2.1 Ilokano

Verbs in Ilokano are marked with voice, aspectual, and mood distinctions by means of different sets of affixes (Table 10.2). For actor voice, the affixes may either be ⟨um⟩ ‘inc’, mang- ‘dist’, or ag- ‘dur’. Undergoer voices are marked with the suffix -en¹² for patient, -an for locative, and i- for circumstancial. As for aspect, perfective is marked by the infix ⟨in⟩, and imperfective is typically marked by reduplicating the first CVC¹³ sequence of the stem. For the irrealis mood, Ilokano shows the optional use of the enclitic =(n)to, which is a variant of the adverb into that indicates future time.

These grammatical specifications on the verb are marked by combining the verbal affixes. To illustrate, the verb stem gatang ‘buy’ marked with ⟨um⟩ for actor voice (inchoative), in combination with the CVC reduplication for realis imperfective, yields the form g⟨um⟩at~gatang ‘⟨av.inc⟩ipfv~buy’. As for marking realis perfective, the aspectual infix ⟨in⟩ comes first before the voice infix ⟨um⟩, and this ordering of the verbal affixes in Ilokano has led to the syncopation of the vowel u in ⟨um⟩, and the subsequent assimilation of n in ⟨in⟩, leading to the form ⟨im⟩⟨m⟩. Thus, marking the same verb gatang ‘buy’ with actor voice, realis perfective yields the form g⟨im⟩⟨m⟩atang ‘⟨pfv⟩⟨av.inc⟩buy’

For distributive and durative verbs, marking aspectual distinctions does not reflect the same level of agglutination as inchoative verbs. In particular, the affixes used to mark realis perfective are portmanteau forms that combine the infix ⟨n⟩ (a reduction of ⟨in⟩) and the voice prefixes mang- for distributive and ag- for durative. This leads to the perfective forms nang- and nag- respectively. Realis imperfective and irrealis forms are more transparent, reflecting the expected combination of the voice and aspectual affixes. To illustrate these derivations, takaw ‘steal’ is derived in the actor voice distributive form as mang-takaw ‘av.dist.ntrl-steal’, nang-takaw ‘av.dist.pfv-steal’, mang-tak~takaw ‘av.dist-ipfv~steal’, and mang-takaw=to ‘av.dist-steal=irr. Surat ‘write’ is derived in the actor voice durative form as ag-surat ‘av.dur.ntrl-write’, nag-surat ‘av.dur.pfv-write’, ag-sur~surat ‘av.dur-ipfv~write’, and ag-surat=to ‘av.dur-write=irr’ (Rubino 2000: lxvii).

The forms mang- and ag- that mark actor voice distributive and durative are historically derived from a combination of the actor voice affix ⟨m⟩ (a reduction of ⟨um⟩) with the prefixes pang- and pag-. These latter prefixes carry the basic distributive and durative senses, and at present are also used to nominalize verb forms in Ilokano. These prefixes, moreover, are reflexes of Proto Malayo-Polynesian (PMP) *paN- and *paR- respectively, and the resulting portmanteau forms *maN- and *maR- are also reconstructed for PMP (Wolff 1973:72–74). The realis neutral form ag- in Ilokano, shows a further reduction of PMP *maR- to its current form ag-. The Ilokano verbal morphology is summarized in Table 10.2, with sample verbs to illustrate the various derivations discussed above.

2.2 Ibatan

Verbs in Ibatan are marked with the same distinctions as those discussed for Ilokano, but by different sets of affixes (Tables 10.4 and 10.5). Given the genetic relationship between the two languages, a number of affixes are identical in the two languages, namely the undergoer voice affixes -en¹⁴ ‘pv’, -an ‘lv’, and i- ‘cv’, as well as the actor voice distributive prefix maN-.¹⁵ The actor voice infix ⟨om⟩ in Ibatan is also phonologically similar to Ilokano ⟨um⟩, where o is pronounced as a high, back, rounded vowel, but it is represented orthographically with the vowel o. Ibatan also shows the use of the future adverb anchi as the enclitic =(a)nchi to optionally mark irrealis, parallel to the development of Ilokano into.

Ibatan differs from Ilokano in terms of the ordering of the aspectual and voice affixes. Where Ilokano reflects the sequence ⟨im⟩ ‘pfv’ + ⟨m⟩ ‘av’, Ibatan show the reverse order, that is, ⟨om⟩ ‘av’ + ⟨in⟩ ‘pfv’. This sequence is actually a retention of the ancestral system reconstructed for PMP (Ross 2002), and the current ordering observed in Ilokano constitutes an innovation shared among many Northern Luzon languages (Reid 1992).

