Portuguese Morphophonology: A Generative-Markedness Approach

Introduction

§6.0.0 Nominals Identified: This chapter addresses the number and gender morphology and phonology related not only to nouns but also to adjectives, articles, demonstratives, pronouns, and infinitives. They are all nominals because, by and large, they share both syntactic and morphological properties. Syntactically, almost all of them can be subjects of verbs, objects of verbs, and objects of prepositions, as Table 6.1 demonstrates:

§6.0.1 Table 6.1 is divided into two sections: Open-ended Nominals and Grammatical Nominals. Sets 1, 2, and 3 (nouns, adjectives, and infinitives) are open-ended both because they are bona fide words—they may begin or end utterances—and because their inventory is open-ended. At any time, new nouns, adjectives, and infinitives may be coined, either from Portuguese sources or from words barrowed from other languages. The grammatical nominals (Sets 4–9) are paradigmatic forms with a closed inventory, and some are restricted to specific semantic and syntactic circumstances. They are essentially nominals, however.

§6.0.2 Set 1: There is no controversy as far as the sentences in set 1 of Table 6.1 are concerned since pinguim is a noun, and it is an ideal subject, verbal object, and an object of a preposition.←2 | 3→

§6.0.3 Set 2: Velho ‘old’ is incontrovertibly an adjective, and one might argue that in these sentences some noun has been deleted. In addition, it is of course true that sentences like these would only make sense in the real-world if listeners knew exactly of whom or about what (i.e. some actual noun) the speaker was talking. Still, it is the case that in Portuguese most adjectives can take the place of whatever nouns they might describe, and, once they do, they function as subjects as well as verbal and prepositional objects.¹ And this is precisely one of the two reasons² why both nouns and adjectives are nominals.

§6.0.4 Set 3: Infinitives function as nouns because they can occur as sentence subjects, direct objects, and objects of prepositions.³←3 | 4→

§6.0.8 Set 7: The 3^rd person pronoun (in this case, ela⁴) functions in all varieties of Portuguese as a subject or as the object of prepositions. The verbal direct object ela has become wide-spread in Brazilian Portuguese, at least in casual speech. Speakers of European Portuguese do not use subject pronouns as direct objects, they would say Quero-a for ‘I want her’.

§6.0.9 Set 8 demonstrates the 3^rd person direct object pronoun. It cannot occur as either object of a preposition or as a subject. The pronouns of Sets 7 and 8 are high valence nouns insofar as they, like infinitives, never function as adjectives. On top of that, they are never modified by adjectives without a copula: Ele é velho ‘He is old’ but *Ele velho foi comigo ‘*Old he went with me’. And whereas quero-os means ‘I want them’, quero os velhos only means ‘I want the old ones’—the ←4 | 5→direct object os is homophonous but not homologous with the definite article os. (See Table 6.18 and §6.2.1.0–.1, §6.2.5.3.0–4.2.)

§6.0.11.0 The other reason for considering nouns, adjectives, definite and indefinite articles, nominative pronouns, demonstratives, possessives, 3^rd person direct object pronouns, and infinitives together is that at least some of each category’s forms can reflect gender and/or number.⁵ Table 6.2 shows how all these forms reflect gender and number:

←5 | 6→§6.0.11.1 Table 6.2 shows that for all nominals standard orthography reflects plurality with a word-final –s (<z#). Beyond this, in Sets 1, 2, 4, 8, 9 the vowel –o (<o#) indicates [-feminine] gender. In the eight cases where [+feminine] gender can be indicated, –a (<a#) occurs. On the other hand, potentially word-final –e (in Sets 6 and 7) indicates [-feminine] gender, and the indefinite article (Set 5) appears to have no morpheme indicating [-feminine] gender. The 3^rd person direct object ←6 | 7→pronouns in Set 8 appear to be mere sequences of regular gender ([±feminine]) and number [±number].⁶ Set 9: The 1^st person plural form, like all possessive pronouns, reflects the number and gender of the possessed item(s) with the typical gender and number reflections.

§6.0.12.0 Tables 6.1 and 6.2 use the markedness strategies advanced in Chapter 1 and put to use there, as well as in Chapters 2, 3, 4, and 5. That is, the marked nominal cases are [+feminine] and [+plural], whereas the unmarked cases are [-feminine] and [-plural]. Such an analysis predicts: (A) In representative texts, plural nominals will be less frequent than non-plurals. (B) Non-plural nominals will be more widely interpretable than the plurals. (C) The marked gender ([+feminine]) will occur less frequently than its negation [-feminine]. (D) The marked gender will be restricted to a subset of referents. (E) The unmarked gender will refer to a more general set of referents. That is, [-feminine] may include [+feminine] referents, whereas the converse does not normally obtain: [+feminine] may not, in general, indicate non-feminine entities among particular words’ referents.⁷ This approach provides an economic and elegant analysis of Portuguese morphophonology.⁸

§6.0.12.1 Concerning the frequency of the marked cases, the Portuguese phonological dataset used in this study provides a convenient set of texts on which to test these predictions. The definite and indefinite articles and demonstrative pronouns are easily isolable in the corpora and co-occur with and modify or stand in for nouns. Table 6.3 displays their relative frequency in the dataset.

←7 | 8→Table 6.3 bears out the prediction of text frequency. Of the 530 nominals counted in the dataset, 298 (56.23%) were [-feminine] and 426 (80.37%) were [-plural]. Beyond their textual frequency, marked and unmarked nominals’ ranges of reference are inversely related to their markedness. Two Portuguese proverbs and modifications thereof demonstrate this principle.

§6.0.12.2.0 The two standard Portuguese proverbs occur in unmarked cases: Proverb 1 unmarked gender, Proverb 2 unmarked ←8 | 9→number. In the modified set of proverbs, the subject of Proverb 1 has changed from [-feminine] to [+feminine], and the subject of Proverb 2 has changed from singular to plural.

