A prosodic account of complex predicate acquisition in Mam: A Mayan language

Abstract

The Mayan language Mam uses complex predicates to express events. Complex predicates map multiple semantic elements onto a single word, and consequently have a blend of lexical and phrasal features. The chameleon-like nature of complex predicates provides a window on children’s ability to express phrasal combinations at the one-word stage of language development. The ubiquity of complex predicates in the adult language insures that children will produce complex predicates as some of their first words. The verb complex in Mam has obligatory inflections for aspect, person, and to a degree direction. The inflections vary in degree of attachment between syllable segments, affixes, and clitics. Inflections with vowels are phonologically free, while inflections without vowels attach as either syllable segments or affixes. The Mam verb complex requires the addition of a phrasal layer to prosodic models of lexical acquisition. The paper used this extended version of prosodic theory to make five predictions for the acquisition of the verb complex. The paper analyzes production data for three children between 2;0 and 2;8 acquiring the northern variety of Mam spoken in San Ildefonso Ixtahuacán, Guatemala. The children’s production data for both the intransitive and transitive verb complexes support all five predictions to some degree. The children produced prefixes more frequently on vowel-initial stems than on consonant-initial stems, and they produced imperative suffixes more frequently than prefixes on consonant-initial stems. The children exhibited developmental differences and produced phrasal contractions that the prosodic theory did not predict. The results underline the need to integrate prosody into models of morphosyntactic development, and highlight the significance of complex predicates for theories of language acquisition.

Keywords

Mam Maya child language acquisition complex predicate prosody morphosyntax

Introduction

Complex predicates map multiple semantic elements onto a single word. As Amberber et al. (2010) state ‘The enduring theoretical interest in complex predicates is undoubtedly due to the fact that in some aspects they pattern with prototypical words, whereas in other aspects they pattern with prototypical phrases’ (p. 3). The lexical aspect of complex predicates insures that their grammatical elements have the predictability of lexical features. The phrasal aspect of complex predicates stems from their combination of multiple semantic elements. The chameleon-like nature of complex predicates provides a window on children’s ability to express phrasal combinations at the one-word stage of language development. The ubiquity of complex predicates in the adult language insures that children will produce complex predicates as some of their first words.

Complex predicates vary across languages due to differences in their constituent elements and degree of inflection. The traditional linguistic distinctions between the attachments of phonemes within a word, the attachments of affixes and clitics to words, and the attachments between words in phrases do not capture the variety of attachments within complex predicates. Traditionally, affixes exhibit stronger attachment to words than clitics (Spencer & Luís, 2012). Affixes, but not clitics, are prone to allomorphic changes (Zwicky & Pullum, 1983, p. 505). Affixes, themselves, exhibit different degrees of lexical integration as seen in the contrast between the words dogs and trees. These words have the same plural allomorph [z], but the plural suffix produces a final consonant cluster on dogs, but not on trees. The contrast between these noun forms is irrelevant to the rule for plural formation but becomes apparent when observing the plurals that children produce. Children acquiring English produce inflections more frequently on words with final vowels than on words with final consonants (Bernhardt & Stemberger, 1998; Song et al., 2009).

This paper explores the role of morpheme attachment in the acquisition of complex predicates in the Mayan language Mam (ISO-639-3, mam). Verb complexes in Mam, like those in other Mayan languages, have obligatory inflections for aspect, agreement, and clause type (Kaufman, 1990). Agreement is marked with a combination of ergative and absolutive cross-reference markers and person enclitic. Mam incorporates directional modifiers in the verb complex that specify both spatial and aspectual features of the action (England, 1983, pp. 171–172). Examples of intransitive and transitive Mam verb complexes are shown in (1).

(1) Mam verb complexes (c.f. England, 1983, pp. 166–167)

a. Intransitive b. Transitive

ma chin b’éet=a ma chin tzaj t-tzyú-Ɂn=a

[prox abs1 walk=enc] [prox abs1 dir erg2-grab-dep=enc]

‘I walked’. ‘You grabbed me’.

I use England’s (2017) notation in this paper to distinguish between three forms of attachment. Spaces indicate attachments for phonological words, hyphens indicate affix attachments, while equal signs indicate clitic attachments. The three forms of attachment represent different degrees of cohesion within the verb complex. Affixes have the strongest attachments, while phonological words and clitics are more loosely attached. The verb complex functions syntactically as a verb within a clause; its inflections for aspect, person and to a large degree direction are obligatory. I added an acute accent to vowels with primary stress and square brackets in the morpheme gloss to indicate the verb complex.

Verb complexes in Mam vary in form depending on which of their component inflections have vowels (England, 1983, p. 40). None of the inflections that precede the verb stems in (2) have vowels. The primary stress remains on the verb stems in (2), as in (1). The dependent suffix is absent in (2b) because the verb complex lacks a directional. The enclitic is absent in (2b) because the subject is not first or second person. The initial inflections in (2) attach more strongly to the verb stem than the initial inflections in (1).

(2) Mam verb complexes with bound morphemes

a. Intransitive b. Transitive

n=0=b’éet=a n=0=t-tzyú

[ipfv=abs2=walk=enc] [ipfv=abs3=erg3-grab]

‘You are walking’. ‘S/he is grabbing it’.

Initial inflections with vowels result in phrasal verb complexes, as in (1), while initial inflections without vowels result in lexical verb complexes, as in (2). The distinction between inflections with and without vowels is orthogonal to the functions of the inflections. The aspect and absolutive person markers in (1) have vowels, while the aspect and absolutive person markers in (2) lack vowels. The differences between form and function in the case of the Mam verb complex make it possible to examine whether children acquire verb inflections as a single functional category, for example, aspect, or based on their forms.

Mam verb complexes also make it possible to assess the role of position relative to form and function in language acquisition. The directional modifier in (1b) precedes the verb stem. Directionals attach as suffixes on imperative verbs, as in (3). The directional suffix -x ‘away’ in (3a) is attached to the directional suffix -ka ‘down’. The enclitic attaches after the directional suffixes. Apart from the imperative contexts, directional suffixes appear on intransitive verbs of motion and other directionals, as in (3a).

(3) Directional suffixes on imperative verbs in Mam (England, 1983, p. 174, 176)

a. Intransitive b. Transitive

chi mok’ée-0-ka-x=a 0=0-tx’éeɁma-n-ka=ya

[abs5 crouch-imp-dir-dir=enc] [abs3=erg2-cut-imp.dep-dir=enc]

‘Crouch down (you all)!’ ‘Cut it down!’

In this paper, I use the prosodic approach to explore the effects of morpheme attachment on children’s productions of verb complexes in Mam. Prosodic approaches unify syllable structure with phrase structure, providing a comprehensive description of child language (Demuth, 2018; Kehoe, 2018; Shattuck-Hufnagel, 2020). Kehoe’s review of prosodic research on child language over the past three decades observes a large degree of variation between children’s initial word forms that reflects the variation in prosodic features in adult languages. Studies of children acquiring Germanic languages (e.g. Demuth & Fee, 1995; Fikkert, 1994) found that the children built prosodic structures in a bottom-up fashion, producing individual syllables before combining them into feet and eventually prosodic words. Alternatively, studies of children acquiring Romance languages (e.g. Correira, 2009; Lleó, 2006; Santos, 2006) found that children acquiring Portuguese and Spanish take a top-down approach and produce multisyllabic words before producing monosyllabic words.

Kehoe (2018) observes that some of the differences between reports reflect differences in whether the investigators included or excluded phrasal forms, for example, French au revoir ‘good-bye’ (p. 179). She does not comment on the prevalence of inflected forms in these studies. Evidence from the acquisition of the verb complex in Mam thus contributes key insights on children’s ability to construct highly inflected phrasal forms in support of the top-down model of prosodic development.

The next section of the paper introduces the prosodic approach to language acquisition and shows how I tailor the prosodic model to fit the Mam verb complex. Using these modifications, I derive five predictions for the acquisition of the Mam verb inflections, which I then test with production data from three children. I assess the success of the prosodic approach in the discussion section and point to some of its limitations. I conclude with a look at the broader significance of the Mam data for crosslinguistic research on the acquisition of complex predicates (c.f. Sarvasy, 2021a).

The prosodic approach to the acquisition of the Mam verb complex

Prosodic approaches use the prosodic hierarchy shown in (4) to describe children’s linguistic development (Archibald, 1995; Demuth, 1996, 2001, 2018; Fee, 1995; Fikkert, 1994; Gerken, 1996; Kehoe, 2018). Demuth (2014, p. 20) comments that

although many other factors (frequency, processing load, etc.) may also contribute to a child’s variable use of a consonant or morpheme, the phonological/prosodic environment is critical for predicting where this might be more likely to occur. Knowing about the prosodic structure of a language then facilitates making crosslinguistic predictions about how a particular form will be realized at early stages of phonological development.

