Linguistics in Amsterdam 4-2 (september 2011)Magaly Grández Ávila: Language transparency in Functional Discourse Grammar: The case of Quechua1

To refer to this article use this url: http://www.linguisticsinamsterdam.nl/04/nr02/a02

4 Accounting for degrees of transparency within levels

A number of properties at the Morphosyntactic Level and within the Phonological Level are expected for a language to be considered transparent.

4.1 At the Morphosyntactic Level

next section

The following properties within the Morphosyntactic Level are expected for Quechua to be considered a transparent language:

4.1.1 No expletive elements

Expletive elements are introduced at the Morphosyntactic Level in situations in which no interpersonal or representational material, necessary for the formation of an appropriate structure, is inserted in a certain slot. Expletive forms may replace an argument or a predicate as in the following English examples:

(Hengeveld & Mackenzie, 2008: 10)

In (53), the expletive pronoun it serves as an argument, which is required to fill the obligatory Subject slot in the clause. In (54), the expletive form there is inserted in the absence of an Ascriptive Subact to fill in the Subject slot, which at the same time requires the introduction of copula be as a support verb in the predicate slot.

According to Weber (1989), in Quechua, the existence or presence of an entity must be asserted by means of the copula verb ka-. But, unlike the English dummy verb be in (2), ka- functions in this case as a lexical verb within a regular one-place predication frame, in which the existence itself is ascribed to that entity by means of the copula, used as a lexical expression, comparable in English to a lexical verb such as ‘exist’, rather than as an expletive element, as shown in (55):

(Weber, 1989: 24)

In this respect, Quechua can be considered to be transparent. With respect to event-descriptions, such as those realized by weather predicates like (53) above, Quechua makes use of a zero-place predication frame which, unlike English, does not require the insertion of an expletive element in the subject slot at the Morphosyntactic Level, as can be seen in (56) and (57) below:

(Weber, 1989: 30, 529)

Despite the absence of an expletive element in the subject slot, Quechua is not completely transparent in this respect due to the presence of an expletive default third person marking on the verb as in (57), which appears in (56) incorporated in the perfect marker –sha.

4.1.2 No tense copying

Tense copying is a mechanism of operator agreement between main and subordinate clauses, whose realization pertains to the Morphosyntactic Level. It occurs when information pertaining to the tense operator of the main Clause is copied to the tense operator of the subordinate clause as a result of a copying rule applying to the grammar of a given language, in which case transparency with respect to the Representational Level, where the nature of the operators is triggered, results affected. In Quechua, such a copying rule does not apply. Instead, it does have an absolute tense system, expressed in main clauses, and a set of relative tense markers, namely an anterior (ANT), simultaneous (SIM) and posterior (POST) marker, which are restricted to subordinate clauses, as in the examples below. In this respect, Quechua can be considered then to be transparent.

(Weber, 1994: 107, 114, 102)

4.1.3 No raising

The morphosyntactic phenomenon referred to as ‘raising’ involves cases in which a constituent semantically belonging to a subordinate Clause appears as a constituent of a superordinate Clause. As Hengeveld and Mackenzie (2008:368) explain, the triggers for raising may be interpersonal, representational, or morphosyntactic in nature. Whatever the nature of the trigger for this dislocation is, it clearly affects the semantic and syntactic integration of constituents and so the ideally one-to-one mapping between units at all levels expected for a transparent grammar.

According to Weber (1994: 99-121), Quechua does present the phenomenon of raising in which case not only the Subject of the subordinate Clause can raise, but also the Object. Object raising occurs especially with infinitival complement clauses, namely with complement taking predicate such as muna ‘want’ and puyri ‘be able’, as shown in the following examples (Weber 1994:104):

As for Subject raising, it generally occurs with complement of sensorial verbs, in which case the Subject of the subordinate clause raises to the Object position of the superordinate clause, taking the accusative marker –ta, used as Object marker in Quechua, as shown in the following examples (Weber 1989:290)

As for the trigger of raising in Quechua, it appears to be semantic in nature, as it applies with a limited number of complement taking predicates, in which case the raised Subject or Object bear accordingly the semantic function of Actor or Undergoer and denote an entity that is actively involved in bringing about the situation denoted by the entire sentence, as Bresnier (1988, cited in Hengeveld and Mackenzie 2008: 370) also suggests for Tuvaluan. Thus, as for what the phenomenon of raising concerns, Quechua cannot be regarded as transparent.