The Morphophonological Structure of Open-Ended Nominals

§6.1.0 It is obvious from Table 6.2 and the discussion and tables following it that the structure of most nominals is:←9 | 10→

Nominal Pattern ⁹¹⁰1:

Now, using morphophonemic transcription, Table 6.5 shows the structure of the open-ended items in Table 6.2—all of which have monomorphemic roots.

§6.1.1 Table 6.5 shows that the plural morpheme is unequivocally <z# for open-ended nominals. It also represents plurality on a significant portion of the grammatical nominals as well (see Tables 6.22, 6.24, 6.25). Indeed, nominal pluralization is largely, but not entirely, unproblematic.¹¹ Notwithstanding the nominal plural morpheme’s ←10 | 11→unambiguous identity and the transparency of the forms in Table 6.5’s Sets 1–3, nominal pluralization impedes a phenomenon addressed in the next section: stem vowel metaphony.

←11 | 12→The words analyzed in Table 6.6 do not exhaust the category of singular/plural root vowel metaphony in Portuguese. They come from a list of 32 nouns (Cunha, 1969: 127) of this type.

§6.1.2.1 The 4^th column of Table 6.6 identifies /ᴐ/, rather than /o/, as the vowel underlying both the singular and the plural forms of these words.¹² This tack acknowledges that the vowel in these nouns was a short (i.e. open, lax) vowel in Latin, /ŏ/, which regularly became /ᴐ/ in Portuguese (Williams, 1962: 36).¹³ In addition, this identification is the cornerstone of a simple rule of stem vowel metaphony:

Rule 6.1 The Closing of Nominal Class 1 (N₁) Vowels:

The environment or determinant of this rule (i.e. the material to the right of the slash) is for nominals in class N₁¹⁴ that are neither plural nor feminine. The non-feminine suffix must be preceded by at least one but no more than two consonants.¹⁵ The disjunction in the determinant indicates that the [-feminine] gender morpheme can be either an inflecting morpheme or a derivational one.¹⁶ The determinand, i.e. the ←12 | 13→specification to the left of the arrow indicating a transformation, more than adequately describes a stressed /ᴐ/, which in the environment of the determinant loses its lax quality ([+wide glottis]) and becomes [o]‌.

§6.1.2.3 The phenomenon generated by Rule 6.1 can be explained in the following manner: The presence of a [+high] back vowel [ŭ] in a syllable following one containing a non-high back vowel, especially a labial vowel such as [ᴐ], will cause the formants of the open vowel to approximate those of [o]‌.¹⁷ In the evolution of Vulgar Latin into modern Portuguese, the word-final vowel, i.e. the morpheme <o# or +o#, indicating [-feminine] gender reduced from [o] to [ŭ] earlier than it did when it was followed by the plural suffix. Words inflected for plurality, i.e. ending in <z#, end with a potentially closed syllable, and a contoidal coda such as this can impede or at least retard the rising of a tautosyllabic vowel.

§6.1.2.4 Word-final atonic vowels, especially [ŭ], undergo considerable reduction and a high degree of effacement (§3.2.1.2, §3.3.2.8, and Table 3.18b), but such is not the case when the plural suffix follows this [-feminine] morpheme. And the reason for this is that the deletion of a word-final vowel such as [ŭ] may leave a word with a somewhat aberrant syllable coda, i.e. a word-final syllable ending in other than a vowel or /n, r, l, z/ e.g. povo ‘folk’: #pᴐv++o#<∅# ➔ [ˡpo.vŭ] ➔ [ˡpov]. Loss of the vowel in a plural form would end the word with a much more aberrant consonant cluster: povos ‘peoples’: #pᴐv++o#<z# ←13 | 14→➔ [ˡpᴐ.vŭš] ➔*[ˡpᴐvš], were this gender indicating vowel deleted.¹⁸ The atonic vocalic conservatism of the potentially closed word-final syllable of plural nouns, as opposed to the volatility of the singular forms with their open syllables, contributed to the maintenance of the tonic open vowel in the preceding syllable.

§6.1.2.5 The data and analysis in §6.1.0–.2.3 account for a huge amount of the phenomena related to nominals’ grammatical number, but these data are far from closing the case on Portuguese pluralization and nominal stem vowel metaphony. Nominal gender also plays a role in nominal stem vowel metaphony, and, even in the unproblematic nominals, gender is significantly more complex.

§6.1.3.0 Simple Nominal Gender: At their most basic, nouns that refer to animate beings (human or otherwise) are inflected for gender, which can be [±feminine]. This inflection is systematic insofar as it corresponds to the actual gender of the [+animate] being particular words denote. Table 6.7 contains examples of inflected nouns:

←14 | 15→Several of these items, amigo, moço, and porco, function readily as adjectives. And all but pombo are among the Aurélio’s 3,000 most frequent words. Save porco/porca, these words are Class N nouns, i.e. their [-plural] manifestations undergo no-stem vowel metaphony.

§6.1.3.1 Stem Vowel Metaphony Revisited: All N₁ stems are exempted from metaphony when they occur with a [+feminine] suffix.¹⁹ Table 6.8 contains examples of this phenomenon:

§6.1.3.2 The metaphony displayed in Table 6.8 is most prevalent in adjectives, especially given the high-frequency (fn. 13) adjective-forming suffix +ᴐz +. The adjectives, and however many nouns of the type porco ⇔ porca there may be, undergo root vowel metaphony when their roots are inflected for [-feminine] and [-plural]. Rule 6.1 needs no ←15 | 16→alteration to account for the non-metaphony of [+feminine] inflected (or derived) N₁ words.