What suffix rule applies to making rival into rivalrous (i.e., is this legitimate)

I was annoyed to find my neuroscience textbook transforming the noun "rival" into an adjective in

...IT changes systematically when people and animals report switches in rivalrous percepts.

I checked my 1913 Webster's Revised Unabridged Dictionary and it had no such suffix production on the noun rival. It suggested a verbal noun form that would apply, i.e., rivalling. One might say then "report switches in rivaling percepts" (looks more like an adjective in context). I could not find "rivalrous" in an online text version of the Oxford English Dictionary I have (but might not be the entire Oxford). Now that I think of it, saying simply "rival percepts" would be correct.

I checked with Google Ngram American English and rivalrous showed no usage really until a spike shortly after 1920, then declined to almost nil usage but began to increase after 1940, reaching a very small decimal value (1e-6 percent) by 2019. British Ngram no usage until began to climb after 1960, reaching a similar "peak" of 1e-5 percent or the like (have trouble counting the decimal places).

My Princeton WordNet resource has rivalrous:

Overview of adj rivalrous

The adj rivalrous has 1 sense (no senses from tagged texts)

1. emulous, rivalrous -- (eager to surpass others)

But WordNet admits it is very rare:

wn rivalrous -famla

Familiarity of adj rivalrous

rivalrous used as an adjective is very rare (polysemy count = 1)

The usual production for noun to adjective is

The simplest way to turn a noun into an adjective is to add suffixes to the end of the root word. The most common suffixes used to create adjectives are -ly, -able, -al, -ous, -ary, -ful, -ic, -ish, -less, -like and -y. For example, turn the noun "danger" into the adjective "dangerous" by adding the suffix -ous.

"rivalrous" appears to me to be an erroneous production which is somehow creeping into usage. But how should one produce an adjective from rival?

Solution 1:

Executive summary

1. English usage of the word rivalrous is rare (about one per hundred uses of rival) in the Google Ngram corpora of published material today and non-existent in corpora consisting of social interaction general language (SUBTLEX-US).
2. Usage of rivalrous has remained rare for hundreds of years very likely through the linguistic phenomenon of blocking, i.e., the conscious or unconscious resistance of language users to a new form synonym for an established form of a particular word meaning. Here we are talking about the adjective describing the condition of being in competition, which is succinctly covered by the word rival itself. This is a particularly potent mechanism when the proposed synonym is less efficient, e.g., longer, more syllables and when the formation is not accomplished through a commonly familiar route (see next point).
3. The word rivalrous is not formed by a standard root + -ous suffix attachment rule, so it appears even more anomalous (being unfamiliar to begin with, given the rare use) when encountered. By rule we mean here the observable commonly used pattern of construction (whether unconsciously or consciously applied) by language users, inferred from present-day English corpora and the descriptive framework of current grammars of English, including as a specialized area in the latter the classic The Categories and Types of Present-Day English Word-Formation by Hans Marchand, 1960.
4. The word rivalrous does not seriously violate the grammatical rule system of English, but the data show that most writers prefer the word rival when describing entities standing in competition for superiority (and rivalrous is not found at all in corpora sourced from spoken conversation). If your goal is to put your ideas into words with simplicity and intelligibility (as H.L. Mencken wrote in The American Language in 1919), you have no reason to use the word rivalrous. I should note that Mencken was not hostile to evolving language, saying, without sarcasm, that "the American vulgate is not only constantly making new words, it is also deducing roots from them," and, criticizing overly prescriptive grammar, he opined that no first-rate writer has ever written a textbook on the art of writing.
   Usage
   Interrogating the Google Ngram Viewer (more on Google Ngrams below) shows rivalrous appearing about once for every 100 appearances of the word rival as of 2019. The usage ratio rivalrous/rival has been growing approximately linearly with time about 0.0125%/year over the last 80 years, i.e., since around 1940. However, use of the adjective rivalrous remains relatively rare.