(4) The Prosodic Hierarchy (after Selkirk, 1996, p. 190)

Utterance Utt

Intonational phrase IP

Phonological phrase PPh

Prosodic word PW

Foot Ft

Syllable σ

Mora μ

The prosodic hierarchy assumes that a prosodic word contains at least one foot, the foot contains at least one syllable, and the syllable contains at least one mora (Hayes, 1995; Lahiri, 2001; Selkirk, 1984). Kehoe (2018, p. 167) observes that the prosodic word is based on evidence from minimality constraints, phonotactic generalizations, and the application of phonological processes, the foot is the unit of stress assignment, and the syllable is based on evidence from phonotactic generalizations. Demuth (1996), Fee (1995), and Fikkert (1994) proposed a model of children’s early prosodic development, in which children initially produce words as single syllables before they develop a foot structure that allows them to put together two syllables or two mora to form a foot. Demuth labels children’s binary foot structure a minimal word. In accordance with the prosodic hierarchy, children’s minimal words are headed by a strong syllable.

Stress in Mam

Syllable strength is based on stress in Mam. Lexical stress in Mam is sensitive to both syllable weight and position. It falls on the long vowel in a word, or if the word does not have a long vowel, stress falls on the vowel preceding the last glottal stop in the word. Otherwise, stress falls on the vowel preceding the last consonant in the root (England, 1983, pp. 37–38). Accordingly, the primary stress falls on the long vowel of the verb root -béet ‘walk’ in (1a), and the vowel of the verb root -tzyú ‘grab’ preceding the glottal stop in (1b). The preceding inflections in (1a and b) are unstressed, evidence that they are part of a single word.

Demuth, Fee, and Fikkert predict that children in the minimal word stage of acquiring Mam will produce the verb complex as a single stressed syllable. Mam is historically related to K’iche’ and has a prosodic structure that is similar to the structure in K’iche’ in many respects (Pye, 2017). To the degree that Mam stresses the bimoraic syllable of the verb stem, a prosodic model predicts that children will initially reduce the verb complex to this stressed syllable. On this basis, I derive the first test of the prosodic approach shown in (5).

(5) Prediction I

Children in the minimal word stage produce the Mam verb complex as a single, stressed syllable, e.g. the syllable b’éet in (1a) and the syllable t-tzyú-Ɂn in (1b).

Syllable structure

One strength of the prosodic approach is that it unites developments in syllable structure with developments in phrase structure. Children acquiring English, for example, produce many words with CV and CVC shapes (Bernhardt & Stemberger, 1998; Vihman, 2018, p. 190). Such observations predict that children will simplify the stem syllable t-tzyú-Ɂn in (1b) by omitting the prefix t- and reducing the consonant cluster /tzy/ to a simple stop, for example, /t/ (Pye et al., 2017). Children could apply the same reduction to the syllable coda in (1b) by reducing the suffix -Ɂn to /Ɂ/ or /n/. Together, these processes combine to predict the children reduce the verb complex in (1b) to a form like /tú-Ɂ/.

Some prefixes in Mam have allomorphs that are conditioned by the initial segment of the stem they attach to. The first-person ergative prefix is n- before consonant-initial stems and w- before vowel-initial stems. The second- and third-person absolutive person markers are conditioned by the initial sound of the following stem. The tz’- allomorph occurs on vowel-initial stems, and the zero allomorph appears on consonant-initial stems. The tz- allomorph only appears on the stems -úul ‘arrive’ and -íky’ ‘pass’, while the k- allomorph occurs on stems in the potential mood (England, 1983, p. 56).

Verb complexes with vowel-initial verb stems are shown in (6). The absolutive person marker tz’- in (6a) prosodifies with the vowel-initial intransitive verb stem, and results in a syllable with a CVVC structure. The person marker w- prosodifies with the vowel-initial transitive verb stem, producing a CVVɁC syllable structure.

(6) Vowel-initial verb stems in Mam

a. Intransitive b. Transitive (England, 1983, p. 169)

n=tz’-éel ma 0=tzaj w-íi-Ɂn=a

[ipfv=abs3-leave] [prox abs3=dir erg1-carry-dep=enc]

’S/he is leaving’. ‘I brought it’.

If the acquisition of inflection was strictly a matter of function, attachment to vowel-initial or consonant-initial verb stems would not affect children’s morpheme productions. Children would simply omit all the inflections in the verb complex and produce just the verb roots -éel in (6a) and -íi in (6b). If the acquisition of inflection is sensitive to prosodic structure on the other hand, children would be more likely to produce inflections that prosodify with their minimal words.

Demuth (2014) suggests that pretonic function words and unstressed word-initial syllables are both unfooted, and therefore subject to omission by children. She attributes differences in children’s productions to the claim that word-internal syllables prosodify at the lexical level, whereas function words prosodify at the phrasal level. She does not address syllables that contain a prefix such as the C-VC syllables in the Mam verb complex. Selkirk (1996, p. 188) proposed enriching the prosodic hierarchy by adding the category ‘prosodic clitic’ to account for ‘a morphosyntactic word that is not itself a PWd’ [prosodic word, cp]. The prosodic clitic is not a prosodic word because it is unstressed. It is not joined as a foot to a prosodic word because it does not act as a typical syllable in a word.

I will draw upon Selkirk’s prosodic clitic types to analyze the Mam prefixes. Selkirk distinguishes between what she labels free clitics, internal clitics and affixal clitics (7). I analyze the Mam prefixes that attach to vowel-initial stems, as internal clitics (7b). Internal clitics prosodify with the prosodic word. In particular, the prefix may form part of the stressed syllable that heads the prosodic word. I analyze prefixes that attach to consonant-initial stems, as affixal clitics (7c). The affixal clitics are not prosodically integrated into the prosodic word to the same degree as internal clitics. I discuss examples of free clitics later.

(7) Prosodic clitics (after Selkirk, 1996, p. 188)

a. Free clitic ( CL ( lex )_PW )_PP

b. Internal clitic (( CL lex )_PW )_PP

c. Affixal clitic (( CL ( lex )_PW )_PW )_PP

One significant difference between the attachments of internal clitics and affixal clitics is that affixal clitics, but not internal clitics, will be subject to consonant cluster reduction by children. If consonant cluster reduction affects affix omission then children are expected to produce prefixes on vowel-initial verb stems like those in (6) more frequently than prefixes on consonant-initial verb stems like those in (1). This argument leads to the prediction in (8).

(8) Prediction II

Children will produce prefixes more frequently on vowel-initial stems than on consonant-initial stems.

Phonological and syntactic words

The discussion to this point has shown how the prosodic approach uses prosodic distinctions between attachment types to predict weak syllable omission within words, as well as affix omission. Together, these processes make precise predictions about children’s productions of verb stems in Mam, but they do not predict the production of the initial inflections in the verb complexes.

The verb complexes in (1) and (2) function as single words. The inflections for aspect, person and to a large degree direction are obligatory. The same absolutive person markers that attach as prefixes to intransitive verb stems (as in 6a) attach as prefixes to directionals inserted within the verb stem (as in 6b). Verb complexes project the arguments for subject and object. The Mam verb complex does not allow the extensive degree of insertion that the English auxiliary construction was playing allows, for example, was often not actually playing. Mam’s rule for stress assignment provides further evidence that the verb complexes are single words. The attachments between the inflections within the verb complex are stronger than the attachments between syntactic words.

The verb complex in (1) is not a single word at the level of phonology. The preceding inflections are not prosodic words by the standards of the prosodic hierarchy because they have short vowels and are not stressed. England (1983, p. 40) states that the aspect and person markers with vowels, as seen in (1), are phonologically free. Mam limits insertion within the verb complex to a small set of directionals (as in 6b) and adverbs. England (1983) notes that ‘If a transitive verb has a directional, the potential mode suffix is the intransitive form and it follows the directional stem’ (p. 173). The verb complex in (9) shows the adverb tzan ‘well’ inserted between the directional and the verb stem. This evidence demonstrates that the inflections that precede the verb stem have a weaker degree of attachment than affixes, but a stronger degree of attachment than syntactic words.

(9) Adverb insertion (England, 2017, p. 508)

n=0=xiɁ=tzan n-má-Ɂn=a t-ée=ma

[ipfv=abs3=dir=well erg3-say-dep=enc] erg3-dat=clf

‘and I said to him’

{AILLA MAM007R011.137-138}

Selkirk (1996, p. 196) proposed that free clitics (FC) attach to the phonological phrase rather than to the prosodic word (8a). FC are not attached to a foot, and are therefore not stressed. Selkirk (1996, p. 194) compares the free clitic in the English phrase can pile to the unstressed initial syllable in the word compile. I applied Selkirk’s distinctions between FC, internal clitics and affixal clitics to the analysis of the verb complex in (1b) as shown in (10).