4.1.4 No grammatical gender, declination or conjugation

Morphosyntactic phenomena such as grammatical gender, declination and conjugation involve the application of inflectional operations to lexical roots, e.g. nouns, verbs, adjectives, which are not required by the basic meaning of the concept they express, but rather, by the grammatical environment in which they occur. Such phenomena are generally required in order to create fully form words, ready to be integrated into discourse. As these morphosyntactic operations cannot be mapped directly onto units from the Interpersonal or Representational Level, they are not expected in transparent languages.

Grammatical gender refers to the grammatical class of the noun, which gets reflected on the behaviour of associated words by means of the morphosyntactic mechanism known as gender agreement. As Hengeveld and Mackenzie (2008:396) explain, gender agreement is automatic and triggered by inherent features of the head noun that cannot be predicted on the basis of its meaning. Quechua does not bear grammatical gender, but it does present a gender classifier which specifies the sex of the entity being referred to by the noun it modifies forming a compound, as in warmi dansa ‘woman dancer’.

As for declination, it refers to inflectional operations on nouns, pronouns and adjectives in order to indicate features such as number, case, gender and possession. Languages may present sets of declined forms according to word patterns, which may result into complex declension systems. For instance, Latin presents a system of five different declension paradigms that distinguish seven different grammatical cases according to number, depending on the ending forms. Quechua does not present a declension system as such. Nouns carry distinctive morphemes in order to indicate number, case and possession, which are not subjected to a morphological paradigm and are agglutinating in nature, which means that they can be mapped directly onto units from the higher levels of representation.

With respect to conjugation, it refers to inflectional operations on verbs, which serve to indicate features such as person, number, gender, tense, aspect, mood and voice. Verbs may be subjected to conjugation paradigms according to its formal patterns, leading to a complex verbal system, as is the case in Romance languages. In Quechua, there is not a conjugation system a such. Verbs carry distinctive morphemes in order to indicate person, number, tense or/and aspect, which are agglutinating in nature, that is they can directly be mapped onto Interpersonal or Representational units. In conclusion, Quechua can be said to bear transparency with respect to the morphosyntactic pheonomena in question.

4.1.5 No agreement (but pronominal arguments)

As explained by Hengeveld and Mackenzie (2008:350), agreement is a mechanism by which information properly pertaining to a single element of the construction under consideration is copied to one or more other elements. Rules of agreement basically operate at the Morphosyntactic Level as they are applied, where relevant, once all slots in a template have been filled by material from the Interpersonal and Representational Level. This explains why such a mechanism is not expected in transparent languages.

Of special interest in this section is the distinction made in FDG between agreement and cross-reference, associated with the expression of participants, which is relevant not only at the Clause layer, but also at the Phrase layer. At the Clause layer, languages may present argument agreement on verbs, which is non-referential but syntactic in nature, as they are strictly the result of copying rules from a target to other associated forms. Subject agreement on verbs, for instance, applies especially in non pro-drop languages such as French and English. Contrarily to agreement, cross-referencing marking is capable of referring by itself and therefore constitute the bound expression of Referential Subacts. In Quechua, person marking on the verb is cross-referential in nature as it is sufficient by itself and may optionally be expanded by a lexically realized argument (see section 3.1.1 above). The same is true at the Phrase layer, for instance in nominal constructions involving possessions, in which case the optional (pro)nominal possessors are cross-referenced on the possessum by means of suffixation of possessive person markers (see 40 above), which have referential force by themselves and therefore reflect the presence of Referential Subacts.

It is important to mention that neither argument agreement nor operator agreement rules such as those related to gender and number appear to apply in Quechua. In this respect, Quechua can be regarded then as transparent.