Because rival is used as a noun or as an adjective, you might expect at least twice as many appearances as rivalrous, which is solely an adjective, other considerations being equal. However, the observed 100:1 rival:rivalrous ratio dwarfs the ratio 2:1, so the additional part of speech rôles of rival do not explain the difference in rate of appearance vis-à-vis rivalrous. You can search Ngram with query "rival_*", i.e., append underscore + asterisk to the word rival, to see usage frequencies for rival as a noun, adjective and verb on the same graph. Though rival is also used as a verb, the verb use appears about a third as frequently as its use as a noun or adjective.

check social interaction corpus

In order to get an idea of the possible usage of rivalrous in another language register (i.e., one less academically sourced), we downloaded and searched a copy of the SUBTLEX-US corpus from the SubtlexUS – Lexique website, approximately 51 million words in lines from 8,388 different subtitle files from films and television programs: Brysbaert, M. & New, B., 2009, Moving beyond Kucera and Francis: A Critical Evaluation of Current Word Frequency Norms and the Introduction of a New and Improved Word Frequency Measure for American English. Behavior Research Methods, 41 (4), 977-990.

Using the Python programming language (on a Linux machine, but this commonly used open-source programming language also runs on Windows) and its support for pattern matching (regular expressions), we found 307 occurrences of rival- stem words (in the 50 million word SUBTLEX-US corpus). The forms and counts were as follows:

'rival': 178, 'rivals': 61, 'rivalry': 52, 'rivalries': 5, 'rivaling': 5, 'rivaled': 3, 'rivaly': 1, 'rivalled': 1, 'rivalli': 1

There were no occurrences of the word rivalrous. That outcome is not entirely unexpected, given the Google Ngram analysis above, i.e., rivalrous is a rarely used word and when it does appear, it is predominately in published academic work in social science, economics and in occasional perception-related studies, rather than in social interaction (the latter the focus of the SUBTLEX-US corpus).

Grammar.morphology.blocking

In the context of morphology, David Crystal (A Dictionary of Linguistics and Phonetics, 6th Edition) defines blocking as the prevention of word formation due to the existence in the language of a word with the same meaning as the one to be formed. He provides an example, "Although we may obtain curiosity from curious, English does not allow gloriosity, because glory already exists in the language, and therefore blocks it."

We believe that the very small usage of rivalrous vs rival as an adjective in English is probably a result of blocking. In Natural Selection in Self-Organizing Morphological Systems, c. 2011, Lindsay and Aronoff suggest that the “primary driving force behind competition in the lexicon of a language is that, in general, languages do not tolerate true synonymy...one stem + affix combination will be preferred over another.” The Ngram and SUBTLEX-US data above imply that most users of English find the “slot” for the concept “of the nature of rivals” (in particular the slot related directly to forms from the root rival) already occupied by rival used as an adjective.

This proposed competition between rival stem + affix combinations probably is related also to economy, e.g., which form is easier to say or which has fewer syllables (see Chapter I, The Principle of Economy Applied to Words, in Herbert Spencer's 1884 Philosophy of Style for an interesting discussion of the subject). That being said, Zipf (George Kingsley Zipf, Human Behavior and the Principle of Least Effort, 1949) suggested that there is a tension between, on the one hand, the speaker's desire to use a simple construction (e.g., easier to pronounce), and on the other hand the improved probability of decoding by the hearer if more pieces of information are available (making the word distinct from another).

To take a rough look at this tension, we graphed (using NLTK, the Natural Language Toolkit by Steven Bird and collaborators, in a Python, NumPy, Matplotlib computing environment) the word length frequencies from the venerable 1961 Brown Corpus, the first million-word electronic corpus of English, produced from 500 published sources categorized in 15 genres, e.g., news, learned (academic), fiction, etc.:

If you filter out so-called stopwords, e.g., 'I', 'me', 'my', 'myself', 'here', 'there', 'when' (179 words in the English portion of the Stopwords Corpus by Porter el al distributed with NLTK) and punctuation, the top four word lengths, 3, 2, 4, 1 in the graph above, are replaced by 5, 4, 6, and 7 and the total word count drops from 1,161,192 to 509,267. In English, it appears that the competing motivations of ease of production and ease of understanding settle out in the 4-7 character range for words referring to other than the simple concepts loosely categorized by stopwords.