(10) Prosodic Structure of the Mam verb complex in (1b)

((ma)_FC (chin)_FC (tzaj)_FC (t-(tzyú-Ɂn)_PW )_PW )_PP=a

Selkirk does not follow the traditional distinctions that linguists make between words, clitics and affixes, which explains the discrepancies between her prosodic divisions and England’s more traditional divisions between morphemes. I will adopt a hybrid notation that uses England’s traditional morpheme distinctions indicated by spaces, hyphens and equal signs, and Selkirk’s prosodic divisions indicated by parentheses for the examples in this paper. The prefix t- in (10) is an affixal clitic, while the suffix -Ɂn is an internal clitic. I label the prosodic divisions within the verb complexes but not within other words in the examples.

This analysis treats the inflections that precede the verb stem as FC that attach to the phonological phrase rather than the prosodic word. The analysis predicts that children would produce the FC later than the affixes that attach to the verb stem because children have to project a higher structural level of the prosodic hierarchy. I propose that the person enclitic = a attaches outside of the phonological phrase containing the rest of the verb complex for reasons that I discuss below. The Mam verb complex allows a direct test of the effects of prosodic cohesion realized by FC, internal clitics, and affixal clitics.

Directionals

As seen in the verb complexes in (1) and (3), the directionals both precede and follow the verb stem. The directionals are derived from a set of intransitive motion verbs. Most of the directionals express the direction that the agent takes in the course of the action, although some of the directionals express aspectual information. The directionals in verb complexes have short vowels in place of the long vowels that occur in the corresponding motion verbs (England, 1983, pp. 167–168). More radical phonological reductions occur in the directional suffixes that follow the verb stem and other directionals (169). England lists the motion verbs along with the directionals that precede and follow the verb stem, as shown in (11).

(11) Motion verbs and directional clitics and suffixes in Mam (England, 1983, pp. 167–168)

Motion Directionals Motion Directionals

Verbs Pre Post Verbs Pre Post

‘go’ xiɁ xi -x ‘go out’ eel el -al

‘come’ tzaaj tzaj -tz ‘go in’ ook ok -k

‘go down’ kub’ kub’ -ka ‘remain’ kyaj kyaj -kaj

‘arrive there’ poon pon -pon ‘finish’ b’aj b’aj -b’aj

‘arrive here’ uul ul ‘pass by’ iky’ iky’

‘go up’ jaaw jaw ‘return’ aaj aj

Only the motion verbs in (11), but not other intransitive verbs, take directional suffixes in non-imperative contexts (England, 1983, p. 169). Directionals within the verb complex also take these suffixes (168). The directional suffixes follow the imperative suffix in imperative verbs, as shown in (3) and repeated in (12). The person enclitic follows the directional suffixes.

(12) Directional suffixes on imperative verbs in Mam (England, 1983, p. 174, 176)

a. Intransitive b. Transitive

((chi) (mok’ée-0-ka-x))=a (0=(0-tx’éeɁma-n-ka))=ya

[abs5 crouch-imp-dir-dir=enc] [abs3=erg2-cut-imp.dep-dir=enc]

‘Crouch down (you all)!’ ‘Cut it down!’

The directionals that precede the verb stem differ in form and position from the directional suffixes that follow the verb stem. The directionals that precede the verb stem are FC. The directional suffixes can prosodify as either internal clitics or affixal clitics, depending on whether they form part of the stressed syllable in the verb stem. The directional suffixes in (12) are affixal clitics. The directional suffixes often occur at the end of the phonological phrase, which gives the directional suffixes additional prosodic salience. The prosodic analysis results in the prediction in (13).

(13) Prediction III

Children will produce directional suffixes more frequently than directionals that precede the verb stem.

Dependent and imperative suffixes on transitive verbs

Transitive verbs with directionals require a dependent suffix. Verbs in indicative contexts with directionals have the dependent suffix -Ɂn (1b), while imperative verbs with directionals have the suffix -n (12b). Transitive verbs in imperative contexts without directionals have the imperative suffix -m. Intransitive verbs do not have dependent or imperative suffixes.

The dependent and imperative suffixes attach to the verb stem before the directional suffixes and frequently prosodify with the stressed syllable in the verb root. I analyze them as internal clitics when they form part of the stressed syllable in the verb stem and as affixal clitics otherwise. The suffixes are part of the prosodic word that contains the verb stem even where they do not attach to the stressed syllable, as in (12b). Accordingly, the prosodic approach leads to the prediction in (14).

(14) Prediction IV (Intransitive verb)

Children will produce the dependent and imperative suffixes on transitive verbs more frequently than prefixes attached to consonant-initial transitive verb stems as affixal clitics.

The person enclitic

Mam lost its original second-person ergative and absolutive markers when it extended its third-person markers to second-person contexts. Mam innovated the use of a person enclitic to distinguish first- and second-person contexts from third-person contexts (England, 2017). It also distinguishes the first person, plural exclusive from the first person, plural inclusive (15). The person enclitic is used with both intransitive and transitive verb stems and combines with both the ergative and absolutive person markers. It is also used with the ergative person markers on nouns to cross-reference the possessor. The person enclitic has two allomorphs: -a following consonants and -ya after vowels (12).

(15) Enclitic use with person markers (England, 1983, p. 56)

Person Ergative Absolutive Enclitic

1 singular n-/w- chin =a/=ya

2 singular t- 0-/tz-/tz’-/k- =a/=ya

3 singular t- 0-/tz-/tz’-/k-

1 plural, exc q- qo =a/=ya

1 plural, inc q- qo

2 plural ky- chi =a/=ya

3 plural ky- chi

The person enclitic follows all other suffixes in the verb complex. The enclitic does not syllabify with the verb stem. The evidence from glide epenthesis is significant in this regard. Other vowel-initial suffixes in Mam undergo a process of vowel synthesis rather than epenthesis. The vowel in the passive suffix -eet, for example, synthesizes with the vowel in the verb -pajóo ‘choke’, yielding the passive verb stem -pajóot (England, 1983, p. 46). The absence of vowel synthesis for the person enclitic shows that it does not prosodify with the verb stem. It could be a free clitic and attached to the phonological phrase containing the verb complex, but other evidence suggests that this is not the case. For example, the transcriptions show that adult speakers omit the person enclitic when the verb is followed by a nominal classifier or adverb.

The evidence suggests that the person enclitic prosodifies above the phonological phrase that contains the other parts of the verb complex. It is not a prosodic word because it is unstressed, and it is distinct from Selkirk’s FC because it does not attach directly to the phonological phrase. The transcription evidence suggests the person enclitic is both prosodically and pragmatically weak. The person enclitic frequently appears on imperative verbs addressed to children, for example, néj-a, wait-enc, ‘Wait!’ I suggest renaming Selkirk’s type of FC ‘phrasal clitics’ because they attach to the phonological phrase and reserving the term ‘free clitic’ for cases like the person enclitic in Mam, which do not attach to the phonological phrase. I will continue using Selkirk’s terms in this paper to maintain consistency between Selkirk’s examples and the initial inflections in Mam.

Evidence from the acquisition of the person enclitic has significant implications for the prosodic analysis of the verb complex. Demuth (1996) proposes that children will initially produce minimal words with a trochaic stress pattern. Assuming that children treat the stressed syllable in the Mam verb complex as the head of a binary foot, this analysis predicts that they would reduce the verb complex in (1a) to a trochaic minimal word containing the last two syllables, for example, b’éet = a. However, if the person enclitic attaches above the phonological phrase as I suggest, children would reduce the complex to the stem b’éet. The prosodic analysis predicts the children produce the person enclitic infrequently even though they hear it frequently in the final position of imperatives. This argument leads to the prediction in (16).

(16) Prediction V

Children will produce the person enclitic less frequently than the directional suffixes.

This section of the paper described the prosody of the component morphemes in the Mam verb complex, and derived five predictions for their acquisition. The prosodic approach of Demuth, Fee and Fikkert predicts that features of the minimal word will dominate children’s early production of the verb complex. This prediction suggests that children will reduce the verb complex to its stressed syllable and any internal clitics that form part of the stressed syllable. Accordingly, children are predicted to produce prefixes on vowel-initial verb stems and suffixes on vowel-final verb stems. The directionals that precede the verb stem are FC that attach to the phonological phrase rather than the prosodic word containing the verb stem. Accordingly, children are predicted to produce the directional suffixes more frequently than the directionals that precede the verb stem. The enclitic attaches outside of the phonological phrase and therefore is predicted to be a late acquisition.

The following section of the paper describes the methods used for collecting and analyzing the acquisition data for Mam. In the succeeding sections, I use the children’s production data to test the five predictions made by the prosodic approach. I assess the success of the prosodic approach in the discussion section of the paper and point to phenomena that require further analysis. I conclude with a look at the broader significance of the Mam data for crosslinguistic research on the acquisition of complex predicates.

Method

The Mam acquisition data for each child are based on four separate recording sessions made in San Ildefonso Ixtahuacán, Guatemala, between 2005 and 2007. The recordings took place in and around the children’s homes. We recorded the children with an Olympus digital voice recorder, an Edirol R-1 digital audio recorder, and a Panasonic GS150 MiniDV camcorder. The recordings and transcriptions are available through the Acquisition of Mayan Languages Collection of Clifton Pye at the Archive of the Indigenous Languages of Latin America (AILLA) at the University of Texas, Austin.

