• No se han encontrado resultados

4.1 La Evaluación y la Investigación

5.1.3 Momento oportuno para la evaluación de impacto

This subsection reports on assimilatory processes in loanwords. Although the elicitation context is not conducive for phonological processes such as assimilation and lenition as the setting is semi-formal and such processes appear mostly in relaxed speech (cf. Gurevich 2011: 1565), a number of assimilation processes appear in the corpus.

The cross-linguistically most common phonological process, assimilation, refers to a process by which segments come to share certain features (Zsiga 2011: 1919). It could be partial where one or more features are shared or complete where all features are shared resulting in a false geminate (ibid). The assimilation processes in the corpus are divided into four types, presented in the subsections below.

3.1.2.2.1 Voice assimilation

Voice assimilation refers to a process whereby adjacent consonants agree in voicing. As can be seen in the examples in (14), the voiceless obstruents /f/ and /k/ trigger voice assimilation in the first three cases and target the voiced /d/ and /z/. In (d), the voiced alveolar fricative spreads voice to the voiceless alveolar stop.

(14) a) headphone hitfon ~ hitifon

b) foolscap fuluskaab

c) vodka vootka ~ votika

d) pizza biidza

This voice assimilation is attributed to the cross-linguistic markedness requirement that obstruent clusters agree in voicing where the last obstruent functions as the trigger (cf. Itô, Mester & Padgett 1995; Lombardi 1999). To check whether this represents a case of TETU or it is an active AA constraint, I checked AA native words and found that this constraint is attested in AA as the examples in (15) show.

95

(15) ʔagsaam > ʔaksaam ‗sections‘ cf. gisim ‗section‘ wagt > wakt ‗time‘ cf. ʔawgaat ‗times‘

ʔafḍaʕ > ʔavḍaʕ ‗more terrible‘, cf. faḍiiʕ ‗terrible‘ bit-ziid > bidziid ‗you m.s. increase‘

Moreover, I checked the whole corpus of loanwords to see if there were any cases that violate this constraint. No such cases were found except for clusters involving the voiced labial /b/. The behaviour of /b/ is also attested in native words, as in ʔakbar ‗older‘ xuṭb[e] ‗engagement‘, where voice does not spread to the preceding voiceless stop. Therefore, I argue that /b/, being the only labial stop, could be underlyingly unspecified for voice in AA (cf. Abu-Mansour 1996). More loanwords that show that /b/ does not spread voice are given in (16).

(16) Voiceless obstruent-b clusters

facebook feesbuk

football faṭbul

spare sbeer

The corpus also shows that voice does not spread from a sonorant to an obstruent in obstruent-sonorant clusters. The same is found in native words. This again suggests that sonorants in AA are not specified for voice underlyingly (cf. Abu-Mansour 1996). A few examples are given in (17).

(17) Voiceless obstruent-sonorant clusters cholesterol kulistrool

christmas kriṣmaṣ

syringe srindʒ[e]

3.1.2.2.2 Place assimilation

Place assimilation is rare in the corpus and appears in two words given in (18).

(18) convoy kamboy

baking powder bikimbawdar

The trigger is a labial while the target is a nasal. This is consistent with cross- linguistic literature on assimilation and is reminiscent of the markedness hierarchy where dorsal and labial sounds are more marked than coronals (Mohanan 1993).This

96

implies that coronals are the least marked in AA. This is motivated phonetically as coronals have weak cues of place of articulation, which are superseded by the stronger cues of labials and dorsals (Hall 2011).

3.1.2.2.3 Complete assimilation

Complete assimilation appears in two words only given in (19).

(19) charleston ʃallistun

stainless staallis

In the first example, the lateral functions as a trigger and the rhotic is the target. This is unnatural as it is expected that the weaker lateral sound assimilates to the stronger rhotic sound (cf. Heselwood et al. 2011, cited in El-Ramli 2012). This is also attested in native words such as /jiɣfir lak/ > jixfillak ‗he forgives you m.s.‘. It seems that the position of the trigger is the decisive factor with regressive assimilation overriding progressive one.

Moreover, it is noticed that the two cases of complete assimilation affect consonants that are very similar (see Table 3.4 above) where it obtains between /r/ and /l/ on the one hand, and between /n/ and /l/ on the other (cf. Zsiga 2011: 1925). In fact, it is believed that all types of assimilation in AA tend to require great similarities between triggers and targets; the more similar they are the more likely assimilation is expected to occur (cf. Zuraiq & Abu-Joudeh 2013 for other Jordanian dialects). More evidence for this comes from the fact that obstruent gutturals such as /ʕ/ do not spread voice like other obstruents in obstruent clusters (e.g. tʕaani ‗you suffer‘. Some researchers, e.g. Zawaydeh (1999), argue that /ʕ/ is a sonorant. Rather, I think that the similarity between this sound and other obstruents is not enough to trigger assimilation. Note that assimilation in all the cases above is regressive, which is in line with markedness criteria.

A last type of assimilation also appears in consonant-to-vowel assimilation as in (20) below.

(20) a) baby bubbu

b) pedicure budikiir

97

d) balance balanṣ

e) jersey dʒurzaay[e]

In (20a-b), the labial consonant triggers rounding of the vowel. Similarly, the non- primary [guttural] consonants /r/ and /ṣ/ lower and back the vowels in (20c) and (20d). Finally, vowel rounding in (20e) is induced by the alveopalatal affricate, as will be discussed in §3.2.2.5 below. Note that cases of final devoicing could also represent a case of assimilation to silence (Hock 1999, cited in Iverson & Salmons 2011).

To account for assimilation, I follow Zsiga (2011) and argue that it is perceptually and articulatorily motivated. If perceptual cues of a segment are weak or overlap with an adjacent segment, a listener might misperceive the signal and produce the other segment (ibid). Articulatorily, it renders production easier.26