What makes Ibatan unique, not only in comparison to Ilokano but also to its sister Batanic languages, is its two distinct but parallel paradigms of verbal affixes, where the use of a particular set typically depends on the etymology of the stem. This is observed in the paradigms for actor voice durative and realis imperfective. For marking durative verbs, Ibatan reflects two sets of prefixes, namely pay- (along with may- ‘av.dur.ntrl’ and nay- ‘av.dur.pfv’) and pag- (along with mag- ‘av.dur.ntrl’ and nag- ‘av.dur.pfv’). For marking realis imperfective, Ibatan shows different reduplication patterns, namely CV(y)/CVCV and CVC. Native Ibatan stems are marked with the paradigms pay- for ‘dur’ and CV(y) or CVCV for ‘ipfv’ (Table 10.4). As an example, the native Ibatan verb disna ‘sit’ occurs as may-disna for ‘av.dur.ntrl-sit’ and may-di~disna for ‘av.dur-ipfv~sit’. In contrast, loanwords, typically of Ilokano origin (but also stems from other source languages (SL), such as Filipino, English, and Spanish), are generally marked with pag- for ‘dur’ and CVC for ‘ipfv’ (Table 10.5). To illustrate, the Ilokano loanword kalap ‘fish’ is derived as mag-kalap for ‘av.dur.ntrl-fish’, and mag-kal~kalap for ‘av.dur-ipfv~fish’. The co-existence of these parallel paradigms in Ibatan is clearly an outcome of contact-induced change, where non-native stems are marked with non-native morphology. To further illustrate these parallel paradigms, (2a) and (2b) show the prefixes nay- and nag- marking native abang ‘(ride on a) rowboat’ and non-native lampitaw ‘(ride on a) motorized boat’ respectively.

(2) a. Native actor voice durative prefix nay- (Maree 2007:174)

Nayabang si adi a nangay do Calayan.

Nay-abang	si	adi	a	nangay	do	Calayan
dur-rowboat.ivb	det	younger.sibling	lk	went	det	Calayan

‘Younger sibling rode on a rowboat going to Calayan.’

b. Non-native actor voice durative prefix nag- (Maree 2007:174)

Naglampitaw si adi a nangay do Calayan.

Nag-lampitaw	si	adi	a	nangay	do	Calayan
dur-motor.boat.ilo	det	younger.sibling	lk	went	det	Calayan

‘Younger sibling rode on a motorized boat going to Calayan.’

The two sets of durative prefixes in Ibatan can be traced from two sources, both descended from PMP *paR-. The paradigm consisting of the forms pay- ‘dur’, may- ‘av.dur.ntrl’, and nay- ‘av.dur.pfv’ are directly inherited, as evidenced by the final consonant y, which is the regular reflex of PMP *R in the Batanic languages. The non-native paradigm consisting of the counterpart forms pag-, mag-, and nag- respectively is argued to be transferred from Ilokano, albeit with subsequent adaptation into the Ibatan system. Not only do the forms reflect g as the reflex of PMP *R, a feature of Ilokano,¹⁶ but the distribution of the prefixes with mostly Ilokano stems clearly points to Ilokano as the source of this paradigm (see Sections 3 and 4).

This non-native durative paradigm has become regularized in Ibatan, and has come to apply generally to loanwords, including those from English, Filipino, and Spanish (Table 10.3). Its usage and distribution are discussed in detail in Section 3.

As mentioned, these parallel durative paradigms are a unique feature in Ibatan, which is not observed in other Batanic languages such as Ivatan, a closely related language spoken on Batan Island, Batanes. Both native vidi ‘return’ and Spanish eroplano ‘(ride an) airplane’ take the native verbal prefix nay- (3).

(3) Ivatan: native may- with non-native stem

Nay-eroplano si Maria ta nayvidi du Basco. (elicited)

Nay-eroplano	si	Maria	ta	nay-vidi	du	Basco
dur-airplane.spa	det	Maria	because	dur-return.ivv	det	Basco

‘Maria took the airplane because she returned to Basco.’

3.1 Unexpected Formations

The first category among the small set of unexpected formations involves hybrid forms, or combinations of native and non-native material. Type 1 involves non-native morphology used with native stems (14 of 1436 stems, or merely 0.97 %) and Type 2 involves native morphology used with non-native stems (62 of 1436 stems, or 4.32 %). Table 10.7 gives some examples.

Evidently, the development of the non-native durative paradigm in Ibatan has arisen mainly through indirect transfer, that is, via the transfer of complex loanwords (Seifart 2015), as evidenced by the significant number of Ilokano stems that occur with the paradigm. The existence of hybrid formations, however, suggests other mechanisms that must have operated in driving this particular contact-induced change.

Type 1 hybrid formations constitute only a very small fraction of the overall distribution (only 0.97 % of all instances of durative formations indicated in Maree et al. 2012). Such kinds of formations raise an important question about how non-native morphology comes to be extended to native stems. As for Type 2 hybrid formations, while they occur more frequently than Type 1 forms, they still constitute a very small portion of the overall distribution (4.32 %). These two types of hybrid formations, along with other unexpected distribution, although very few in number, point to further complexity in Ibatan in terms of diversity of structures, as discussed below.