The children interacted with other family members, the investigators, and random visitors during the recordings. The investigators lived in the same town as the children and interacted with them in Mam. The sessions included play with toys, natural objects, and picture books. The children were being raised in monolingual Mam households although their parents use Spanish outside the home. WEN and CRU are girls, while JOS is a boy. Table 1 shows the general measures for the children for each of the four sessions included in the analysis.

Table 1.

General measures for the Mam children.

Child	Ages	Duration	No. of utterances	IV tokens	TV tokens	Percent verbal
WEN	2;0.2	1:2:36	1021	97	264	35
	2;0.25	1:2.33	1153	81	243	28
	2;1.7	1:2.32	849	71	74	17
	2;1.21	1:2.32	1157	206	215	36
CRU	2;5.12	1:9:52	619	61	14	12
	2;5.26	31:38	187	21	15	19
	2;6.10	1:2:18	589	34	119	26
	2;7.1	42:05	643	61	20	13
JOS	2;6.17	1:3:01	607	67	47	19
	2;7	51:46	886	134	110	27
	2;7.14	1:2.32	720	104	85	26
	2;8.4	1:5.29	1238	213	192	33

IV: intransitive verb; TV: transitive verb.

The number of utterances produced by the children shows that WEN and JOS were more talkative than CRU. As a result, CRU produced fewer verbal utterances than WEN and JOS, and therefore fewer contexts for verb inflections. As it happens, she produced enough verbal utterances to evaluate the similarity between her verb morphology and that of the other two children. I note some individual differences between the children’s productions in the following sections.

Less than a third of the children’s utterances contained verb complexes. Mam does not have a copula, which reduces the number of verbal utterances that children and adults produce. The verbal utterances the children produced include many repetitions of such exclamations as ‘Look!’ and ‘Wait!’. The sparseness of the children’s verb productions requires that every effort be made to extract as much information as possible from every verbal utterance. The children show tremendous consistency in the forms of the verbs that they produced, both individually and across the children.

Prediction I: children’s minimal words

Prediction I states that children in the minimal word Stage produce the Mam verb complex as a single, stressed syllable (5). This section provides examples of the children’s verb production to illustrate the phenomenon, and then examines the frequency of the children’s minimal word productions. The children produce the verb complex most frequently as the verb root containing the stressed syllable.

I provide examples of WEN’s production of verb complexes in (17). The first line shows the child’s utterance, while the second line shows the equivalent adult utterance with its prosodic structure. In (17a), WEN produced a phrase that includes the recent past marker ma together with the stressed syllable of the verb phrase and the direct object. The motion verb -xíi ‘go’ is suppletive and has the third person singular form txíi ‘s/he/it goes’. WEN produced this verb as /chi/ in (17a). In addition, she reduced the noun n-yamád=a ‘my call’ to its stressed syllable /máa/. I did not analyze the children’s production of nouns due to space limitations. In (17b), WEN produced the stressed syllable of the verb phrase but omitted the ergative prefix t-. She reduced the initial consonant of the verb root /ky’/ to /k/ in this utterance.

(17) WEN’s production of consonant-initial verb roots (2;0.2; MW080107)

a. Intransitive b. Transitive

maa chi maa teɁ kee jha

((ma)_FC (txíi)_PW)_PP (w-éky’) (n-(yamád))=a (t-(ky’ée)_PW)_PW (jál)

[prox abs3_go] erg1-mine erg1-call=enc [erg3-not_want] nh

‘I made a call’. (Literally ‘A call of mine went’.) ‘She does not want it’.

I provide examples of CRU’s production of verb complexes in (18). In (18a), CRU produced the stressed syllable of the verb complex along with the person enclitic. She omitted the preceding first-person absolutive person marker chin. In (18b), CRU contracted the directional txi ‘away’ with the initial consonant of the verb root -xó and added the person enclitic. She omitted the aspect marker ma, the first-person ergative marker n- and the dependent suffix -Ɂn. CRU is more limited than WEN in her phrasal productions and may have compensated by contracting the directional txi ‘away’ with the initial consonant of the verb.

(18) CRU’s production of consonant-initial verb roots

(2;5.26; MC161005) (2;6.10; MC301005)

a. Intransitive b. Transitive

kuɁa. o chixha.

((chin)_FC (kúb’)_PW)_PP=a (o) ((ma)_FC (txi)_FC (n-(xó-Ɂn)_PW)_PW)_PP=a

[abs1 go_down=enc] o [prox abs3_dir erg1-throw-dep=enc]

‘I am going down’. ‘Oh, I already threw it away’.

I provide examples of JOS’s production of verb complexes in (19). In (19a), JOS produced the stressed syllable of the verb complex along with the potential suffix -il. He omitted the potential prefix k-. In (19b), JOS produced a verb complex in which the preceding directional bears a directional suffix. He produced the stressed syllable containing the verb root -má ‘take’ and the dependent suffix -Ɂn. He omitted the object prefix t- on the directional as well as the subject prefix t- on the verb stem. He also omitted the final person enclitic. He may have lengthened the vowel in the verb root to maintain syllable weight in order to compensate for the omission of the glottal stop in the dependent suffix.

(19) JOS’s production of consonant-initial verb roots (2;6.16; MJ021005)

a. Intransitive b. Transitive

chuxh jawal pe? koxh maan?

(chúx) (k-(jáaw-il)_PW)_PW (b’é)? ((ok)_FC (t-(kuɁ-x))_FC (t-(má-Ɂn)_PW)_PW)_PP=a?

quick [pot.abs3-go_up-pot] road [pot erg3-dir-dir erg2-take-dep=enc]

‘Quick, is it going up the road?’ ‘Will you take it away?’

These examples show that the children produced the stressed syllable in all of these examples, with the possible exception in (18b), in which CRU contracted the verb root with the preceding directional clitic. All three children produced some contracted forms, but CRU produced them more frequently than WEN and JOS. Her contractions preserve part of the stressed syllable but omit the dependent suffix. (A reviewer noted that this type of phrasal contraction is rare in children acquiring English.) The children simplified syllable onsets by omitting prefixes from the directional and the verb stem. With the exception of WEN’s utterance in (17a), the children omitted the initial aspect markers.

As a test of Prediction I, I counted the indicative verb forms that the children produced in their first sessions. I did not include the forms of the children’s imperative verbs because imperative verbs in adult speech generally lack preceding inflections. Including the children’s imperative verbs would skew the count in favor of Prediction I. The results are shown in Table 2. The production of the verb root predominates in WEN and CRU’s productions, whereas JOS produced transitive verbs with ergative prefixes and other inflections that are not included in the table. The children omitted the verb complex entirely in a few contexts.

Table 2.

Children’s indicative verb forms.

Child	Type	Omission	Root	Root + dir	abs + root	erg + root	asp + root
WEN	IV	3	19	1	2		1
WEN	TV	1	4			5	1
CRU	IV	1	48		5		13
CRU	TV		2			1
JOS	IV		11		10		10
JOS	TV	2					1

IV: intransitive verb; TV: transitive verb.

These results support Prediction I in that WEN and CRU limited their production of the verb complex to the stressed syllable containing the verb root. The variety of verb forms that JOS produced reflects the degree to which he had moved beyond the minimal word stage.

Prediction II: children’s production of vowel-initial and consonant-initial verb stems

Prediction II states that children will produce prefixes more frequently on vowel-initial stems than on consonant-initial stems (8). Here, I present data on the children’s production of verb complexes with vowel-initial verb roots such as -éel ‘leave’ or -íi ‘carry’. The vowel-initial stressed syllable allows children to produce prefixes more frequently. These productions offer the primary evidence that the children can access the morphosyntactic features for agreement when it coincides with the children’s prosodic development.

I provide examples of WEN’s production of verb complexes with vowel-initial verb stems in (20). In (20a), WEN produced the stressed syllable of the vowel-initial intransitive verb along with the third-person absolutive marker tz’-, which she produced as /t/. In (20b), WEN produced the stressed syllable of the vowel-initial transitive verb along with the first-person ergative markers w- and the dependent suffix -Ɂn. She omitted the preceding directional txi in this utterance even though she produced it, as a verb in (17a). She also omitted the person enclitic = a. These examples are further evidence of WEN’s limited ability to produce multiword utterances, which also limits her ability to produce directionals.

(20) WEN’s production of vowel-initial verb roots (2;0.2; MW080107)

a. Intransitive b. Transitive

tel wiin

((n=(lay)_PW)_PW (tz’-éel)_PW)_PP (jál) (i) ((txi)_FC (w-íi-n)_PW)_PP=a

[ipfv=cannot abs3-leave] nh pur [abs3_go erg1-carry-dep=enc]

‘It cannot leave’. ‘I am going to carry it’.