4.1.6 Phrase marking through clitics rather than head marking through affixes

A clitic is a bound morpheme that constitutes a morphosyntactic word in itself that may function at a phrasal or clausal level, which is phonologically dependent in the sense that it needs to bind (cliticize) to an element, known as host, within the phrase or clause, irrespective of the nature of that element. At the Phrase layer, clitics may fulfil various functions as a result of specifications made at the higher levels. Quechua does present a number of Phrase-layer clitics, which may serve to express semantic functions, as explained in section 3.2.4, but also interpersonal functions such as Topic, in which case the topical marker –qa is used to mark information contained in a phrase as the most topical element in discourse, as shown in the following example (Weber 1989: 408):

Contrarily to cliticization, affixation is a morphological operation which is sensitive to nature of input, in the sense that affixes, i.e. morphemes with grammatical content, may only occur in conjunction with a stem of a given class. Languages that present head marking through affixes, instead of phrase marking though clitics, lack transparency in the sense that affixation generally involves the application of morphosyntactic rules and specification that do not bear a pragmatic or semantic counterpart, affecting then the ideally one-to-one mapping between units at all levels of representation. The following example from Jarawara (Dixon 2000, cited in Hengeveld and Mackenzie 2008:307) illustrates how the nucleus or head of the Noun Phrase is marked by means of affixation of a gender marker, which is determined by the gender of the possessor rather than by the inherent gender of the inalienably possessed noun:

In Quechua, phrase marking by means of cliticization, rather than head marking affixation, is a salient feature that also characterizes languages such as Kharia (Leufkens, this issue), which accounts for transparency in their grammars.

4.1.7 No fusional morphology

As explained by Hengeveld and Mackenzie (2008: 301), fusional languages are semantically opaque as there is no one-to-one relation between a unit of form and a unit of meaning, as illustrated in the following Spanish example:

Quechua is a language which does not present a fusional but an agglutinating morphology, which means that there is a one-to-one relation between morphemes at the Morphosyntactic Level and units at the Representational and Interpersonal Levels, as can be noted in the example below, whose representation at each of the levels mentioned before clearly reflects the morphogical transparency of this language:

[Note 10]
(Weber, 1989: 88)

However, Quechua does also present certain degree of opaqueness in its morphology, especially with respect to transitional markers. A transition is defined by Weber (1989) as a complex of verbal suffixes that function together to indicate the person of the object and subject, as well as the tense/subordination relationship. In most cases, transitions are agglutinating in nature, which means that the boundaries between suffixes forming a transition are clear-cut and so semantically transparent, but in other cases, especially those involving a third person object, the distinctions between the subject and object person and the tense/subordination relationship is conveyed by means of a single form, as shown in the example below (Weber 1989:79):

In the morphology of Quechua, we also find cases of stem alternation in verbs affecting the ideally one-to-one mapping between units of meaning and units of form. Stem alternation is a property of certain verbal roots which derive historically from the combination of a monosyllabic verbal root and a derivational suffix which is normally subjected to morpho-phonemic lowering when followed by certain other suffixes. For example, the final high vowel of the verb miku- ‘eat’, which derives historically from mi-+ku-, is lowered to /a/, and so miku- becomes mika-, when followed by one of a certain group of suffixes which triggers the property of lowering, such as the directional suffix -mu ‘afar’, as shown in (74) below:

(Weber, 1989: 29)

Thus, stem alternation in verbs occurs basically because several of the incorporated derivational suffixes, such as –ku in miku- ‘eat’, present the property of lowering in different morphological contexts and carry such property into the verb stem when the verb+suffix becomes one unit.

In conclusion, Quechua can be said to be transparent, though to a certain extent, with respect to the feature of fusion.

4. 2 At the Phonological Level

The following properties within the Phonological Level are expected for Quechua to be considered a transparent language:

4.2.1 No diphtongization or nasalization

As stated by Andersen (1972:11), the phenomenon of diphthongisation can be understood in several ways. From a synchronic point of view, it can refer to the process whereby a diphthongised vowel (a diphthong) derives from a single underlying segment (a monophthong) as a result of the application of certain phonological rules. For instance, the alternation of the Spanish phonemes /e/ and /o/ and their corresponding diphthongised forms /je/ and /we/, attested in derivational and inflectional processes whereby the original stress is shifted, has been the reason of a number of studies attempting to account for an underlying phonological rule, which is, undoubtedly, confronted with a number of exceptions. As suggested by Malkiel (1966: 433), the native child and the second language learner have to deal with the difficulty of discriminating cases in which such a rule is required, and be able to remember, for example, that a verb like defender /defenٰder/ ‘to defend’ has a diphthongised form such as defiendo /deٰfjendo/, when conjugated in the first person of the indicative present, whereas a verb like ofender /ofenٰder/ ‘to offend’, phonologically similar to the former one, does not undergo diphthongisation when conjugated. The existence of phonological operations, which do not have a representational or an interpersonal basis, affects the transparency of a language.

