’s observed pattern; it is a problem of word retrieval rather than grammar; it involves the exchange of a word, yenkrá, for another phonetically similar, yené. That there is inflectional morphology involved -exchanged for an adverb- does not obliterate the fact that DE knows the Boruca phrase structure in that he left what was intended as a verb in final position, that is he kept the SOV order intact.

On line d. Doña Paulina recognized a syntax error, namely the absence of the marker ang following the subject (di? tsasúh róhk ki) in accordance with a rule of Boruca syntax that requires subjects of subordinate clauses to be ang-marked (cf. Quesada-Pacheco 1995: 119f, Quesada 1996). This is one of the few syntax errors of DE. Finally, on line e. she, a bit exasperated, asked ‘one what?’ (¿un qué?). This is precisely the line where the chaos begins. After that, as was discussed above, it is impossible to follow the story.

5. Conclusion

The admittedly brief analysis reveals an interesting case of a terminal speaker, who in only twelve years of interrupted practice shows a command of his language that very much resembles that of a rusty speaker. Even if the difference between these categories/labels were of degree and not discrete -as the definitions in the literature lead one to believe, e.g. Sasse (1992a, b)- DE’s performance shows an extremely accelerated state of attrition. Sasse (1992b) mentions the following seven features typical of semi-speakers: loss of subordinative mechanisms, loss of systematic integration, breakdown of grammatical categories, agrammatism (total disintegration of the morphological system and “rampant” analogy), word retrieval problems, extreme phonological variation, and phonological hypercorrection. All these features but one, namely word retrieval problems, are absent in DE’s speech. As we saw in 4. it is this problem plus the inability to connect speech, that characterizes this terminal speaker’s performance. The reason for this, I suspect, is one of individual character. DE’s personality was characterized as a rather laconic and contemplative fellow by Constenla & Maroto (1979: 44) and the present author can verify that assessment. Age cannot be adduced here since Doña Paulina, the fluent and lucid oldest speaker, seems not to be affected by this, as is also the case of Doña Rafaela, who is currently helping Quesada-Pacheco (personal communication) translate St. John’s Gospel. Elmendorf (1981) reports another situation, where a terminal speaker of Wappo was able to produce fluent spontaneous speech after sixty years of interruption. Elmendorf’s remark that “in any investigation of terminal speakers, it will always be necessary to consider specific details of the social biographies of these individuals” (1981: 45) proves especially adequate in the present case. It seems, thus, that personality traits such as talkativeness and/or outgoingness also come into play in the process of language decay.




1995  “The socio-historical background of creoles”. In Arends, Muysken & Smith (Eds.): 15-24.


1995  Pidgins and Creoles: an introduction. Amsterdam. Benjamins.


1995  “Mixed languages and language intertwining”. In Arends, Muysken & Smith (Eds.): 41-52.


1991  Einführung in die Sprachkontaktforschung. Darmstadt. Wissenschaftliche Buchgesellschaft.


in press “On the syntactic expression of topic and rheme in the Salishan languages. To appear in Proceedings of the Workshop on Focus: University of Massachusetts at Amherst.


1989  “The structural conse­quences of language death”. In Dorian (Ed.): 181-196.


1979  Leyendas y tradiciones borucas. San José. Editorial de la Universidad de Costa Rica.


1992  “Language shift and language death: the case of Rama in Nicaragua”. International Journal of the Sociology of Language 93: 11-26.


1991  “A changing language situation: the decline of Dyirbal”. Language in Society 20: 183-200.

DORIAN, N. (Ed.)

1989  Investigating obsolescence. Studies in language contraction and death. Cambridge. Cambridge University Press.

1977  “The problem of semi-speaker in language death”. International Journal of the Sociology of Language 12: 23-32.


1981  “Last speakers and language change: Two Californian cases”. Anthropological Linguistics 23 (1): 36-49.


1886. Colección de documentos para la historia de Costa Rica. 10 vols. París-Barcelona: Viuda de Tasso.


1875  “Tribus y lenguas indígenas de Costa Rica”. In Fernández (ed.) Vol 2: 303-486.


1992  “The process of language death in a Mayan community in Southern Mexico”. International Journal of the Sociology of Language 93: 53-66.


1989  Bilinguality and Bilingualism. Cambridge: Cambridge University Press.


1991  Principles of Historical Linguistics. Berlin. Mouton de Gruyter.


1992  “Discourse structures, cultural stability, and language shift”. International Journal of the Sociology of Language 93: 67-73.


1988  Lectures on language contact. M.I.T. Press.


1920  Zentral-Amerika. Vol. 1 (Die Sprachen Zentral-Amerikas.) Verlag Dietr.

MUYSKEN, P. & SMITH, N. (Eds.)