I provide examples of CRU’s production of verb complexes with vowel-initial verb stems in (21). In (21a), CRU produced the stressed syllable of the vowel-initial verb root -éel ‘leave’ with the initial consonant /t/ that I interpret as her production of the third-person allomorph tz’-. She omitted the preceding imperfective proclitic n = . Note that CRU also reduces the Spanish loan word botón ‘button’ to its stressed syllable as well. In (21b), CRU produces the stressed syllable of transitive verb -íil ‘see’ with the preceding aspect and ergative markers as well as the person enclitic.

(21) CRU’s production of vowel-initial verb roots)

(2;5.26; MC161005) (2;6.10; MC301005)

a. Intransitive b. Transitive

naa tel toni. man tila?

(nada) (n=(tz’-éel)_PW)_PW (ti) (n-(botón))=a ((ma)_FC (t-íil)_PW)_PP=a

no [ipfv=abs3-leave] dem erg1-button=enc [prox erg3-see=enc]

‘No, my button is coming off’. ‘Did you see it?’

I provide examples of JOS’s production of verb complexes with vowel-initial verb stems in (22). In (22a), JOS produced the stressed syllable of the intransitive verb -óok ‘enter’ together with the third-person ergative marker t- and the purpose adverb i ‘so that’. In (22b), JOS produced the stressed syllable of the vowel-initial transitive verb -áaj ‘want’ together with the second-person ergative prefix t-. He omitted the person enclitic in this utterance. These examples are further evidence of JOS’s ability to produce multiword utterances.

(22) JOS’s production of vowel-initial, monosyllabic roots (2;6.16; MJ021005)

a. Intransitive b. Transitive

i tok mahɁ. kal taj wey’ ki.

(i) (t-óok)_PW (máal). . . (tqáal) ((t-áaj)_PW)_PP=a (w-éeky’)=a (kyjakín)

so_that [erg3-enter] dem what [erg2-want=enc] erg1-with=enc so

‘So that that enters . . .’ ‘What do you want with me so?’

I analyzed the effect of the initial verb segment on the children’s production of preceding morphology by counting how often the children produced or omitted a preceding inflection on the vowel-initial and consonant-initial verb stems. I performed this analysis separately for the children’s intransitive (Table 3) and transitive verbs (Table 4). The results show that the children produced more preceding inflections on vowel-initial verbs than on consonant-initial verbs. The results are similar between sessions and across the children. The difference between vowel-initial and consonant-initial stems is less clear for the transitive verbs because the children produce fewer transitive verbs, but the results for the transitive verbs are consistent with the results for the intransitive verbs.

Table 3.

Children’s inflection production on vowel-initial and consonant-initial intransitive verb stems.

WEN	Age	Vowel-initial		Consonant-initial
Session	Age	Use	Omit	Use	Omit
MW080107	2;0.2	5	2	2	13
MW010207	2;0.25	4	6	1	16
MW130207	2;1.7	2		4	11
MW270207	2;1.21	4	2	4	33
	Totals	15	10	11	73
CRU	Age	Vowel-initial		Consonant-initial
Session	Age	Use	Omit	Use	Omit
MC021005	2;5.12	7	1	5	35
MC161005	2;5.26	3		5	7
MC301005	2;6.10	1		6	11
MC211105	2;7.1	1		2	11
	Totals	12	1	18	64
JOS	Age	Vowel-initial		Consonant-initial
Session	Age	Use	Omit	Use	Omit
MJ021005	2;6.17	11		27	13
MJ161005	2;7	13	1	41	26
MJ301005	2;7.14	9	2	50	27
MJ201105	2;8.4	17	2	60	46
	Totals	50	5	178	112
Totals		77	16	207	249

Table 4.

Children’s inflection production on vowel-initial and consonant-initial transitive verb stems.

WEN	Age	Vowel-initial		Consonant-initial
Session	Age	Use	Omit	Use	Omit
MW080107	2;0.2	4		4	11
MW010207	2;0.25	3			1
MW130207	2;1.7	2			8
MW270207	2;1.21	4		5	22
	Total	13	0	9	42
CRU	Age	Vowel-initial		Consonant-initial
Session		Use	Omit	Use	Omit
MC021005	2;5.12	3		1	2
MC161005	2;5.26				3
MC301005	2;6.10	1
MC211105	2;7.1			1	1
	Total	4	0	2	6
JOS	Age	Vowel-initial		Consonant-initial
Session		Use	Omit	Use	Omit
MJ021005	2;6.17	12		6	6
MJ161005	2;7	7		11	13
MJ301005	2;7.14	5	1	8	5
MJ201105	2;8.4	15	2	14	15
	Total	39	3	39	39
Totals		56	3	50	87

I tested the effect of the initial segment on the children’s production of preceding inflections by combining the number of uses and omissions for intransitive and transitive verbs for each child. A chi-square test showed a significant effect for each child (WEN Χ² = 51.26, p < .05; CRU Χ² = 33.11, p < .05; JOS Χ² = 36.67, p < .05). JOS produced enough tokens to allow separate tests for his intransitive and transitive verbs. The tests showed a significant effect for both verb types (intransitive verbs Χ² = 43.27, p < .05; transitive verbs Χ² = 56.67, p < .05). I conclude that the children’s production of the preceding inflections on vowel-initial and consonant-initial verb stems supports Prediction II of the prosodic approach.

Prediction III: directional position

Prediction III states that children will produce directional suffixes more frequently than directionals that precede the verb stem (13). This prediction is based on the difference between the preceding directionals realized as FC and the directional suffixes realized as internal and affixal clitics. The results support this prediction.

The children’s production of directionals was largely determined by their frequency of use and position. The children focused their production on three directionals that were frequent in conversation: xi ‘away’, tzaj ‘toward’ and kub’ ‘down’. The children produced directionals preceding the verb stem as well as the directional suffixes, but WEN and CRU produced directionals more frequently and accurately as suffixes. The children’s productions exhibited this pattern for both the intransitive and transitive verb complexes. The number of the children’s directional productions and their percentage use in obligatory contexts are shown in Table 5.

Table 5.

The number of tokens and percent use in obligatory contexts of children’s directionals.

Child	Age	Intransitive verb complex				Transitive verb complex
		Preceding		Suffix		Preceding		Suffix
		No.	Percent	No.	Percent	No.	Percent	No.	Percent
WEN	2;0	1	0	13	54	31	23	94	51
CRU	2;6	4	50	13	100	3	67	20	70
JOS	2;7	46	91	115	83	93	88	119	86

The data support a developmental interpretation in that the two youngest children favor the production of directional suffixes over directionals that precede the verb stem in accordance with Prediction III. CRU and JOS at 2;6 produced directionals as preceding forms and suffixes on transitive verbs equally as a percentage of obligatory contexts, but CRU produced more tokens of directional suffixes. All three children produced the directional suffixes at similar percentages of obligatory contexts on intransitive and transitive verbs. This similarity suggests that they treated the directional suffixes as a single inflection that applied to both intransitive and transitive verbs.

There is evidence of productivity in that the children combined the directional suffixes with different stems. The examples in (23) show the range of intransitive verb complex forms with directionals that WEN produced in a single recording session when she was 1;10. The production in (23a) contains just the verb root. It omits the aspect marker, the absolutive person marker, and the preceding directional. The examples in (23b) and (23c) show that WEN added the same directional to two different motion verbs -jáaw ‘go up’ and -kúb’ ‘go down’. In both cases, WEN applied the appropriate morphophonological changes to derive the combined forms -jáax and -kúx with some minor adjustments. WEN used both motion verbs with and without directional suffixes.

(23) WEN’s intransitive verb complex forms with directionals (MW081106; 1;10.2)

a. Intransitive Verb Root

tak.

= ((ma)_FC (tz’-el)_FC (tz’áq)_PW)_PP

[prox abs3-dir fall]

‘It fell’.

b. Motion Verb + Directional Suffix

axh

= ((ma)_FC (chin)_FC (jáa-x)_PW)_PP=a

[prox abs1 go.up-dir=enc]

‘I went up there’.

c. Verb Complex (Aspect + Person + Motion Verb + Directional Suffix)

ma kuɁxh

= ((ma)_FC (0=kú-x)_PW)_PP

[prox abs3-go.down-dir]

‘It went down there’.

Prediction IV: the dependent and imperative suffixes

Prediction IV states that children will produce the dependent and imperative suffixes on transitive verbs more frequently than prefixes attached to consonant-initial transitive verb stems, as affixal clitics (14). The results offer marginal support for this prediction.

I illustrate the children’s use of the dependent suffix using one remarkable exchange that WEN produced in a single recording session when she was 1;9.21 (MW271006). WEN produced a variety of transitive verb stems that illustrate the range of transitive verb stems that all three children produced. WEN’s utterances are shown in (24).

