The role of applicative morphology in marking the telicity of applicative verbs in Ruruuli-Lunyala

This study examines the role of the morphology of applicatives in marking the telicity of applicative verbs in Ruruuli-Lunyala. Applicative verbs in Bantu languages have largely been investigated as allowing a new object Noun Phrase (NP) within the subcategorization of their base verb; this leads to a change of valency with the new Noun Phrase (NP) often giving a certain thematic role. However, less attention has been put on the role of the applicative in supporting the notion that argument projection may be aspectually determined. All applicative verbs in Ruruuli-Lunyala can be used with specialised overt telicity markers kakyarumwei ‘completely’ / ’very’/ ‘a lot’ and were ‘for nothing’. The research findings indicate an interesting relationship between these degree modifiers and applicative morphology. Thus, applicative verbs can be classified into two types, namely atelic and non-atelic applicative verbs. The post-verbal ‘applied object’ position /applicative adverbial position can be called the ‘applied constituent position’.


INTRODUCTION
The objective of this paper is to examine the role of the morphology of applicatives in marking the telicity of applicative verbs in Ruruuli-Lunyala. We examine how the form and function of applicatives in the morphological processes of applicative derivation mark telicity in Ruruuli-Lunyala.
Ruruuli-Lunyala is a tonal Bantu language of the Niger-Congo language family spoken by some of the inhabitants of River Nile-Lake Kyoga basin of Central Uganda Bureau of Statistics (2014) and estimate that there are 190,000 Ruruuli speakers. Uganda Bureau of Statistics (2014) also mentions 47,699 Lunyala dialect speakers. This means that there are roughly 237,699 speakers of Ruruuli-Lunyala 1 . Applicatives "involve a derived verb form combined with a subject semantically identical to that of the non-derived form of the same verb, and with an applied object representing a participant that cannot be encoded as a core argument of the same verb in its non-derived form" (Creissels, 2004: 3). In contrast, Kifle (2012: 106) refers to an applicative as a grammatical expression that morphosyntactically codes an altered construal of an event. When applicative morphemes are attached to certain verbs, they may add an argument that increases the valency of the clause from transitive to ditransitive. Valency increase is also possible with intransitive verbs as applicative verb licenses the presence of an additional argument with the syntactic role of the object (Bostoen and Mundeke, 2011).
However, the above valency-based definitions seem to assume that applicative constructions are always valency-increasing. This is not entirely accurate because the applicative construction may modify verb valency by either adding or suppressing an object, depending on the base it attaches (e.g. in Shiwilu, Kawapanan) (Valenzuela, 2016). In Yagua (Peba-Yaguan), the applicative marker may neither bring about valency change nor affect the semantic role of an object, as it encodes a greater sense in "intensity" (Payne, 2000). It may also not modify verb valency at all in some Bantu languages (for example, in Luganda, Swahili and Bemba) (Marten and Mous, 2015). In some cases it is the pragmatic function of the applicative verb that brings about a non-valency-changing environment. In such cases, the applicative verbs can be encoded by extra inferential effects derived from the predicate as instruction of concept strengthening: "The hearer is entitled to construct a concept which is "stronger" than a potential concept constructed from a corresponding base verb" (Marten and Mous, 2015:7). Applicative constructions aimed at encoding a strengthened concept and results in valency-preserving form can be illustrated in the following examples.
Examples 1 and 2 given above involve cases where applicative verbs are used without an attendant change in valency, and one relies on the pragmatic interpretation to draw distinct meanings from them. Although sentences (1a) and (1b) have the same verb valency, the verb in the former is in its basic form, while the verb in the latter is an applicative verb. Marten and Mous (2015: 11) attribute the difference in meaning in the principle of concept strengthening such that simply "eating" differs from the more emphatic "in_the_room_eating".
Sentence (2) is an example of "double applicative" 2 construction involving an augmentative applicative verb somerera "read diligently". Marten and Mous (2015: 13) explain that augmentative applicative verbs "are formed by extending the base verb twice", and this implies som-a "read" becomes som-er-er-a "read diligently". The purpose of augmentative applicative verbs, according to Marten and Mous (2015: 13), "is either to introduce a new object, similar to the simple applied form, or … to encode a strengthened concept". Marten and Mous"s views widened Ashton et al. (1954)"s who considered augmentative verbs as mere extension of the verb-action in time and duration. The notion of concept strengthening can as well be analyzed as concept reducing, as used in, for example, Bemba (Bantu); sunk "push", sunk-il-il "push a little at a time" and pyang "sweep a little at a time" (Sharman, 1963: 68). This implies that concept strengthening/concept reducing and verb valency change can be either exclusively or integratively analysed in applicative constructions. Despite the above valencybased, pragmatic and concept strengthening definitional arguments about applicative constructions, they do not fully explore the notion that argument projection may be aspectually determined in some Bantu languages. Isingoma and Jane (2012) argue that despite different studies making theoretical generalizations based on the empirical patterns of particular languages, there exists considerable variation in the behaviour of applicatives in different languages. To show these variations, Jerro (2015) suggests that each applicative in a language encodes its own idiosyncratic restrictions on object symmetry. In relation to this, some Bantu languages allow adding an extra object to unergative verbs (Bresnan and Moshi, 1993). Furthermore, Pylkkänen (2008) shows that languages like Kichaga allows high applicatives, while Chichewa allows low applicatives only. Dixon (2012) justifies such variations by explaining that different languages can exclusively permit applicative constructions with transitive and intransitive verbs, or intransitive verbs, or transitive verbs. As a result, those variations justify the fact that their conclusions are not entirely satisfactory to account for applicative constructions in Ruruuli-Lunyala.

