This thesis investigates some phonological alternations found in Hazaragi. Hazaragi is an understudied Eastern Iranian language spoken primarily in Afghanistan and in the Quetta city of Southern Pakistan. The main focus is on how vowels alternate in prefixation, suffixation, and cliticisation, and how these alternations can be systematically analysed within a feature based framework of phonology. The theoretical approach adopted is the Featurally Underspecified Lexicon (FUL) model (Lahiri2018 and the references within). One of the most striking patterns in Hazaragi involves the surface alternations of vowels in affixes, particularly in aspectual prefixation, the ezafe clitic, and the perfective suffix. While these vowels often appear to alternate unpredictably across these morphological constructions this thesis shows that the alternations follow systematic, phonologically governed patterns. The puzzle is most clearly illustrated in the imperfective, imperative and subjunctive prefixes {mV-, bV-}, where vowel alternation does not have a clear pattern and the prefix vowel surfaces as [i e o u]. This happens in two different environments, one where the verbal stem has the same preceding vowel is straight forward. But the scenario changes when there is an /a/ in the verbal stem. Given the other vowels, we would predict that an /a/ would surface as the prefix vowel. However, instead one of the vowels [i e o u] surfaces. These alternations cannot be explained through morphological conditioning or syntactic structure, pointing instead to an active role for phonological features. The same types of alternations are seen in what is known as the ezafe construction, where attachment of the ezafe -i to a vowel final stem triggers different outcomes, /i+i/ results in a single [i], while /a+i/ yields [e]. The same holds true for the perfective, where the past participle suffix -a displays the same phonological pattern observed in the ezafe construction, with /a+i/ > [e] and yet another where the vowel /a/ attaches to /u/ of the suffix and gives [o] (/a+u/ > [o]). These patterns raise key theoretical questions about feature interaction and the role of segmental context in phonological computation. In order to analyse these phonological alternations, I first developed a comprehensive phonological foundation for the language. This begins with the construction of a segmental inventory. A detailed phonological analysis of the phonemic system is presented, highlighting several areas of contradiction in the previous literature. Notably, the exact number of consonants and vowels, the presence of /h/ and /v/ as the consonants in the language, the phonemic status of the retroflex stops /t̪ d̪ /, and the controversy between the presence of velar fricatives /x ɣ/ or uvular fricatives /χ ʁ/ is clarified. Additionally, a pilot phonetic study of Hazaragi vowels supports the analysis of Hazaragi vowels which are eight in number and are distinguished from each other in quality. This led to the development of a full feature matrix in the light of FUL model, for both consonants and vowels. This matrix plays an important role in all subsequent alternation analyses and represents the first formal featural matrix of Hazaragi in the literature. Two key phonological rules were applied: (i) the Obligatory Contour Principle (OCP), which blocks adjacent identical segments, as in /i+i/ > [i]; and (ii) Tongue Height (TH) Matching, a novel rule that prohibits sequences of the features [LOW] and [HIGH] to solve the core phonological phenomena, beginning with an investigation of the ezafe construction. This rule is formulated within the FUL model’s ternary condition of match, mismatch, and no mismatch. A match occurs when there are same features as in /i+i/; a mismatch arises when the features are in direct contradiction, seen in /a+i/ (/a/ has the Tongue Height feature [LOW], while /i/ has [HIGH]); and the most important is the no mismatch condition, which permits combinations where features are neither identical nor contradictory as in [e] (it is a mid vowel, which is unspecified for the Tongue Height feature, neither [HIGH] nor [LOW]). These same rules, OCP and TH Matching are also applied to the perfective paradigm. There, they account for alternations such as /a+u/ > [o], further supporting the TH Matching rule. This reinforces the proposal that a sequence of feature [LOW] followed by [HIGH] is disallowed. In such cases, the grammar, following FUL resolves the feature conflict by selecting a mid vowel [e o], which represents a no mismatch configuration. The most significant theoretical contribution of the thesis is found in the analysis of prefix vowel alternation. Drawing on data from the imperfective, subjunctive and imperative paradigms, and equipped with our new formulated TH Matching rule, I demonstrated that the prefix vowel in Hazaragi is underspecified for the feature [HIGH]. The surface realisation of the prefix vowel [i e o u] is determined by the phonological features of the following stem. Two mechanisms are identified: regressive assimilation from the stem vowel, and Tongue Height feature transfer from the stem initial consonant. In particular, it is shown that consonantal Tongue Height feature plays an important role in the surfacing of the prefix vowel in the context of the verbal stem vowel /a/, supporting FUL’s claim that aperture features apply equally to both consonants and vowels. This finding supports one of FUL’s central theoretical claims: that consonants and vowels share the same set of PLACE features. The thesis also identifies the relevance of the ternary condition in FUL, match, mismatch, and no mismatch, in predicting prefix vowel realisation. For example, when a stem-initial consonant has Tongue Height [HIGH], the prefix vowel surfaces as [i]; when it has Tongue Height [LOW], the prefix vowel surfaces as [e]; and when the consonant is unspecified for Tongue Height, a no mismatch condition allows either [i] or [u] to surface depending on labiality. These patterns are unified under a single featural explanation and shown to reflect the systematic structure of the Hazaragi grammar. Taken together, this thesis makes several important contributions. Empirically, it is the first comprehensive phonological analysis of Hazaragi affixal morphology, based on original data and grounded in a theoretically informed feature model. Theoretically, it supports key tenets of the FUL model including the assumption of same features for the consonants and vowels, the relevance of underspecification, Tongue Height features of the consonants and the role of ternary matching conditions in resolving surface variation.