In deriving the basic durative meaning, loanwords occur with the non-native paradigm as expected, but in more complex formations that also involve other affixes, the native morphology is used. Table 10.8 gives some examples, where bosel ‘(develop) buds’, kamoras ‘(become sick with) measles’, darop ‘attack’, and tiro ‘shoot’ are all loanwords that are marked with the non-native mag- for the basic durative form but take the native paradigm may- when combined with other native affixes such as the distributive cha- and the reciprocal sin- along with reduplication to mark additional meanings of the verb.

Such cases suggest how morphology, even in agglutinative languages that have relatively transparent compositionality, such as Ibatan, encodes meanings on the basis of patterns of combination, irrespective of the discrete functions of the component elements (cf. Word and Paradigm approach by Hay and Baayen 2005, Ackerman et al. 2009, etc.). That is, more complex derivations in Ibatan appear as combinations involving native morphology, and these apply even for loanwords that are known to take non-native morphology in basic derivations. The sentences below illustrate this further. The Ilokano verb labang ‘dappled’ in (5a) and Spanish tiro ‘shoot’ in (6a) occur with mag-/nag- in the basic durative form, but (5b) shows may-cha-laba~labang ‘have irregular patches’ involving the native prefix may- in combination with cha- and CVCV reduplication to further derive the distributive meaning, and (6b) shows may-sin-ti~tiro ‘throw at each other’, again involving the native may- with the affix sin- and CV reduplication to derive the reciprocal meaning to the verb.

(5) a. Non-native mag- with non-native stem (Complex loanword)

Maglabang kodit kwaya, ta nadoplagan. (Maree et al. 2012: labang)

Mag-labang	kodit	kw=aya	ta	nadoplagan
dur-dappled.ilo	skin	1sg.gen=ref	because	scalded.

‘My skin becomes dappled because it was scalded.’

b. Native may-cha-RDP- with non-native stem (Type 2 hybrid formation)

Maychalabalabangayaw basket kwaya. (Maree et al. 2012: labang)

May-cha-laba~labang=aya=w	basket	kw=aya
dur-dist-rdp~dappled.ilo=ref=nom	basket	1sg.gen=ref

‘My basket has irregular patches of color.’

(6) a. Non-native mag- with non-native stem (Complex loanword)

Nagtiro so amang so pirpiroka. (Maree et al. 2012: tiro)

Nag-tiro	si	amang	so	pirpiroka
dur-shoot.spa	det	father	det	pirpiroka.bird

‘Father shot the pirpiroka bird.’

b. Native may-sin-RDP- with non-native stem (Type 2 hybrid formation)

Maysintitiro saw mangalkem so bwa. (Maree et al. 2012: tiro)

May-sin-ti~tiro	sa=aw	mangalkem	so	bwa
dur-rec-.spa rdp~throw	3pl.nom=ref	old.men	det	betel.nut

‘The old men threw betel nuts at each other.’

To illustrate further, Table 10.9 presents various derivations involving the durative paradigms found in Maree et al. (2012). The diversity of structures that can co-occur with the non-native durative prefixes is evidently limited compared to those that combine with the native may-. Such restricted distribution of the non-native paradigm indicates that it is not yet fully parallel with its native counterpart, especially with structures involving more complex morphological combinations that encode further semantic specifications on the verb.

There are also a few cases where both native and non-native durative prefixes can be used with the same verb, but appear to encode divergent meanings. An example is the Spanish word kwarto ‘room’, where mag-kwarto in (7a) means ‘make a room’, encoding dynamicity, while nay-kwarto in (7b) means ‘have a room’, encoding a stative sense.

(7) a. Non-native mag- with non-native stem (Complex loanword)

Magkwarto ka so rakoh. (Maree et al. 2012: kwarto)

Mag-kwarto	ka	so	rakoh
dur-room.spa	2sg.nom	det	big

‘Make a big room.’

b. Native nay- with non-native stem (Type 2 hybrid formation)

Naykwarto so anem bahay ko, ki dedekey. (Maree et al. 2012: kwarto)

Nay-kwarto	so	anem	bahay	ko	ki	de~dekey
dur-room.spa	det	six	house	1sg.gen	but	rdp~small

‘My house has six rooms, but they are small.’

Another example is in expressing direction/goal. The sentences in (8a) and (8b) involve the native Batanic word songet ‘forested area’. Songet also happens to be a place name in Babuyan Claro, and when derived to mean ‘to go to Songet’, it takes the non-native prefix mag- in combination with the directional pa-, as shown in (8a). In contrast, when referring to its general sense as ‘forested area’, the stem takes the native prefix may-pa-, as shown in (8b).²⁰

(8) a. Non-native mag-pa- with a proper noun (Type 1 hybrid formation?)