METHODOLOGY
We used corpus and participant observation methods of data collection since grammatical analyses should be arrived at inductively, through observations of a corpus of recorded discourse, supplemented by direct observation of how the language is used in the community (Dixon, 2012). One needs to gather a broad database which should contain numerous genres, and thereafter supplementary data should be gathered through participant observation (Bowern, 2015). One should represent multiple speakers as well as control for other variables, such as gender, age place of residence as well as length of texts.
We used Ruruuli-Lunyala-English Dictionary (RLED) corpus data which was compiled by a Ruruuli-Lunyala language documentation project: A comprehensive bilingual talking Ruruuli/Lunyala-English dictionary with a descriptive basic grammar for language revitalisation and enhancement of mother-tongue based education (PI Saudah Namyalo, Makerere University, funded for 2017-2020, Volkswagen Foundation). The RLED corpus consisted of 159,641 words from 74 written Ruruuli-Lunyala texts. The corpus was produced by speakers of Ruruuli-Lunyala from four districts of Uganda, namely Nakasogola, Kayunga, Kiryandongo and Buyende.

The morphology of applicatives
The morphology of applicatives in Ruruuli-Lunyala is derivational. The applicative suffix in Ruruuli-Lunyala has one deep structure but three surface realizations: The deep structure is -il-, while the surface form can be the suffix allomorphs -ir-/-i-, -er/-e-and -r-. At the surface level, /r/ can be realized as an applicative marker after a pattern of morphological processes involving suffixation, vowel harmony, segment deletion and compensatory lengthening as elaborated under "Infixation vis-à-vis Metathesis arguments". The: applicative suffixes -ir-ander-have morpho-phonemically conditioned variants -ior -e-respectively.

Atuhairwe 37
The choice of the applicative suffix is dependent on vowel harmony within the verb root. On the one hand, if the preceding vowel within the verb root is /i/, /u/ or /a/, the applicative suffix is -ir-(cf. Baker et al., 2012;Bostoen and Mundeke, 2011;Ngonyani, 1998;Nurse and Philippson, 2006;Marten and Mous, 2015). The applicative marker -ir-in Ruruuli-Lunyala is illustrated as follows.

Ruruuli-Lunyala (Bantu, Uganda; Primary Source)
Basic form Derived form (i) sumita pierce" sumit-ir-a "pierce for/at" (ii) langa "announce" lang-ir-a "announce for/at" (iii) pima "measure" pim-ir-a "measure for/at" (iv) suna "pinch" sun-ir-a "pinch for/at" At least one of the vowels, /a/, /u/, and /i/ appears in the preceding syllable in the verb root/base of the examples shown above. One of the common characteristics which all the sentences share is the applicative suffix -ir-.
On the other hand, if the preceding vowel within some verb roots is /e/ or /o/, the applicative suffix is -er-. This is backed up with examples as follows:

Ruruuli-Lunyala (Bantu, Uganda; Primary Source) Basic form
Derived form (i) goda "bend" god-er-a "bend for/at" (ii) kona "hit" kon-er-a "hit for/at" (iii) beyega "deceive" beyeg-er-a "deceive for/at" (iv) kema "groan" kem-er-a "groan for/at" With respect to the examples (4 i-iv) shown above, either /e/ or /o/ appears in the preceding syllable in the verb root. One of the common characteristics which all these verbs share is the applicative suffix -er-. Both applicative suffixes -ir-and -er-have their morphophonemically conditioned variants -i-and -erespectively. These variants surface under two circumstances: Either the applicative construction is in the perfective or the applicative marker involves double suffixation. The applicative marker preceding the perfective marker loses the consonant /r/ but preserves the vowel. In such an event, there are two vowels adjoining each other between the remaining vowel and the vowel attached to the perfective marker after the deletion of the consonant /r/. Below are the examples showing the applicative marker variants -i-and -e-:

Basic form
Derivation process Derived form(APPL+PFV) (i) soma "study" soma-er-ire som-e-ire "studied for/at" (ii) leeta "bring" leeta-er-ire leet-e-ire "brought for/at" (iii) tangaran-a "meet" tangarana-ir-ire tangaran-i-ire "met for/at" (iv) iruk-a "run" iruk-ir-ire iruk-i-ire "ran for/at" -i-and -e-are used in the final derived form instead of -irand -er-respectively. In both roots, iruk and som, there is addition of a corresponding applicative suffix -ir-and -erat the deep structure level. However, the applicative verbs, for instance, are iruk-i-ire and som-e-ire instead of iruk-ir-ire and som-er-ire, respectively. There is the deletion of the consonant /r/ from the applicative marker at the deep structure level. As a consequence, the surface structure shows only the vowels -i-and -e-. This is because all liquids preceding the perfective marker -ire are deleted in Ruruuli-Lunyala. The deletion can be attributed to Ruruuli-Lunyala phonotactic constraint which restricts two liquids near each other involving verb extension elements 3 . As a consequence, the initial vowel of the perfective marker and the applicative vowel keep in their "stand-alone" positions.
In relation to the above, the construction involving the surface realisation of the applicative suffix -er-is also found in monosyllabic verbs. With the exception of the liquid /l/-initial monosyllabics like lya "eat", all monosyllabic verbs use the applicative suffix -er-in the applicative construction in the present tense. Relatedly, the applicative suffix -er-in monosyllabic verbs is reduplicated in the perfective to form double applicative structure -er-er. (Double applicative structure is elaborated in results and discussion). However, the liquid /r/ that precedes the perfective marker -ire is deleted. These changes in the form of derived verbs involving monosyllabic verbs are illustrated below:

Ruruuli-Lunyala (Bantu, Uganda; Primary Source) Basic form Derived 'present' Derived 'perfective'
The surface realisation of the applicative is consistentlyer-in the present tense in all the monosyllabic verbs mentioned above except in example (6iv). The clear change of form between a derived applicative verb in the present tense and the one in the perfective involving the monosyllabic verbs is a consistent observation. Excluding liquid-initial monosyllabic verb lya-"eat" 4 , we found that all other monosyllabic verbs behave alike. We postulate that monosyllabic derived verbs behave like augmentative verbs by extending the verb base twice in the perfective. Therefore, it can be interpreted that the change in tense involving monosyllabic verbs in Ruruuli-Lunyala occurs with a change in the form of an applicative marker from single (in present) to double applicative (in the perfective).

Infixation vis-à-vis Metathesis arguments
In addition to being realized as a suffix, the applicative marker in Ruruuli-Lunyala can arguably occupy the position of an infix after base modification. On the one hand, there is a possible scenario that there occurs the insertion of what looks like the "infixes" -ri-or -reexclusively within /z/-ending roots that have three or more syllables. These "infixes" also adhere to the same principles of vowel harmony governing the suffix applicative marker construction: If the preceding vowel within the verb root is /i/, /u/ or /a/, the applicative infix isri-and if that vowel is /e/ or /o/, the applicative infix is -re-. On the other hand, there can be an argument that the derived forms are a result of metathesis as illustrated in the data below:

Basic form
Presumed derivation Actual derived form (i) ongereza "lobby" ongerez-er-a ongerereza "lobby for/at" (ii) bonereza "punish" bonerez-er-a bonerereza punish for/at" (iii) angiriza "praise" angiriz-ir-a angiririza "praise for/at" (iv) ikiriza "accept ikiriz-ir-a ikiririza "accept for/at" With respect to the above examples, the applicative marker is highlighted in bold in the columns of presumed and actual derived forms. It is seemingly realized as an "infix", and placed immediately before the sibilant /z/ in the last syllable of the root under the actual derived forms.
The argument that the derived forms are a result of metathesis acknowledges the fact that suffixation is the primary morphological process of the applicative construction (s). This implies that verbs like ongere-re-za "lobby for/at" and etegere-re-za "assess for/at" were presumably ongerez-er-a lobby for/at and etegerez-er-a "assess for/at" such that the actual derived verbs are a result of the base modification process by metathesis. Firstly, the syllable /za/ and the suffix /er/ interchange positions. Secondly, the distinction between presumed and the actual derived verb is observed through segmental metathesis of the suffix, that is, -er-changes into -re-, while -ir-changes into -ri-. In principle, metathesis involves switching of two or more segments in the base, and these segment(s) are sounds that "must be next to each other" (Haspelmath and Sims, 2010:334;Langdon, 1976). In this, Haspelmath and Sims put emphasis on the "local" rearrangement of order of two (6) Ruruuli-Lunyala (Bantu, Uganda; Primary Source) Basic form Derived 'present' Derived 'perfective' (i) sy-a "grind" sy-er-a "grind for/at" si-er-e-ire "ground for/at" (ii) nyw-a "drink" nyw-er-a "drink for/at" nyw-er-e-ire "drunk for/at" (iii) bbw-a "tie" bbw-er-a "tie for/at" bbw-er-e-ire "tied for/at" (iv) li-a "eat" li-ir-a "eat for/at" li-ir-i-ire "ate for/at" The surface realisation of the applicative is consistently -er-in the present tense in all the (6) Ruruuli-Lunyala (Bantu, Uganda; Primary Source) Basic form Derived 'present' Derived 'perfective' (i) sy-a "grind" sy-er-a "grind for/at" si-er-e-ire "ground for/at" (ii) nyw-a "drink" nyw-er-a "drink for/at" nyw-er-e-ire "drunk for/at" (iii) bbw-a "tie" bbw-er-a "tie for/at" bbw-er-e-ire "tied for/at" (iv) li-a "eat" li-ir-a "eat for/at" li-ir-i-ire "ate for/at" The surface realisation of the applicative is consistently -er-in the present tense in all the uruuli-Lunyala (Bantu, Uganda; Primary Source) ic form Derived 'present' Derived 'perfective' -a "grind" sy-er-a "grind for/at" si-er-e-ire "ground for/at" yw-a "drink" nyw-er-a "drink for/at" nyw-er-e-ire "drunk for/at" bw-a "tie" bbw-er-a "tie for/at" bbw-er-e-ire "tied for/at" -a "eat" li-ir-a "eat for/at" li-ir-i-ire "ate for/at" surface realisation of the applicative is consistently -er-in the present tense in all the sounds or group of sounds, which are in contiguous relationship.
The metathesis argument would also be advanced to explain the insertion of applicative markers -ir-and -erwithin verb roots whose last syllable ends with the noncausative semi-vowel /j/. In relation to the above, there is the insertion of the presumed "infixes" -ir-and -er-within some verb-roots that end with the glide /j/. In such verbs, the applicative marker immediately precedes the glide /j/ in a non-causative environment. The applicative construction involved in this kind of presumed infixation is also vowel harmony-governed as earlier observed as regards -ri-/-re-metathesis arguments. This is illustrated as follows:

Ruruuli-Lunyala (Bantu, Uganda; Primary Source)
Wabasiryanga pamba o-a-basy-ir-a=nga pamba 2sgS-PST-sleep-APPL-FV=HAB 10.cotton "You would sleep on cotton." In the example above, the verb-root basy 'sleep" uses the presumed "infix" -ir-to form an applicative verb-stem basir-y-a 'sleep on'. This is the actual derived form that changed from the presumed form basy-ir-a 'sleep on". The applicative affix permits the object pamba 'cotton" to acquire transitive object status in a valency changing construction. Other verbs behaving in the same applicative infix formation as basya "sleep" include the following: 9. Ruruuli-Lunyala (Bantu, Uganda; Primary Source)

Basic form
Presumed derivation Actual derived form (i) bbeesy-a "comfort" bbeesy-er-a bbeesery-a "comfort for/at" (ii) tesy-a "dirty" tesy-er-a tesery-a "dirty for/at" (iii) weisy-a "run out of" weisy-ir-a weisiry-a "run out of (for/at)" (iii) maisy-a "throw" maisy-ir-a maisiry-a "throw for/at" With respect to examples (9i-iv), basic di-syllabic verbroots ending with /j/, adjacent to the sibilant /s/, use the presumed infix -ir-or -er-in applicative construction. The interchange of position between the glide /j/ and the verb extensions -erand -ir-is a true argument of metathesis. It is the base modification that produces the actual derived verbs. The glide in these roots cannot be detached from the entire root without changing the thematic meaning of the word. In verbs similar to the ones given above, detaching the glide from its root would render it non-existent in Ruruuli-Lunyala. We take the applicative marker as a suffix since metathesis processes involved in the applicative construction are maximally local, simple and logical.
In contrast to the presence of Ruruuli-Lunyala presumed "infix" applicative construction in the preceding paragraph, there are other /j/-ending verb-roots whose glide is a verb extension with a causative effect. In this case, the glide can be detached from the verb-root and the thematic meaning of the verb is maintained. For Atuhairwe 39 example, both naab "take a bath" and naaby "bathe" have similar thematic meaning, but the /j/ in the latter is a verb extension. This means that naaby is a verb stem while naab is the root. The root without the final /j/ is an intransitive verb. The addition of /j/, which has a causative effect, makes the verb transitive as further illustrated in an applicative construction below: 10. Ruruuli-Lunyala (Bantu, Uganda; Primary Source) Abakunaabirya. a-ba-ku-naab-ir-y-a REL-3pl-who-PROG-bathe-APPL-CAUS-FV "Those who are bathing (someone) for you." In example 10 above, the /j/ in naaby "bathe" has a causative effect (causing one to take bath), and it is not part of the root. The applicative marker, -ir-, is placed between the verb-root and causative /j/. The interpretation in this case is that the applicative marker is not a presumed "infix"; instead, it is a regular suffix. There are many verbs that behave in this way and below are more examples: (11) Ruruuli-Lunyala (Bantu, Uganda; Primary Source)