1986  Substrata versus Universals in Creole genesis. Amsterdam. Benjamins.

1995a   “The study of pidgin and creole languages”. In Arends, Muysken & Smith (Eds.): 3-14.

1995b   “Reflexives”. In Arends, Muysken & Smith (Eds.): 271-288.


1996  “Enfoque y topicalidad en boruca: el caso de ang”. Estudios de Lingüística Chibcha  15: 43-58.

2000  “Another look at Spanish-speaking (Central) America”. Hispanic Research Journal 1 (3): 229-242.

in press “En route to subject-coding: the case of Boruca”. To appear in Theoretical Linguistics 27.


1995  “Boruca ki: de definido a específico y la hipótesis de las funciones contiguas”. Estudios de Lingüística Chibcha 14: 89-99.


1995  Hablemos boruca. San José. Ministerio de Educación Pública.


1992a   “Theory of language death”. In Brenzinger, M. (Ed.). Language death: Factual and theoretical explorations with special reference to East Africa. Berlin. Mouton de Gruyter: 7-30.

1992b   “Language decay and contact-induced change: similarities and differences”. In Brenzinger, M. (Ed.). Language death: Factual and theoretical explorations with special reference to East Africa. Berlin. Mouton de Gruyter: 59-80.


1985  Young People’s Dyirbal. An example of language death from Australia. Cambridge: Cambridge University Press.

SELIGER, H & VAGO, R. (Eds.)

1991  First language attrition. Cambridge. Cambridge University Press.


1988  Language Contact, Creolization and Genetic Linguistics. Berkeley. University of California Press.


1991  “Terminal-fluent speaker interaction and the contextualization of deviant speech”. Journal of Pragmatics 15: 153-173.


1953/1970     Languages in contact. The Hague. Mouton.


1991  “Language death and disappearance: causes and circumstances”. In Robins, R & Uhlenbeck, E. (Eds.). Endangered languages. Oxford. Berg: 1-18.


[1]    This paper is a revised version of a paper that had originally been written for the International Congress of Americanists 1997, in Quito, which the author could not attend due to a last minute situation. The paper was thus not read at that forum. Between July 1997 and its present form two things have happened. Sadly, the speaker whose speech constitutes the subject of this study, Don Ernesto González, passed away in 1998. Second, this paper has circulated among colleagues for suggestions, among them the most valuable coming from Miguel Quesada Pacheco. The author dedicates this paper to the memory of the former and acknowledges the comments from the latter.

[2]    Usually the terms abandoned and target language are used to refer to the host and the guest languages, respectively (cf. Sasse 1992a: 12-13). While I see no objection to the term ‘abandoned’, I think that ‘target’ is not faithful to the contact situation bringing about language shift and death, in that it presupposes willingness on behalf of the shifting population to adopt the guest language. And this is by no means always the case.

[3]    As pointed out in the next paragraphs, the structural changes in a dying language reveal intense contact; it is not difficult that such phenomena as underdifferentiation of phonologi­cal oppositions, overgeneraliz­ation of unmarked features as well as overgeneralization of marked features, and the so-called “acts of reception” (cf. Campbell and Muntzel 1989) resemble structural interference, not surprisingly from the imposed language.

[4]    Thomason & Kaufman (1988: 120) regard the number of speakers as an indicator of healthiness or likelihood to cause interference.

[5]    The number of speakers in the middle of the XXth C. is unknown. Quesada-Pacheco (1995: 13) mentions the existence of a few monolingual Boruca speakers even during the first half of the XXth C. The last speakers, Don Ernesto González and Doña Paulina Leiva, told this author that many people used to speak the language when the two of them were little, and that it was precisely the punitive measures of the school system caused the language to be reduced to the family context.

[6]    The Borucas had been bilingual since the colonial period and preference about the use of Spanish has been reported as early as late XIXth C. (cf. Gabb 1875/1976: 405), but the decisive language shift took place during this century, precisely in DE’s generation. Quesada-Pacheco (1995: 13) mentions the existence of a few monolingual Boruca speakers even during the first half of the XXth C.

[7]    The following notational conventions will be used: ... is a short pause, // is a long pause [...] a much longer pause; boldface will be used for case of code switching, and extratextual comments will appear in boldface and brackets.

[8]    That is, in cases of code-switching the speaker retrieves the item that first appears in his bilingual competence when under communicative pressure. In this case, however, there seems to be no two items available corresponding to two codes; rather, one of the codes has decayed in such a way that its items are non-existent anymore. This poses the question of whether there is actually code-switching here.

[9]    This is why in the free translation third person participants are marked [?].