Magpa-Songet dana sa. (elicited)

Mag-pa-Songet	dana	sa
dur-dir-Songet.ivb	already	3pl.nom

‘They are already going to Songet.’

b. Native may-pa- with native stem (Native formation)

Maypasonget si anang mabekas. (elicited)

May-pa-songet	si	anang	mabekas
dur-dir-forested.area.ivb	det	mother	morning

‘Mother is going to the forested area in the morning.’

Ibatan also has instances of doublets, where a particular form is actually descended from two different sources. An example is the verb boya ‘to see, to meet, to watch’, where the Batanic languages and Ilokano share cognate forms. Ivatan vuya, Itbayaten vooya, and Ibatan boya²¹ are all cognates carrying the meaning ‘to see, to meet’. The Ibatan stem takes may-, as illustrated in (9a). The semantics of the word has also been expanded to include the meaning ‘to watch’, but in this particular sense, the form takes the non-native prefix mag-, as seen in (9b). This particular meaning of the form has been transferred from Ilokano, where the Ilokano word buya²² means ‘to watch’.²³ It is only the difference in meaning and the use of the non-native prefix that indicates that mag-boya is a complex loanword instead of a Type 1 hybrid formation.

(9) a. Native may- with native stem (Native formation)

Mayboya tanchi andelak. (elicited)

May-boya	ta=anchi	andelak
dur-meet.ivb	1pl=fut	tomorrow

‘Let’s meet tomorrow.’

b. Non-native mag- with non-native stem (Complex loanword)

Magboya kami so sine do Sabado. (elicited)

Mag-boya	kami	so	sine	do	Sabado
dur-watch.ilo	1pl	det	movie	det	Saturday

‘We will watch a movie on Saturday.’

This also relates to near-homophonous pairs of words that have arisen out of contact, where native Ibatan terms have come to share near-similar forms with Ilokano loanwords (only differing in terms of stress placement). Despite the similarity, however, the meanings and etymologies are kept distinct not only by maintaining the difference in the placement of stress, but also by the use of native and non-native prefixes, as illustrated in Table 10.10. The forms babang, barot, sagot, and talon occur with both native and non-native morphology, keeping the meanings and etymologies separate.

The cases described above clearly illustrate how the distribution of the durative paradigms in Ibatan, while relatively straightforward in the majority of cases (including doublets and near-homophonous terms that have different etymologies), can still be unpredictable for a small set of stems that constitute hybrid formations. As a final point, there are also instances where both the native and non-native durative prefixes appear to be used interchangeably (Table 10.11). It is not certain whether these are instances of stable variation in Ibatan, or if these constitute change in progress, where particular groups of speakers may tend to prefer the use of one particular paradigm over the other.

Thus, while the non-native durative paradigm has not yet been fully integrated into the morphological system of Ibatan given its limited distribution, not just in terms of the stems it occurs with but also the kinds of other structures it can combine with, it has added to the morphological complexity of Ibatan through contact-induced change. That is, Ibatan exhibits diversity of structures that are not seen in either Ilokano or its sister Batanic languages (see Section 2). This clearly runs in contrast with the usual claim in the literature that language contact results in a reduction of morphological complexity, and/or convergence between the languages in contact (cf. Gumperz and Wilson 1971, Matras and Sakel 2007, Gardani et al. 2015, etc.), which is often explained as a “by-product of the trend to syncretise the inventory of constructions across the languages in a bilingual’s repertoire” (Matras 2015:54). The case of Ibatan demonstrates that equal emphasis should be put on the nature and kinds of complexity that may arise in contact-induced change (cf. Bakker et al. 2011, Meakins et al. 2019, etc.).

3.2 Ongoing Cross-Linguistic Influence

So far, we have seen the general distribution and usage of the parallel durative paradigms in Ibatan, informed by data from Maree et al. (2012). These patterns constitute apparent contact-induced change that has become more or less stable in Ibatan. Synchronically, however, further variation in the usage of the paradigms can be observed among individual speakers.

Van Coetsem (1988, 2000) argues that individual speaker-based psycholinguistic mechanisms are linked to particular contact outcomes. His framework centers on the psycholinguistic notion of language dominance, which underpins the individual’s agentivity in bi-/multilingual speech. Language dominance has to do with the person’s relative proficiency in the different languages in their repertoire, where the dominant language is oftentimes the language they are most proficient in, typically their first language. However, it must be noted that dominance is not static and can vary across a person’s lifetime. Therefore, a person’s dominance may shift to their second language, and this is dependent on factors beyond language proficiency, such as exposure, frequency of use, and domain/context of use, among many others (cf. Silva-Corvalán and Treffers-Daller 2016, Treffers-Daller 2019, etc.). Therefore, contact effects vary as a person becomes more dominant in the recipient language (RL).