Applicative suffix -r-
There are some cases in Ruruuli-Lunyala where the surface realisation of an applicative is a -r-. The applicative suffix -ris an applicative marker allomorph from -er-/-ir-and found in applicative constructions involving verb-roots/verb-bases that end with the liquid /l/. There happens suffixation, segment deletion, after which, compensatory lengthening of the root vowel takes place in the present tense. The final derived verb depicts an applicative marker which has lost its vowel such that -eror -irin the basic form becomes /r/ in the derived form. All verb roots in this category have their final /l/ seemingly change to /r/ because the /l/ in the basic form and /r/ in the derived form appear to be in contrastive distribution. The applicative marker in such cases must follow a long vowel. To back up our argument, we mention /l/-ending verb-/bases which have an already existing long vowel in the root: A good example is the tense vowel /o/ as in simoola 'speak", which changes into simoo-r-a 'speak for/at'. By vowel harmony, simoola "speak" adds the applicative -er-(simool-er-a). Then /l/ is deleted (simooøer-a) because Ruruuli-Lunyala does not allow two liquids in syllables in contiguous relationship involving verb extension. Next is the deletion of /e/ because two root vowels cannot be followed by another vowel in Ruruuli-Lunyala. The resultant derived verb can now be realized as simoo-r-a "speak for". This process can be reduplicated in other verbs. For example, sosotoola 'unwrap (food)' becomes sosotoo-r-a 'unwrap (food) for/at". Related to the above, there are other word-forms whose applicative marking involves the applicative suffix -r-. A case in point are /l/-ending disyllabic-roots, whose initial syllable contains the semi-vowel /w/ as in twala 'take" that becomes twa-r-a "take for/at" and lwala 'fall sick" becomes lwa-r-a "fall sick for/at". Vowel harmony in twala 'take", for example, adds applicative -ir-(twal-ir-a). Then /l/ is deleted (twaø-ir-a) because of the already mentioned restriction about two liquids in syllables in contiguous relationship. After this, there occurs the deletion of /i/ because of the underlying semi-vowel qualities in the preceding syllable structure, which contains semi-vowel /w/. This semi-vowel is as a result of compensatory lengthening in terms of retiming of segments; namely, underlyingly twala itself comes from tuala: As the high vowel /u/ changes to the glide /w/, the vowel /a/, which follows the glide, gets lengthened so as to preserve the bimoraic status of the first syllable in tuala (cf. Katamba 1989: 171-172). The resultant derived verb can now be realized as twa-r-a "take for/at" after applicativisation.
It is also notable that in cases where the basic /l/ending verb-root has one mora vowel-structure, there is vowel lengthening in the process of applicative construction. With respect to this category of verbs, kola 'work" becomes koo-r-a 'work for/at", gula 'buy" becomes guu-r-a 'buy for/at" and mala 'finish" becomes maa-r-a 'finish for/at". Based on vowel harmony gula 'buy", for example, adds the applicative -ir-(gul-ir-a). Then /l/ is deleted (guø-ir-a) because of the already mentioned restriction about liquids in syllables in contiguous relationship. After this, /u/ adjoins with /i/, and since /u/ is stressed in the root vowel, /i/ is also deleted. Moreover, two dissimilar vowels in form of /ui/ are not allowed in Ruruuli-Lunyala. Next is the compensatory lengthening of the root vowel /u/. The resultant derived verb can now be realized as guu-r-a 'buy for/at". In (12a), (12b) and (12c) are more examples involving /l/-ending verb-roots:

Basic form (present tense) Derived form a(i) foola
"transform" foo-r-a "transform for/at" (ii) duula "brag" duu-r-a "brag for/at" b(i) swala "be ashamed" swa-r-a "be ashamed for/at" (ii) zwala "dress" zwa-r-a "dress for/at" c(i) bala "count" baa-r-a "count for/at" (ii) kul-a "grow" kuu-r-a "grow for/at" The distribution of verbs in (12a), (12b), and (12c) above are arguably cases of -rapplicative marker. All the verbroots within the three groups end with the liquid /l/. In addition to this, all the derived applicative verbs have a two-mora vowel structure positioned immediately before the final /r/ in the verb base. The basic verbs in (12a) form their applicative construction by suffixation and segment deletion. The applicative verb construction in (12b) also involves a suffixation and segment deletion, but this time, in the environment of a semi-vowel /w/. Finally, the applicative construction for the verbs under group (12c) involves /l/-ending verb-roots whose basic form contains a short vowel positioned immediately before the final /l/. Under group (12c), suffixation, segment deletion and compensatory lengthening of the root vowel take place in Ruruuli-Lunyala applicative construction.