(24) WEN’s transitive verb complex forms with directionals (MW271006; 1;9.21)

a. Verb Root + Dependent Suffix + Person Enclitic

oona

= ((k-(x-el))_FC (n-(xó-Ɂn)_PW)_PW)_PP=a

[pot.abs3-dir-dir erg1-throw-dep=enc]

‘I will throw it’.

b. Absolutive + Preceding Directional + Verb Root + Dependent Suffix

texh oon

= ((ma)_FC (tz-el-x)_FC (n-(xó-Ɂn)_PW)_PW)_PP=a

prox abs3-dir-dir erg1-throw-dep=enc

‘I threw it’.

WEN’s productions in (24a) and (24b) show her production of transitive verbs in indicative contexts in which the verb complexes have preceding directionals. WEN omitted the preceding directional in (24a) but produces it in (24b). WEN produced the transitive verb stem with the dependent suffix in both of these cases. All three children produced examples like these, providing evidence that their production of the dependent suffix was independent of their production of a directional.

WEN did not only omit the dependent suffix. In several instances, she produced the dependent suffix in contexts that did not require its use. These contexts all involved the use of the verb -q’í ‘take’, as in (25). WEN’s production of this verb contrasts with her use of its near synonym -tzyú ‘grab’ as seen in (26). WEN’s production in (26) ends with the imperative suffix -m that is used when a directional is not used. Her production in (25) has the dependent-imperative suffix -n that is used when a directional is present. Productions like that in (26) show that WEN could produce the imperative suffix -m, so her use of the suffix -n in (25) appears to be an overgeneralization. It is notable that WEN did not produce the suffix -m when the suffix -n was required. She overgeneralized the dependent-imperative suffix but not the imperative suffix -m.

(25) Dependent suffix overgeneralization (MW080107)

Ɂin mam

= (q’í-m)_PW=a (mamá)

[take-imp=enc] mama

‘Take it mama!’

(26) Imperative suffix production (MW080107)

tum luki

= (tzyú-m)_PW=a (lúki)

[grab-imp=enc] this

‘Grab this!’

To test Prediction IV, I compared the children’s production of the dependent/imperative suffixes on transitive verbs (shown in Table 6) with their production of prefixes on consonant-initial transitive verbs (from Table 4). The results show that WEN and JOS produced the dependent/imperative suffixes in many of their obligatory contexts but were far from consistent in their use. The data in Table 6 include all of the children’s utterances with the verb -kí ‘look’. The children frequently produced various forms of this verb, and many without the dependent-imperative suffix. This verb accounts for the majority of CRU’s transitive verb tokens and explains her limited production of the dependent-imperative suffix shown in Table 6.

Table 6.

Children’s dependent/imperative suffix and prefix production on transitive verbs.

Child	Age	Dependent/imperative suffix		Prefix
Child	Age	Contexts	Percent	Contexts	Percent
WEN	2;0	431	69	51	18
CRU	2;6	46	17	8	25
JOS	2;7	310	61	78	50

These results offer marginal support for Prediction IV. The data for WEN show the strongest support for the prediction. CRU’s low rate of production for the imperative suffix was affected by her production of the verb -kí ‘look’ while JOS’s high rate of prefix production reflects his advanced linguistic development.

Prediction V: the person enclitic

Prediction V states that children will produce the person enclitic less frequently than the directional suffixes (16). The children’s production of imperative verbs provides an optimum context for observing their use of the person enclitic. WEN’s imperatives in (27a and b) both omit the person enclitic. Her production of the transitive verb complex (27b) omits the stressed syllable but includes the dependent-imperative suffix -n as well as the directional suffix -kub’.

(27) WEN’s production of imperative verb complexes (2;1.21; MW270207)

a. Intransitive b. Transitive

taj! nkuɁ!

(tzáaj)_PW=a ((q’óo-n)_PW-kub’)_PW=a

[come=enc] [put-imp-dir=enc]

‘Come!’ ‘Put it down!’

CRU’s utterances in (28a and b) also omit the person enclitics. Her production of the transitive verb complex is notable because it includes the possessed noun n-páy-a (erg1-bag-enc, ‘my bag’). CRU’s example shows that the children omitted the person enclitic on both nouns and verbs. Her production of the transitive verb complex in (28b) contrasts with WEN’s example in (27b) in that CRU’s utterance includes the vowel from the verb root as well as the directional suffix -x ‘away’. Her production omits the imperative suffix, the directional suffix -kub’ ‘down’, and the person enclitic.

(28) CRU’s production of imperative verb complexes (2;6; MC301005)

a. Intransitive b. Transitive

ne! oxh tu pay!

(néej)_PW=a ((q’óo-n)_PW-ku-x)_PW)_PP=a (t-új) (n-(páy))=a

[wait=enc] [put-dep-dir-dir=enc] erg3-in erg1-bag=enc

‘Wait!’ ‘Put it down in my bag!’

JOS’s example in (29a) shows that children as old as 2;8 still omit the person enclitic in simple imperatives. His utterance in (29b), on the other hand, includes the person enclitic as well as the transitive imperative suffix -m.

(29) JOS’s production of imperative verb complexes (2;8; MJ201105)

a. Intransitive b. Transitive

uxh! chuma!

(chúux)_PW=a (tzyú-m)_PW=a

[fast=enc] [grab-imp=enc]

‘Faster!’ ‘Grab it!’

Table 7 compares the results for the children’s production of the person enclitic with their production of the directional suffixes from Table 5. Their production of the person enclitic with intransitive verbs displays a developmental increase, while their production of the person enclitic with transitive verbs does not. CRU’s production of the person enclitic with transitive verbs differs the most from the other children. Her tendency toward verb contraction accounts for most of this difference. There is evidence that the children’s production of the person enclitic differed for the indicative and imperative verbs. WEN averaged around 5% usage of the person enclitic with her intransitive verb complexes and her imperative forms of the transitive verb complex. Her use of the person enclitic with indicative forms of the transitive verb complex was higher at 24%.

Table 7.

Number of tokens and percent production in obligatory contexts of the person enclitic and directional suffixes.

Child	Age	Intransitive verb complex				Transitive verb complex
		Enclitic		Directional		Enclitic		Directional
		No.	Percent	No.	Percent	No.	Percent	No.	Percent
WEN	2;0	3	4	13	54	12	9	94	51
CRU	2;6	4	19	13	100	11	55	20	70
JOS	2;7	23	32	115	83	15	14	119	86

The children’s production of the person enclitic was significantly less frequent than their production of the directional suffixes. I tested the difference between the children’s production of the person enclitic and the directional suffixes by combining the number of uses and omissions for intransitive and transitive verbs for each child. A chi-square test showed a significant effect for each child (WEN Χ² = 102, p < .05; CRU Χ² = 14.99, p < .05; JOS Χ² = 177, p < .05). JOS produced enough tokens to allow separate tests of the difference for his intransitive and transitive verbs. The tests showed a significant effect for both verb types (intransitive verbs Χ² = 52.86, p < .05; transitive verbs Χ² = 124, p < .05). The results support Prediction V.

Discussion

In this paper, I modified the prosodic approach to language acquisition proposed by Demuth, Fee, and Fikkert to suit the Mam verb complex. These researchers focused their studies on the development of what they termed the minimal word, which contains up to two mora or two syllables. I extended this model of prosodic development by adding Selkirk’s prosodic clitic types to account for the forms of inflections in the Mam verb complex. The Mam inflections exhibit four degrees of attachment. The preceding inflections with vowels were analyzed as FC. The inflections that attach to vowel-initial or vowel-final stems were analyzed as internal clitics, while the inflections that attach to consonant-initial or consonant-final stems were analyzed as affixal clitics. The person enclitic attaches above the phonological phrase that contains the verb complex. This prosodic account does not include affixes attached to FC.

The prosodic properties of the inflections are independent of their function. The absolutive person markers with vowels are FC. The person markers without vowels are realized as internal clitics or affixal clitics depending on whether they attach to vowel-initial or consonant-initial stems. The directionals are realized as FC before the verb stem and as affixal clitics after the verb stem. These features make it possible to assess the effects of differences in form and position on morpheme acquisition independently of function.

I derived five predictions for child language acquisition from the prosodic description of the verb complex that I tested with the children’s productions (30). The children’s production data for both the intransitive and transitive verb complexes support all five predictions to some degree. The tests offer the strongest support for the Predictions II and V. The tests for Predictions I, III and IV were supported by the data for the two youngest speakers. The oldest speaker, JOS, had progressed beyond the stage in which minimal words dominated his productions of the verb complex.

(30) Prosodic predictions for the acquisition of the verb complex

Prediction I: Children in the minimal word Stage produce the Mam verb complex as a single, stressed syllable.

Prediction II: Children will produce prefixes more frequently on vowel-initial stems than on consonant-initial stems.

Prediction III: Children will produce directional suffixes more frequently than directionals that precede the verb stem.

Prediction IV: Children will produce the dependent and imperative suffixes on transitive verbs more frequently than prefixes attached to consonant-initial transitive verbs as affixal clitics.

Prediction V: Children will produce the person enclitic less frequently than the directional suffixes.