Particular patterns of language dominance determine the application of what van Coetsem (1988, 2000) describes as borrowing transfer in RL agentivity and imposition transfer in SL agentivity. An individual tends to work within the resources of their dominant language. Thus, when dominant in the RL, they use RL resources but may borrow components, typically vocabulary, from their non-dominant SL (RL agentivity). In contrast, a person who is dominant in the SL has a tendency to impose SL materials, such as phonology and grammar, when they use their non-dominant RL (SL agentivity). In terms of contact-induced outcomes, therefore, borrowing transfer results in largely lexical borrowings, which are sporadic, while imposition transfer tends to result to a “catastrophic modification” of aspects of the RL by means of systematic structural innovations (van Coetsem 1988:25).

Taking this framework to understand synchronic cross-linguistic influence among Ibatan speakers, the variant use of the durative paradigms appears to correlate with language dominance. As presented in Section 1, there are three general groups of Ibatan speakers, namely Ibatan-dominant early bilinguals, Ilokano-dominant early bilinguals, and Ilokano-dominant late bilinguals, and they exhibit variation in their knowledge and use of the durative paradigms, based on a preliminary corpus of Ibatan speech collected during the author’s fieldwork, and supplemented by interviews with Ibatan speakers.

Ibatan-dominant early bilinguals exhibit the general pattern of the durative paradigms described in the previous section. A number of these speakers, in fact, show good awareness of internal structures and etymology, where they identify stems that occur with mag- as non-native, typically from Ilokano, and those that occur with may- as native stems, which they describe as “pure Ibatan.” This indicates that they have good knowledge of both Ibatan and Ilokano, and they clearly maintain the distinction between the two languages by means of the associated morphological structures.

In a similar vein, Ilokano-dominant early bilinguals (or those who have learned Ibatan and Ilokano in their childhood but prefer Ilokano as their everyday language) also appear to maintain the boundaries of the two languages. In Babuyan Claro, these speakers are known for code-switching between the two languages (where Ilokano is the matrix language), described by locals as Ibakano, a blend of Ibatan and Ilokano. Despite their relative dominance in Ilokano, however, they still follow the expected use of the durative paradigms, even in situations where they switch between Ibatan and Ilokano in an utterance, as illustrated in (10). Here, the Ibatan verb may-tay~tagadan ‘(remain) slack’, reflecting the expected use of the durative prefix, is maintained alongside a by and large Ilokano utterance.

In contrast, Ilokano-dominant late bilinguals, who have learned Ibatan in adulthood when they migrated to Babuyan Claro, tend to show structural imposition in their use of Ibatan. In terms of morphology, these speakers exhibit increased usage of the non-native durative paradigm, even with native stems that are expected to occur with the native paradigm. This is illustrated in sentences (11) and (12).

In (11), the Ilokano-dominant late bilingual speaker used the non-native prefix pag- for the native stem chichwas ‘search’ instead of the expected native prefix pay-. In other instances, the same speaker used the expected may- for native stems, as seen in (12). The variant use of the durative paradigms by Ilokano-dominant late bilinguals, illustrated in (11), are regarded by Ibatan-dominant speakers as errors, and have come to be a marker that sets apart this group of speakers. It is however important to highlight the temporary nature of these impositions. That is, as proficiency or dominance in the RL increases, these impositions tend to lessen in the speech of Ilokano-dominant late bilinguals.²⁴

(11) Non-native pag- with native stem (Type 1 hybrid formation)

Gallego (2019): IVB1–20180930_08

Pati iyaw no chitowa aywanaw ki nachipagchichwas.

Pati	iyaw	no	chito=a	aywan=aw	ki	nachi-pag-chichwas
also	dei	det	dog=lk	pet=ref	inv	soc-dur-search.ivb

‘Even the pet dog searched (with him).’

(12) Native may-RDP- with native stem yonot (Native formation)

Myan saw mayyoyonot kan yaw no chitwaw.

Myan	sa=aw	may-yo~yonot	kan	yaw	no	chito=aw
ext	3pl=ref	dur-rdp~go.along.ivb	and	dei	det	dog=ref

‘There they are, going along, including the dog.’

As van Coetsem (2000) argues, language dominance and speaker agentivity do play important roles in explaining individual patterns of cross-linguistic influence and outcomes of contact-induced change. However, this model needs to be further tested and refined. In particular, the notion of language dominance needs to be operationalized more carefully. As seen in this section, language dominance is gradient, and contact outcomes may vary even among non-dominant speakers. That is, certain Ilokano-dominant speakers of Ibatan (i.e. late bilinguals) tend to exhibit structural imposition as predicted by van Coetsem’s SL agentivity, whereas others do not (i.e. Ilokano-dominant early bilinguals). Measuring language dominance in a way that captures such differences would allow us to better understand contact outcomes.²⁵

4.1 Layers of Contact-Induced Change in Ibatan

A context-based framework in analyzing contact-induced outcomes is proposed by Thomason and Kaufman (1988), which centers on the sociolinguistic context of the multilingual community (that is, the intensity and type of contact situation, which result in either language maintenance or shift). In situations of language maintenance, involving “borrowing interference”, the cline goes from light, moderate, to heavy contact, and in situations of language shift, involving “substratum interference” or “interference through shift”, the cline relates to the degree of interference from the source language, which depends on the size of the shifting group and the level of bilingualism of the community. Where the specific contact situation of the community is placed along the cline would determine the particular contact-induced outcomes, namely the transfer of non-basic vocabulary, or the transfer of more structured materials such as phonological, morphological, syntactic, and lexico-semantic features.