Double suffixes in applicative construction
There are lexico-pragmatic cases that involve double suffixation in Ruruuli-Lunyala. In such instances, the applicative marker appears twice in immediate succession. This implies that the surface realisation of the applicative marker is either -er-er-or -ir-ir-. However, there occurs the deletion of /r/ in the first applicative suffix such that the surface realisation of the applicative appears as -e-er-or -i-ir. The resultant applicative verb has the same thematic meaning as the basic verb as illustrated below: 13. Ruruuli-Lunyala (Bantu, Uganda; Primary Source) a. Zubiira omusaale ogwo. zub-i-ir-a o-musaale o-gwo 2sgS-weed-APPL-APPL-FV AUG-3.tree 3-that "You weed around that tree." b. Bakukusekeera.
ba-ku-ku-sek-e-er-a 3plS-2sgO-PROG-laugh-APPL-APPL-FV "They are laughing at you." In example 13a, the applicative marker is realised as double suffix -i-ir-(-ir-ir-) in zub-i-ir "weed around". In Example 13b, double suffix -e-er-(-er-er-) is shown in sek-e-er "laugh at". In both cases, the /r/ in the first applicative suffix is deleted. The deletion creates the long vowel which is observed in applicative construction involving /l/-ending verb-roots. More examples can be shown below:

Basic form Derived form (Present tense) (i) enga "ripen"
eng-e-er-a "be yellowish" (ii) gona "sleep" gon-e-er-a "sleep soundly" (iii) babila "irritate" bab-i-ir-a "have a peppery taste" (iv) sala "pain" Sal-i-ir-a "pain by" As observed in examples (14i-v), there is a thematic relationship between the basic verb and the non-basic verb. For instance, the basic verb gon-a "sleep" and the non-basic verb goneera "sleep soundly" have a thematic relationship. Both of them have the same verb-root gon "sleep". It can be interpreted that the non-basic form extends the semantics of the basic form to bring about contextualised extended meanings. However, eng-e-er-a "be yellowish" and bab-i-ir-a "have a peppery taste" are cases of lexicalisation. A single applicative suffix applying twice on a single verb is also found in languages such as Ciluba (De Kind and Bostoen, 2012), Pogoro (Hendle, 1907), Swahili (Marten, 2003), Luganda (Marten and Mous, 2015) and Tswana (Pacchiarotti, 2018). Apart from extending the meaning implied by basicverb form, the double-suffixes representing the applicative marker can be locational to imply "proximity/nearness to something" as in zub-i-ir "weed around". According to Pacchiarotti (2018: 159), such applicative constructions "over time can lose their intensifying, repetitive, persistive meaning and replace the root" under the process of lexicalisation.

Causative -esyand -isy with applicative reading
The suffixes -isy and -esy are basically causatives but with a secondary function of applicative reading. Ruruuli-Lunyala has ably shown that it can use the two causative suffixes (-esy-and -isy) in an applicative construction. This is specifically used in respect of applicative construction involving the referent of the Noun Phrase (NP) that is attached as a weapon, tool, or material used in the action of the verb (cf. Peterson, 2007). We illustrated instrumental applicatives with causative suffixes in sentence (15) (cf. Saeed, 2016). Sentence (15a) shows comitative na 'with" taking the instrumental argument ekiso "machete" in a peripheral function. In sentence (15b), the peripheral argument is put in the transitive object function by the causative suffix -isy, which has an applicative reading. A similar situation can also be found in non-Bantu languages, for example, Javanese (Austronesian language family) whereby suffixes -i andaké can depict a causative/applicative syncretism (Hemmings, 2013).