The results underline the need to integrate prosody into models of morphosyntactic development. The Mam children were especially sensitive to the difference between inflections attached to vowel-initial stems, for example, C-VC, and inflections attached to consonant-initial stems, for example, C-CVC. Such data offer support for a prosodic distinction of the type that Selkirk makes between internal clitics and affixal clitics. The children’s omission of the preceding inflections with vowels support Selkirk’s category of FC. The prosodic hierarchy proposed by Selkirk and others must be amended to account for the prosody of the person enclitic in Mam. Selkirk does not discuss the case of FC with affixes.

Despite these successes, the prosodic approach that I proposed for Mam does not explain two elements of the children’s verb complexes. The first element is the children’s production of contracted forms of the verb complex. Contraction was frequent in CRU’s productions, but WEN and JOS produced contracted forms as well. There are cases in which contraction led to the omission of the imperative suffix but the production of the following directional suffix (28b). Other cases included the production of a preceding directional followed by the omission of an initial consonant in the verb stem (18b).

Children’s contractions have not drawn much attention in the acquisition literature (Bernhardt & Stemberger, 1998, pp. 476-483). A reviewer for this paper suggested a possible connection between the Mam examples of contraction, and a child’s production of the English word balloon as [bu:n]. The difference between the Mam contractions and the English example is that the Mam children are contracting preceding and following clitics with the verb stem, whereas the English example contracts segments within a single prosodic word.

Demuth (2014, p. 19) observes that one explanation for the early use of articles in Spanish in contrast with German is that the Spanish articles prosodify with the following noun, whereas German articles are separate prosodic words. Children acquiring Spanish contract the article with the noun and produce la muñeca ‘the doll’ as [ameca]. Such observations suggest that children’s contractions provide independent evidence about the prosodic structure of the adult language, and show that contraction processes vary across languages. It is possible that the frequency of contraction varies with the degree of clitic attachment (31). The Spanish evidence shows that children contract internal clitics, while the Mam evidence shows that children contract free and affixal clitics with verb stems. A full investigation of the children’s contracted forms is a project for the future.

(31) Predicted correlation between contraction frequency and prosodic type

Frequent Contraction Infrequent Contraction No Contraction

internal clitics > affixal clitics > free clitics > prosodic words

The second element that the prosodic account does not explain is the acquisition of morphosyntactic constraints. The children restricted the use of the directional suffixes appropriately to intransitive motion verbs, directionals and imperatives. They do not add directional suffixes to frequent intransitive verbs such as -tz’áq ‘fall’. In (32a), for example, WEN attempted to produce a potential form of the non-motion verb -tz’áq ‘fall’ with a preceding directional -el ‘leave’. In (32b), she produced the motion verb -kúb’ ‘down’ with the directional suffix -x ‘away’. Mam frequently uses intransitive motion verbs to express transitive events, as in (32b).

(32) WEN’s production of directionals with intransitive verbs (2;1.21; MW270207)

a. Non-motion verb b. Motion verb

ke peyɁjh. uxh ka pixh.

((k-el-x)_FC (tz’áq)_PW)_PP (jal) (n=(t-(kúb’-x)_PW)_PW)_PW (qa) (b’íxh))

[pot.abs3-dir-dir fall] nh [ipfv erg3-down-dir] us now

‘It’s going to fall’. ‘Let’s put it down now’.

Likewise, the children restrict the use of the dependent suffix to transitive verb complexes; they do not add the dependent suffix to intransitive verb complexes. More significantly, the children observe the constraints on the use of the ergative and absolutive person markers. They extend the ergative markers to intransitive verbs in the appropriate contexts (Pye et al., 2013). In sum, the children produced frequent cases of morpheme omission but few cases of overgeneralization. This result is explained best by assuming the children have full access to the grammar of the verb complex but are constrained by prosody from producing the entire phrasal verb complex (c.f. Demuth, 2014).

This paper only reports child data for the northern variety of Mam spoken in San Ildephonso Ixtahuacán. England (2017, p. 500) notes that Mam has the greatest degree of internal diversity of any Mayan language. Significant differences exist between the Northern, Central, Southern, and Western varieties of Mam. For example, stress falls on the penultimate syllable in Southern Mam, on the final syllable in Western Mam, and on the last heavy syllable in Northern Mam. Transitive verb roots in Southern Mam always drop the root vowel when the verb takes the dependent suffix -Ɂn. This never happens in Western Mam, and almost never in Northern Mam. To date, no documentation exists for the acquisition of other varieties of Mam.

The Mam results presented in this paper contribute new details to the comparative work on Mayan language acquisition (Pye, 2017). All Mayan languages have verb complexes, but the elements in the verb complex vary from language to language and have different degrees of cohesion. The absolutive markers precede the verb root in K’iche’, Mam, and Q’anjob’al, while they follow the verb root in Ch’ol, Tzeltal, and Yucatec. K’iche’ has directional particles that follow the verb rather than the directional suffixes seen in Mam. The K’iche’ directional particles do not prosidify with the verb at the same level as the directional suffixes in Mam. Children acquiring K’iche’ do not contract the directional particles with the verb in the way that children acquiring Mam contract the directional suffixes (Pye & Pfeiler, 2019). The person enclitic is unique to the Mamean branch of Mayan languages. Despite such differences, children acquiring other Mayan languages produce person markers with vowel-initial verbs more frequently than with consonant-initial verbs and produce verb suffixes more consistently than verb prefixes (Brown et al., 2013; Mateo Pedro, 2015, 2023; Pye, 1983, 2017).

Conclusion

Ken Hale (1998, p. 192) observed that

While a major goal of linguistic science is to define universal grammar, i.e. to determine what is constant and invariant in the grammars of all natural languages, attainment of that goal is severely hampered, some would say impossible, in the absence of linguistic diversity. If English were the only language on the face of the earth, we could not know literally hundreds of things which are permitted, even predicted, by universal grammar and accidentally missing in English, or any other single language.

Complex predicates are prime examples of features that are largely missing in English, and unexamined by theories of language acquisition (Pye, 2021; Sarvasy, 2021a). Complex predicates such as the Mam verb complex constitute a new realm of investigation in acquisition research and offer fundamental insights into the role of prosody in determining the form of children’s phrasal utterances. Complex predicates demonstrate that we still have some way to go to meet Bowerman’s (2011, p. 616) call to ‘guard against parochial explanations of language acquisition’.

Thanks to a special issue on the acquisition of complex predicates in First Language (vol. 41, no. 4, 2021), we know a little more about children’s productions of complex predicates in diverse languages. Complex predicates in the Algonquian language Northern East Cree have preverbs (separated by a space in the orthography) that express a variety of grammatical and lexical meanings (Brittain & Rose, 2021). The Papuan language Qaqet has a complex predicate that compounds prepositions with verb roots without a space in its orthography (Hellwig, 2021). The Papuan language Nungon has a complex predicate based on a serial verb construction in which the verbs are separated by spaces in the orthography (Sarvasy, 2021b). Although the orthographies for each of these languages use spaces to indicate lexical cohesion, the authors do not discuss the strength of attachment the spaces represent.

Given the examples of the children’s production of verb complexes in Mam, it is surprising to learn that children acquiring Qaqet and Nungon generally produce the entire complex predicate rather than just the verb stem, as was the case for children acquiring Mam. The examples for Northern East Cree include a few cases of preverb omissions, but the majority of the child’s utterances contain the preverb and verb. These data suggest that there can be major differences in how much of a complex predicate 2-year-olds produce from the partial examples seen in Mam and Northern East Cree to the complete forms children produce in Qaqet and Nungon.

Such differences reveal a need to better understand the role that prosody plays in promoting morphosyntactic development across the full range of attachment types, including those in complex predicates (Demuth, 2014). Prosody remains one of the least documented features of child and adult languages (Gussenhoven & Chen, 2020). This is especially the case for languages with complex predicates. Emergentist (Behrens, 2021; Bybee, 2001) and generative (Grinstead, 2021; Hyams, 2007) theories of language acquisition do not account for the development of complex predicates. This paper demonstrates why the Mam verb complex provides a critical test of any acquisition theory (Pye, 2022).

Orthography

All Mayan words are shown in the practical orthography developed by the Proyecto Lingüístico Francisco Marroquín (Kaufman, 1976). The orthographic symbols have their standard IPA values except: <b’>=/ɓ/,<tz>=/ts/,<tx>=/tʂ/,<tx’>=/tʂ’/,<ch>=/tʃ/,<ch’>=/tʃ’/,<y>=/j/,<j>=/x/,<jh>=/x^h/,<xh>=/ʃ/,<x>=/ʂ/. I use an acute accent on vowels, for example, á, to indicate primary stress.