Thomason and Kaufman’s (1988) scale that focuses on widespread, community-level contact outcomes relates to the central concepts in van Coetsem’s (1988, 2000) speaker-based framework. That is, in situations of language maintenance, change is primarily seen in the lexicon, and this relates to the mechanisms involved in RL agentivity. In situations of shift, restructuring in the RL can happen via imposition of phonological and grammatical features from the SL, which is akin to the mechanisms governing SL agentivity.

Accounting for contact-induced language change involves linking the individual and the community, and understanding the transition from innovations to widespread change. Van Coetsem (2000) and Thomason and Kaufman (1988)’s models for language contact both put the psycholinguistic and sociolinguistic contexts of the multilingual individual and community at the heart of their frameworks. It then follows that communities with an extremely dynamic socio-political and linguistic landscape such as Babuyan Claro would reflect layers of change that are linked to changes in the patterns of multilingualism of the individual and the community. These phases in the history of Babuyan Claro are summarized in (1) and repeated in (13) below.

The first Ibatans [RL agentivity]. The first group who permanently settled Babuyan Claro in 1869 originally came from Batanes but had been relocated to the Ilokano-speaking islands of Calayan and Camiguin (Maree 2005). It can be assumed that while they were there, they had considerable interaction with Ilokano speakers, but to what degree they learned Ilokano is uncertain. At this stage, it can be argued that loanwords, including complex ones, were introduced into Ibatan, but were fully adapted not just in terms of morphological structure (as Type 2 hybrid formations) but also in terms of phonology²⁸ given the likely individual-level dominance of Ibatan across these Ibatan-speaking first families.

The daya and laod networks [RL agentivity]. As more groups from both Batanic- and Ilokano-speaking backgrounds came to Babuyan Claro, the population on the island slowly grew. In the initial years of the community, ethnographic evidence shows that ethnolinguistic lines were kept more or less separate (Maree 1982), and this can be seen in the emergence of distinct social networks clustered in the geographic regions of daya ‘east’ and laod ‘west’ coinciding with the use of Ibatan and Ilokano respectively. However, the harsh environmental conditions on Babuyan Claro meant that the inhabitants relied on social contact across these networks. Interaction with Ilokano-speaking networks (laod) most likely facilitated the continued transfer of loanwords into Ibatan, which were then fully adapted into the language, under the assumption that the Ibatan-speaking networks maintained their dominance in Ibatan since the arrival of their ancestors on the island. These fully adapted loanwords, which are older, widespread, and more socially integrated (cf. Poplack, et. al. 1988:72), are hence indicative of community-level dominance in Ibatan at this stage.

Ibatans with increased proficiency in Ilokano [SL agentivity]. The daya and laod networks largely correlate with speakers’ language ideologies and use. While the setting in the early years of Babuyan Claro fostered a type of egalitarian multilingualism, where both Ibatan and Ilokano co-existed on a more or less equal footing, the rise in the status of Ilokano in the wider region, and consequently in Babuyan Claro, had profound effects on the patterns of multilingualism on the island around the 1970s. In addition to a significant portion of the population tracing their ancestry to Ilokano (and so maintaining Ilokano as their first language), there were more domains in which Ilokano was used to the exclusion of Ibatan, consequently threatening vitality. As a result, a number of Ibatan families have shifted to Ilokano as their everyday language. The Babuyan Claro community, including Ibatan-dominant speakers, certainly had increased exposure to and proficiency in Ilokano during this period. This either meant a shift in language dominance for some speakers, thereby becoming Ilokano-dominant, or a shift to (near-)symmetrical/balanced bilingualism for others, wherein they have (near-)equal dominance in both languages.

We can assume that this change in the nature of bilingualism drove a different kind of lexical transfer from that of the early stages of the community. That is, loanwords kept their SL morphology instead of being fully adapted into the grammar of Ibatan, driven by the increased proficiency of the speakers in Ilokano. At this stage, increased dominance in Ilokano may have entailed SL agentivity, and the maintained use of Ilokano morphology in Ibatan is indicative of imposition transfer. Moreover, the speakers’ comparable proficiencies in the two languages, including a degree of awareness of morphological structures, must have facilitated the development of the adapted form mag- from the original Ilokano form ag-. That is, the speakers have analogized the Ilokano form ag- on the basis of the native counterpart may-. Since Ilokano ag- forms a paradigmatic relationship with the prefixes nag- and pag-, it is not difficult to analogize the form to be parallel with the native paradigm may-, nay-, and pay-, thus leading to the current form mag-.