Applicatives in degree adverbials
Most applicative predicates can receive a degree argument either for inherent lexical reasons or by virtue of their semantic and/or syntactic context. There exists open and closed degree scales as regards gradable predicates 5 in applicative construction (Caudal and Nicolas, 2005). An adjective like hard is not bound in terms of ranges of degree lexically associated with it. It is said to have an open scale since there is no limit to "hardness". In contrast, there is a maximal degree associated with the verb "sweep". This is because once a room has been swept, there is no further sweeping that can take place. Based on this, the degree scale lexically associated with "sweep" is said to be closed.
In relation to the above, Jackendoff (1996: 27) argues that the degree of flatness, size and redness can be treated as degree scales while temperature is a scalar on a linear scale. In the process, an object can "move" along the scale, and if the "path" has a boundary (reaching hot or big), the sentence is interpreted as telic. In instances where the "path" is non-bounded (going on indefinitely in the "hotwise" or "bigwise" direction), the sentence is atelic (Jackendoff, 1996). On this basis, telicity can account for selectional restrictions and differences in usage governing closed and open degree scales in different languages. As such, there exists closed scalar structures in instances where English adjectives combine with, for instance, the adverb "completely", but not with "very" and "extremely". Additionally, some English adjectives are said to have open scalar structures when they combine with adverbs such as "very" and "extremely" but not "completely". Interestingly, the three adverbs can be represented meaningfully by a single adverb kakyarumwei in Ruruuli-Lunyala. The question of whether open and closed degree scales maintain their true contextual meanings in Ruruuli-Lunyala as in English was examined as below: In Ruruuli-Lunyala, kakyarumwei can be used in both open and closed scales as observed above. It can mean "completely" in closed scales as in (16a) and "very"/"extremely" in open scales as in (16b). Interestingly, in both cases, it is licensed by an applicative marker. Additionally, kakyarumwei can be used to mean "a lot" in the place of "extremely"/"very" with open and closed scale readings. In either of these cases, an applicative marker cannot be done without as illustrated below: In Ruruuli-Lunyala, a single adverb kakyarumwei can be interpreted as "completely" in English in the example 17a above, and as "a lot" in sentence 17b. It is once again licensed by an applicative marker in an intransitive applicative construction.
Based on examples 17a and b above, the distribution and interpretation of degree arguments in terms of open and closed scales can correspond to quantity and intensity scales respectively. With regard to quantity scales, kakyarumwei "completely" is used to imply completion/ total consumption of the incremental theme as in "Peter ate maize completely." In contrast, it can render intensive scales reading. Intensive scales accord intensive interpretation with the semantic equivalence of "very/extremely" as in "This metal is very/extremely hard." At the same time, intensive scales involve verbs morphologically related with an adjective.

Contrasting kakyarumwei 'completely'/'very'/'a lot' and bwereere 'for nothing' with telicity
Kakyarumwei "completely"/"very"/"a lot" is significant in marking telicity of applicative verbs. Bwereere "for nothing" has a unique interpretive and telicity oriented function which by default describes finality of the sentence in question. Although the two share similar syntactic behaviour, particularly in their appearance in the applicative object slot, they interact differently with the internal structure of events. As such, both kakyarumwei "completely"/"very"/"a lot" and bwereere "for nothing" have different semantic and syntactic interpretations.
On the one hand, the presence of kakyarumwei "completely"/"very"/"a lot" influences aspectual properties involving degree arguments and verbal predicate. On the other hand, bwereere "for nothing" has an interpretive function uniquely catering for purposefulness and item price implications of the event in question. We arranged our findings based on Vendler's (1967) verb classification structure whose work has remained formidable in academic circles as cited by many recent scholars (e.g. Bennett, 1988;Pollock, 2012;Rothstein, 2008;Ruben, 2015).
m-many-ir-a kakyarumwei o-kwiruka 1sgS-mait-APPL-FV a lot AUG-running "I know running a lot." We agree with Abusch (1986) that argument gradable predicates do not themselves exhibit properties, but rather encode measure functions which associate objects with ordered values on a scale, or degrees. The inclusion of kakyarumwei "a lot", in the predicate of (18a), creates a "point of relativity" in the verb-event of many 'know". The derived verb many-ir 'know a lot" makes such state verb attain more certainty and gradable predicate status. The applicative verb is hence better understood with intensity interpretation, which indicates maximum degree for an event to be realised as true and complete. The fact that there is maximum degree implies the presence of "minimum degree". In this way, the use of kakyarumwei 'completely" in state verbs marks "completeness" or a point of how "best" an event can be. Without the applicative adverbial kakyarumwei 'completely', the event has seemingly not reached its endstate. Telicity marking, in the context of Ruruuli-Lunyala, would give such state verb, a telic predication as the affected objects do not reach their endstate without the gradable value. Therefore, the construction formed with the derived verb takes a telic predication.

Activity verbs
Similar to state verbs, the presence of kakyarumwei in the predicate of activity verbs, as in (19), necessitates +intensity interpretation. The event in the basic verb rundy 'pull" takes an applicative degree adverbial in the derived form rundiry "pull a lot". There is also a "point of relativity" considering the least amount of force and the maximum one can apply to rope pulling. Thus verbs of activities, like pull a rope, can also have a gradable predicate status. The construction formed with the activities derived verb also takes an atelic predication since heightened intensity does not necessarily mean the ending of an event as illustrated below: o-Musa a-rundy-ir-ire kakyarumwei loole AUG-1-Musa 3sgS-pull-APPL-PFV a lot 9.lorry "Musa pulled the lorry a lot." Kakyarumwei is in an applicative object slot as shown in (19a) above, and is meant to stress more force which Musa applied in the pulling of the lorry.