Footnotes

Acknowledgements

My grateful thanks to the editors of this issue, Lourdes de León and Pedro Mateo Pedro, as well as to the editor of the journal and the three anonymous reviewers for their careful reading of previous versions of this article. Donald Stengel helped with the statistical analyses. The article benefited greatly from their suggestions; the remaining defects are mine. My deepest gratitude is due to families of the Mam children without whom the research would not be possible. I owe a special thanks to the Mam investigators: Ana Elizabeth López Ramirez, Juana Isabel López Morales, Sheny Ortíz García, and Luis Hernandez López Ramirez. Pedro Mateo Pedro helped the Mam investigators record and transcribe the children’s speech. I dedicate this article to the memory of the late Professor Nora England, whose research provides the foundation for all future work on Mam and whose support of native Mayan linguists enriched the field of Mayan linguistics.

Author contributions

Clifton Pye: Conceptualization; Data curation; Formal analysis; Funding acquisition; Investigation; Methodology; Project administration; Resources; Software; Supervision; Validation; Writing – original draft; Writing – review & editing.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Data collection for Mam was funded by grants from the National Science Foundation (BCS-0613120 and BCS-0515120) and the University of Kansas.

ORCID iD

Clifton Pye

References

Amberber

Baker

Harvey

(2010). Introduction. In Amberber

Baker

Harvey

(Eds.), Complex predicates: Cross-linguistic perspectives on event structure (pp. 1–12). Cambridge University Press.

Archibald

(1995). The acquisition of stress. In Archibald

(Ed.), Phonological acquisition and phonological theory (pp. 81–109). Erlbaum.

Behrens

(2021). Constructivist approaches to first language acquisition. Journal of Child Language, 48(5), 959–983.

Bernhardt

B. H.

Stemberger

J. P.

(1998). Handbook of phonological development: From the perspective of constraint-based nonlinear phonology. Academic Press.

Bowerman

(2011). Linguistic typology and first language acquisition. In Song

J. J.

(Ed.), The Oxford handbook of linguistic typology (pp. 591–617). Oxford University Press.

Brittain

Rose

(2021). The development of preverbs in Northern East Cree: A longitudinal case study. First Language, 41(4), 376–405.

Brown

Pfeiler

de León

Pye

(2013). The acquisition of agreement in four Mayan languages. In Bavin

Stoll

(eds.), The acquisition of ergativity (pp. 271–305). John Benjamins.

Bybee

(2001). Phonology and language use. Cambridge University Press.

Correira

(2009). The Acquisition of Primary Word Stress in European Portuguese [Unpublished PhD dissertation]. University of Lisbon.

10.

Demuth

(1996). The prosodic structure of early words. In Morgan

J. L.

Demuth

(Eds.), Signal to syntax: Bootstrapping from speech to grammar in early acquisition (pp. 171–184). Erlbaum.

11.

Demuth

(2001). Prosodic constraints on morphological development. In Weissenborn

Höhle

(Eds.), Approaches to bootstrapping: Phonological, syntactic and neurophysiological aspects of early language acquisition (Language acquisition and language disorders series, Vol. 24, pp. 3–21). John Benjamins.

12.

Demuth

(2014). Prosodic licensing and the development of phonological and morphological representations. In Ferris-Trimble

A. W.

Barlow

J. A.

(Eds.), Perspectives on phonological theory and development: In honor of Daniel A. Dinnsen (pp. 11–24). John Benjamins.

13.

Demuth

(2018). Understanding the development of prosodic words. In Esteve-Gibert

Prieto

(Eds.), The development of prosody in first language acquisition (pp. 207–224). John Benjamins.

14.

Demuth

Fee

(1995). Minimal words in early phonological development [Unpublished manuscript]. Brown University and Dalhousie University.

15.

England

N. C.

(1983). A grammar of Mam, a Mayan Language. University of Texas Press.

16.

England

N. C.

(2017). Mam. In Aissen

England

N. C.

Zavala Maldonado

(Eds.), The Mayan languages (pp. 500–532). Routledge.

17.

Fee

E. J.

(1995). Segments and syllables in early language acquisition. In Archibald

(Ed.), Phonological acquisition and phonological theory (pp. 43–61). Erlbaum.

18.

Fikkert

(1994). On the acquisition of prosodic structure [Doctoral dissertation]. University of Leiden.

19.

Gerken

(1996). Prosodic structure in young children’s language production. Language, 72, 682–712.

20.

Grinstead

(2021). Interface delay. Journal of Child Language, 48, 888–906.

21.

Gussenhoven

Chen

(Eds.). (2020). The Oxford handbook of language prosody. Oxford University Press.

22.

Hale

(1998). On endangered languages and the importance of linguistic diversity. In Grenoble

L. A.

Whaley

L. J.

(Eds.), Endangered languages: Language loss and community response (pp. 192–216). Cambridge University Press.

23.

Hayes

(1995). Metrical stress theory: Principles and case studies. University of Chicago Press.

24.

Hellwig

(2021). Initial observations on complex predicates in Qaqet children’s language. First Language, 41(4), 406–429.

25.

Hyams

(2007). Aspectual effects on interpretation in early grammar. Language Acquisition, 14, 231–268.

26.

Kaufman

(1976). El proyecto de alfabetos y ortografías para escribir las lenguas Mayances. Proyecto Lingüístico Francisco Marroquín and Ministerio de Educación.

27.

Kaufman

(1990). Algunos rasgos estructurales de los idiomas Mayances con referencia especial al K’iche’. In England

N. C.

Elliot

S. R.

(Eds.), Lecturas sobre la lingüística Maya (pp. 59–114). Cirma.

28.

Kehoe

(2018). Prosodic phonology in acquisition: A focus on children’s early words. In Esteve-Gibert

Prieto

(Eds.), The development of prosody in first language acquisition (pp. 166–184). John Benjamins.

29.

Lahiri

(2001). Metrical patterns. Language Typology and Language Universals, 13, 47–67.

30.

Lleó

(2006). The acquisition of prosodic word structures in Spanish by monolingual and Spanish-German bilingual children. Language and Speech, 49, 205–229. https://doi.org/10.1177/00238309060490020401

31.

Mateo Pedro

. (2015). The acquisition of inflection in Q’anjob’al Mayan. John Benjamins.

32.

Mateo Pedro

. (2023). Adquisición del complejo verbal en Chuj. In Buenrostro

(Ed.), Estudios de Lengua y Cultura Chuj (pp. 109–142). Universidad Autónoma de México.

33.

Pye

(1983). Mayan telegraphese: Intonational determinants of inflectional development in Quiche Mayan. Language, 59, 583–604.

34.

Pye

(2017). The comparative method of language acquisition research. University of Chicago Press.

35.

Pye

(2021). Documenting the acquisition of indigenous languages. Journal of Child Language, 48, 454–479. https://doi.org/10.1017/S0305000920000318

36.

Pye

. (2022). Exploring language diversity: Commentary on Kidd and Garcia (2022) First Language, 42(6), 799–803. https://doi.org/10.1177/01427237221084505

37.

Pye

Pfeiler

Mateo Pedro

(2013). The acquisition of extended ergativity in Mam, Q’anjob’al and Yucatec. In Bavin

Stoll

(Eds.), The acquisition of ergativity (pp. 307–335). John Benjamins.

38.

Pye

Pfeiler

Mateo Pedro

Stengel

(2017). Analysis of variation in Mayan child phonologies. Lingua, 198, 38–52.

39.

Pye

Pfeiler

(2019). The acquisition of directionals in two Mayan languages. Frontiers in Psychology, 10, Article 2442. https://doi.org/10.3389/fpsyg.2019.02442

40.

Santos

(2006). Word stress acquisition: A comparison between Brazilian Portuguese and Dutch. In Bamman

Magnitskaia

Zaller

(Eds.), Proceedings of the 30th annual Boston university conference on language development (pp. 525–536). Cascadilla Press.

41.

Sarvasy

H. S.

(2021a). On the acquisition of complex predicates: Introduction to the special issue. First Language, 41(4), 369–375.

42.

Sarvasy

H. S.

(2021b). Acquisition of multi-verb predicates in Nungon. First Language, 41(4), 478–503.

43.

Selkirk

(1984). Phonology and syntax: The relation between sound and structure. MIT Press.

44.

Selkirk

(1996). The prosodic structure of function words. In Morgan

J. L.

Demuth

(Eds.), Signal to syntax: Bootstrapping from speech to grammar in early acquisition (pp. 187–213). Erlbaum.

45.

Shattuck-Hufnagel

(2020). The role of phrase-level prosody in speech production planning. In Gussenhoven

Chen

(Eds.), The Oxford handbook of language prosody (pp. 522–538). Oxford University Press.

46.

Song

J. Y.

Sundara

Demuth

(2009). Phonological constraints on children’s production of English third person singular –s. Journal of Speech, Language, and Hearing Research, 52, 623–642.

47.

Spencer

Luís

A. R.

(2012). Clitics: An Introduction. Cambridge University Press.

48.

Vihman

M. M.

(2018). The development of prosodic structure: A usage-based approach. In Esteve-Gibert

Prieto

(Eds.), The development of prosody in first language acquisition (pp. 186–206). John Benjamins.

49.

Zwicky

Pullum

(1983). Cliticization vs. inflection: English n’t. Language, 59, 502–513.