Younger generations of Ibatan-dominant speakers [RL agentivity?]. Further socio-historical changes in the Babuyan Claro community led to the renewed vitality of Ibatan from the 1980s. Ibatan has now regained its function as the main language in Babuyan Claro, with Ilokano as the second language of the community and the lingua franca of the wider region. Younger generations of Ibatan speakers maintain their dominance in Ibatan, but keep considerable interaction with Ilokano speakers. This maintained social contact across the networks therefore allows for the mechanisms and processes that drive contact-induced change in Ibatan to persist.

Ilokano-dominant late bilingual speakers [SL agentivity]. Given the function of Ibatan as the main language of Babuyan Claro, Ilokano immigrants are learning Ibatan as their second language. As discussed in Section 3.2, the ongoing imposition transfer in the speech of Ilokano-dominant late bilingual speakers constitute the synchronic layer of contact-induced features we see in Ibatan. These features reflect a great deal of variation not only across individuals, but also within individual speakers. Synchronically, since such imposition transfer correlates with the speaker’s (changing) language dominance, such can be transient and tend to be lost as the speaker’s proficiency in Ibatan increases. These Ilokano immigrants constitute a small portion of the population, and their use of Ibatan tends to be dependent on the social networks they form in the community. That is, Ilokanos who form close ties with the daya network of mostly Ibatan-dominant speakers tend to learn Ibatan quickly, whereas those who are more affiliated with the laod network of Ilokano-dominant speakers tend to have lesser proficiency in Ibatan.

Ibatans with increased proficiency in Filipino [RL agentivity]. At present, the patterns of multilingualism in the Babuyan Claro community are shifting again, this time driven by the rising influence of Filipino. This is clearly reflected in how the younger generations of Ibatan speakers have become more proficient in Filipino. As Babuyan Claro became further integrated into the larger nation state, the Ibatans have more exposure to Filipino, not only as medium of instruction in schools, but also as the main language of print, broadcast, and social media. To compare, the older generations of Ibatans still have limited proficiency in Filipino, but a number of younger Ibatan-dominant speakers report preference towards using Filipino as their second language over Ilokano. As it happens, Filipino has forms identical to the non-native durative prefixes, and this must be reinforcing the current use and distribution of the paradigm in Ibatan. As expected, complex loanwords from Filipino, including nonce borrowings, occur with the non-native durative prefixes. Loanwords of foreign origin (typically English) are also introduced into Ibatan indirectly through Filipino, which have already been adapted with Filipino verbal morphology.

Partly through loanwords (and nonce borrowings), and via speakers with increased proficiency in Filipino, the non-native paradigm has come to be extended to include loanwords from other SL s. While it can be analyzed as a repurposing of the paradigm to accommodate non-Ilokano loanwords, the more accurate way to describe the influence of Filipino in this respect is reinforcing the function of the paradigm, given that Filipino shares exactly the same set of durative prefixes.

Linking phases and mechanisms. Changes in the socio-historical landscape of Babuyan Claro are clearly linked to changes in the nature of multilingualism on the island, which are then reflected as layers of contact-induced features in Ibatan. However, these apparent stages in the history of the community are by no means discrete. Even at present, different mechanisms of agentivity apply among different groups of speakers, yielding different outcomes: (1) for Ibatan-dominant speakers, RL agentivity resulting in lexical transfer, but keeping the boundary between Ibatan and Ilokano distinct through the expected use of the parallel paradigms; (2) for Ilokano-dominant early bilinguals, code-switching behavior with Ilokano as the matrix language; and (3) for Ilokano-dominant late bilinguals, SL agentivity resulting in the imposition of Ilokano structures in Ibatan speech, reflected in the variant use of structures.

This dynamic nature of multilingualism can also be seen in items that have been transferred multiple times into Ibatan. One clear example is the complex loanword may-tarabako, from Spanish trabajo ‘work’.²⁹ The degree of adaptation that applied on the loanword indicates that this is an early loan in the language. More recently, Maree et al. (2012) note that the younger generation now prefers to use the form mag-trabaho. This form, aside from the use of the non-native prefix mag-, exhibits a closer phonetic shape to the original Spanish word.³⁰ Such differences in how the word has been adapted into Ibatan shows agentivity at play; speakers with greater dominance in Ibatan are more likely to adapt a form to their dominant Ibatan phonological structure, while those with greater proficiency in the SL³¹ tend to show less modification.

One thing that is apparent in the history of the Babuyan Claro community is that the speakers have continually kept Ilokano and Ibatan distinct. This etymological consciousness shows that the speakers are more or less aware of the differences between the languages in their repertoire, reflected most strikingly in how parallel morphological structures are used and maintained in Ibatan (not just in terms of derivational morphology discussed in this paper, but also in the domain of inflection, such as the aspectual marking described in Section 2.2). It also indicates how this must have been a conscious process for the Ibatans, as a way of flagging their mixed identity (Gallego 2020:107). This essentially relates to the phenomenon of morphological compartmentalization described by Matras (2015:48) for cases where (inflectional) morphology “is replicated along with lexical word forms from another language in situations in which speakers embrace and flag a bilingual identity.”