The syllabic structure of Quechua does not allow for diphthongised vowels to be the head of a syllable, instead, it must necessarily be a monophthong. Semivowels do occur in Quechua but they behave as (approximant) consonants, restricted to the syllable boundaries, i.e. onset/coda, as stated by Weber (1989:450). This explains why the phenomenon of diphthongisation, as explained above, is not found in this language. And the same applies for nasalization.

Nasalization is the process whereby a vowel is pronounced as nasal due to the presence of a nasal consonant in its phonological environment. This process can be accounted for in terms of nasalization rules, which may vary from language to language. For instance, in English, vowels generally nasalized when they precede nasal sounds (Finegan, 2008), whereas in Land Dayak languages vowels typically undergo nasalization after a nasal consonant (Scott, 1964). As in the case of diphthongisation, nasalization also implies the introduction of certain specifications at the Phonological Level, which are language-specific and not requested by the higher levels of representation, affecting, in that way, language transparency. I have not found evidence to account for nasalization rules as part of the grammar of Quechua, though it may well occur due to purely articulatory aspects.

4.2.2 No sandhi rules

Sandhi rules refer to a variety of phonological processes that lead to the alteration of sounds and may occur either at morpheme boundaries, i.e. word-internal sandhi, or at word boundaries, i.e. word-external sandhi. The alteration of sounds due to sandhi rules includes processes such as fusion, assimilation and elision, which inevitably affect the ideal one-to-one mapping between units of meaning and units of form. Thus, sandhi phenomena are not expected in transparent languages.

In Quechua, we find processes leading to sound alteration, operating basically at morpheme boundaries, that is, in the combination between base, affixes and/or clitics. For instance, first person marking on verbs, which is conveyed by means of vowel lengthening, usually leads to lowering of high vowels, as is the case of miku: ‘I eat’, which is pronounced [miko:]. Lowering can also take place in more restricted contexts; for example, the high vowel of certain morphemes becomes /a/ when a certain suffix follows. This is the case of the reflexive -ku, which changes to -ka when followed by the pre-transitional directional suffix -mu, e.g. sha-ku-mu-n, come-refl-afar-3, ‘he comes (from afar)’ is pronounced [ʃakamun] [11]. The phenomenon of nasal assimilation, whereby a final nasal assimilates to the point of articulation of the following consonant, is also common at morpheme boundaries, e.g. tayta-n=paq, father-3.poss=pur, ‘for his father’ is pronounced [tajtampaϰ]. The existence of these and other phonological specifications that leads to sound alteration, operating mainly at morpheme boundaries, can be regarded as a non-transparent feature of Quechua.

4.2.3 No degemination

Degemination is a phenomenon whereby sequences of identical phonemes are articulated as single phonemes. This occurs, for example, in an English word such as horrible, where the coda of the first syllable /hɒr/ functions as the onset of the adjacent syllable /rib/, and so it would be represented at the Phonological Level as follows (Hengeveld and Mackenzie, 2008: 450):

The Output Component (articulator) will reduce the sequence of phonemes /-rr-/ into a single one /-r-/ by means of a process of degemination. In a transparent grammar, degemination is not expected to operate, especially, at morpheme or word boundaries, as it would obscure the one-to-one correspondence between a unit of meaning and a unit of form.

In Quechua, degemination at morpheme boundaries does occur, e.g. ichik ‘little’+ -kuna ‘pl’ = ichikuna ‘little ones’. This is true when an affix such as –kuna is at stake. However, the cliticization of post-positions such as –kama is not subjected to processes of degemination according to Weber (1989:470). Thus, as for what degemination concerns, Quechua can be said to bear certain degree of transparency.