Accomplishments (like cut a tree or dig a hole)
An accomplishment is realized only once the endpoint is o-muyiigi a-tem-e-ire kakyarumwei o-musaale AUG-1.hunter 3sgS-cut-APPL-PFV completely AUG-3.tree "The hunter completely cut a tree." The applicative adverbial kakyarumwei in (20a) above is used with an accomplishment verb to bring out total completion of the event in the semantic function of the gradable predicate. The acceptability of "cut a tree" as a telic predication, is true of an event only if it leads to an endstate in which the affected participant has reached its endpoint. Without kakyarumwei, there is no certainty that the endstate associated with "tree-cutting" in (20a) has been reached. Telicity marking has to imply beyond a mere sequence of distinct change of state eventualities involving quantity degree arguments as in (20a) Barcelona e-wangul-ir-ire kakyarumwei o-mupiira Barcelona 3S-win-APPL-PFV emphatically AUG-3.football "Barcelona emphatically won the football match." Embedded in the notion of situations that take a moment only, verbs of achievements use kakyarumwei as a focusing device to emphasise the maxim point in the event. With respect to (21a) above, to win a football match is momentous only at a time when one team has scored more goals than the other. Kakyarumwei, in this Atuhairwe 43 context, is used to emphasize winning by a wide margin, which notion can only be captured through an applicative construction. It should, however, be noted that endstate is already achieved by the time kakyarumwei is employed to emphasize the momentous completion. In this context, kakyarumwei "emphatically" is used to convey what we call redundant magnitude, which implies the event reached an endstate earlier than thought. As such, derived verbs under achievements verb class are also said to have a telic predication in Ruruuli-Lunyala. Based on the above arguments, the use of kakyarumwei as an endstate determiner makes the verbal predicate acquire a telic status. Conversely, if kakyarumwei is used as a mere intensifier, the verbal predicate takes an atelic status. Therefore, applicative verbs of states, accomplishments, and achievements are telic, while applicative activities are atelic. Whereas Caudal and Nicolas (2005) call completely and partially event descriptor modifiers, we call kakyarumwei "completely" a telicity marker for its role in influencing the internal structure of an event.
Bwereere "for nothing" is at all times used with an applicative construction. It also occupies the applicative object slot, and can be illustrated using the pair sentences of those exemplified under different verbs in Vendler's (1967) taxonomy above.
c. o-muyiigi a-tem-i-ire bwereere o-musaale AUG-1.hunter 3sgS-cut-APPL-PFV "for nothing" AUG-3.tree "The hunter cut a tree for nothing." In all the above sentences (18b, 19b, 20b, 21b) bwereere "for nothing" has a common semantic role: It has a default interpretive function to mean "without reason" or "at no charge" or without gain: Something is done with no reason as in He was fired for nothing; Something is attained without paying any money for it as in I got sweets for nothing; Something is done but no benefit is achieved as in He won the gold medal for nothing. In general, the information given conveys "no purpose" or benefit about the entire outcome of the whole sentence/clause. All these meanings hold true for verbs classified as states, activities, accomplishments and achievements as exemplified above. Despite variations in meanings, bwereere "for nothing" and kakyarumwei "completely/very/a lot" exist in the same syntactic pattern in applicative construction.

Conclusion
Based on the data collected and the analysis done in this study, we conclude that: (i) Applicative verbs can therefore be classified into two types, namely atelic and non-atelic applicative verbs depending on the roles of an applicative adverbial. On the one hand, if the applicative adverbial plays the role of intensifier, the applicative verb is classified as atelic. For instance, the use of kakyarumwei "completely"/"very"/"a lot" with verbs manya "know" rundya "pull" and tema "cut" gives the verbal predicate an atelic status. On the other hand, if the applicative adverbial plays the role of redundant magnitude, the applicative verb is classified as telic. For example, verbs wangula "win", soka "start" and basya "sleep" when used with kakyarumwei "completely/very/a lot", the verbal predicate takes a telic status. In this way, kakyarumwei "completely/very/a lot" stands out as a major classifier of applicative verbs in Ruruuli-Lunyala. Although bwereere "for nothing" is also described as a telicity marker, its default goal function across all verbs in Ruruuli-Lunyala makes it a distinct classifier of applicative verbs.
(ii) Ruruuli-Lunyala applicative form can change from single applicative (-er/-ir) to double applicative (-er-er/-irir) in present and perfective respectively. However, the form of double applicatives in monosyllabic verbs in perfective undergoes segment deletion of the second -r such that the surface realization is either -er-e or -ir-r. Double applicative constructions in augmentative verbforms (-er-er/-ir-ir) also undergo segment deletion although, this time, it is the first -r that gets deleted. The surface realization is then realized as -e-er/-i-ir. The meaning of augmentative form is mainly for lexicalization, extension of verb-meaning, repeativeness or has proximity bearings. The double-applicative forms can, but not always lead to multiple post-verbal arguments.
(iii) We conclude that Ruruuli-Lunyala applicatives can, in in another example of applicative form and tense interface. Applicative suffix -rbasically involves the deletion of the vowel preceding the regular applicative marker -ir-/-er in /l/-ending verb roots/bases. However, in the perfective forms, the regular affixation as regards applicative formation takes precedence.