Ultimately, knowledge of SL structures is an essential part of how morphology is transferred and regularized in Ibatan. The large number of complex loanwords in the language suggests that the durative paradigm has been transferred indirectly. Seifart (2015) proposes that indirect transfer requires particular patterns in corpus frequencies involving pairs of complex and simplex loanwords, under the assumption that the speakers are analyzing non-native morphological structures on the basis of such patterns, but this does not seem to be the central mechanism for Ibatan. Given what we know of the nature of multilingualism in Babuyan Claro, the speakers are already clearly knowledgeable in Ilokano, and so, this must have played a crucial role in the development of non-native morphology in Ibatan. That is, good knowledge of Ilokano, along with the fact that the two languages are genetically related and typologically similar, allows for easier morphological analysis on the part of the speaker, which can then promote morphological productivity for non-native structures. Furthermore, this process entails a certain level of consciousness in the part of the speakers (cf. Thomason 2008, 2015), and that maintaining the distinction between Ibatan and Ilokano was an important motivation in this process.

At the same time, however, there are a few cases where the boundary between the two languages seems to be less clear. Hybrid formations are a clear indication of this. Some of these forms can be considered early loanwords into Ibatan (Type 2), and are indicative of speakers’ shifting knowledge of what counts as loanwords, while others reflect impositions of SL structures (Type 1). While these forms comprise only a small subset of the distribution (5.29 %), it is necessary to understand in more detail how such formations came to be stable in Ibatan, but this remains an open question.³²

4.2 Further Questions

It cannot be denied that outcomes of language contact and change exist within the socio-historical context of the community that use the languages. With context-based frameworks for language contact such as van Coetsem (1988, 2000) and Thomason and Kaufman (1988), contact outcomes are linked to mechanisms that govern language use. In this particular paper, understanding the structural consequences of the transfer of complex loanwords is not only approached as an outcome of language contact, but also through attested tokens of speaker-driven cross-linguistic influence. This case study thus allows us to test the various assumptions proposed in these context-based frameworks based on contemporary patterns of language use.

From what we have seen in Ibatan, there are evident gaps in the frameworks that need to be addressed. For instance, there is still much to know about the linguistic outcomes of symmetrical or balanced bilingualism, where the speakers have (near-)equal dominance in the two languages in their repertoire. Van Coetsem (2000) proposes the neutralization of transfer types, where outcomes linked to both imposition and borrowing transfer may be equally possible. While the literature on bilingualism argues that this is a rare type of bilingualism (cf. Grosjean 1985, etc.), it is still important to consider it within models of language contact to better understand its linguistic consequences. This issue is also deeply connected to the need for a nuanced operationalization of language dominance that goes beyond mere measurement of relative proficiency (cf. Silva-Corvalán and Treffers-Daller 2016, Treffers-Daller 2019, etc.). A gradient approach to language dominance that considers extra-linguistic factors both at the individual and community levels, such as level of exposure, frequency and domains of use, and age of acquisition among many others, definitely allows for a better understanding of the links between bilingual language use and the outcomes of contact-induced change.

Additionally, modeling contact outcomes based on individual speaker behavior, while certainly insightful, does not directly address the propagation of change. This relates to Weinreich, Labov, and Herzog’s (1968) transition problem in language change more generally, and for language contact more specifically, centers on the question of how to link together Muysken’s (2010) aggregates of language contact. If change begins from the variation seen in individual patterns of speech, then what governs the spread of such innovations across the community (cf. Croft 2000)? In language contact, the nature of community bilingualism seems to play an important role (Thomason and Kaufman 1988), but cases where the data does not follow the expected results (as in instances of Type 1 hybrid formations) demand alternative explanations.

All these issues are relevant if we hope to reconstruct past contact scenarios based on contemporary ones. That is, we take the assumption that the mechanisms that apply synchronically must be the same ones that have applied in the past, and this is known as the Uniformitarian Principle. However, the main issue behind this principle is that we cannot assume that the social processes that operate in the present are actually comparable to those that operated in the past. For instance, many of the social concepts and models used to investigate particular linguistic phenomena, such as norms, standards, and prestige, may greatly differ across communities and across time periods (cf. Labov 1994:23, Bergs 2012:96). In reconstructing historical contact scenarios, speaker-based models such as van Coetsem (2000) are within the scope of the Uniformitarian Principle because we can assume that the mechanisms governing human cognition have not changed. At the same time, however, cognitive processes only present one side of the picture. That is, the psycholinguistic notion of language dominance also relies on extra-linguistic factors which are dynamic and are influenced by community-wide factors. There is thus the need to strengthen the current models and frameworks for language contact to better account for these considerations.