While we were examining the Brown Corpus, we found no occurrences of rivalrous, but the word rival appeared, 5 times as an adjective, 4 as a noun, and twice as a verb (we used the Brown Corpus Form C, which tags each individual word with a part of speech class).

Another factor is that, "conscious word formation being imitative" (borrowing an apt phrase from a discussion on word formation by Meagan Ayer at Dickinson's College), the form of a new word is more probable when it mimics a pattern occurring in other words known to the speaker. Of course, speech errors, along with neologisms and borrowings, continually introduce random change into the system (from 2011 Lindsay and Aronoff cited earlier). We will have more to say about this familiar pattern concept later.

It appears that rivalrous seems has been blocked in the sense discussed above for at least 200 years. We processed the Google English One Million corpus with a computer and found the first occurrence of rivalrous in this corpus in the year 1810, specifically, one occurrence out of 113,677,334 words (single ngrams) total, from 960 books scanned. For comparison, there were 2697 occurrences of the word rival that year.

The Google English One Million corpus we used above contains approximately 262 million lines, with each line a word match count for each of the 393 years with samples within the 488 year range between 1520 and 2008 derived from scans of 999,999 books (hence "Google One Million") published in English.

Etymology

rival, rivality
Let us attempt to rough out the etymology of the English word rival, with the goal of understanding how the form rivalrous might arise.

Consulting A Latin Dictionary, founded on Andrews’ edition of Freund’s Latin Dictionary, by Charlton T. Lewis , 1879 Oxford University Press (a dictionary of ancient Latin which cites Roman era authors and works), we find an entry for rīvalīs, an adjective (first part of speech under that entry) derived from the noun rīvus. rīvus has a separate lemma (a headword with its own entry in the dictionary) as a noun with meaning a small stream of water, a brook. The adjective meaning for rīvalīs given by Lewis, as one might guess, begins with the sense of, or belonging to a brook.

The Lewis entry for rīvalīs includes a noun (substantive) second part of speech use. The plural substantive is rīvales (note the suffix change to -ales), those who have or use the same brook, neighbors. The second meaning under the substantive rīvales is a figurative extension of the first meaning, using a singular form identical to the adjective, i.e., rīvalīs, one who has the same mistress as another; a competitor in love, a rival.

Marchand (The Categories and Types of Present-Day English Word-Formation by Hans Marchand, 1960) §4.6.1 tells us that the English suffix -al derives from Latin (Classical, Medieval and Modern) suffix -ālis. We have a clear path, then, from a Latin adjective or noun rīvalīs to an English adjective or noun rival, where the Latin suffix -alīs is replaced by the English -al suffix on the conversion of the loaned foreign word. Johnson (Samuel Johnson, Dictionary of the English Language, Sixth Edition 1785) agrees with this path, i.e., the English adjective rival derives from Latin rīvalīs and his noun sense of rival proceeds from that identically.

Rival is our root, "the base form of a word which cannot be further analysed without total loss of identity...the part of the word left when all the affixes are removed" (David Crystal, A Dictionary of Linguistics and Phonetics, 6th Edition). One of Crystal's senses for a definition of the concept word seems adequate for our purposes, i.e., "words are the physically definable units which one encounters in a stretch of writing (bounded by spaces)...."

By affix we mean the formatives or morphemes that enter into the construction of words. Following Aronoff and Fudeman, What is Morphology?, we regard morphemes as the smallest linguistic pieces with a grammatical function. The authors "take a no-holds-barred approach to linguistics [and] use any tool or method that will tell [them] how language works." That is sufficient definition of linguistics here. When we say "grammatical function," we mean a function that helps us recognize the way a word is formed and understand how it could be used in a sentence, grammar being "concerned with the structure of words (morphology), and of phrases and clauses (syntax)" (quoting 2011 Oxford Modern English Grammar by Bas Aarts).

The affixes that precede the root are called prefixes. Those affixes that follow the root are suffixes (we ignore infixes, which are uncommon in English).

We are interested only in lexical morphemes, i.e., those affixes that produce new words when attached to a root. This contrasts with inflections, which modify a word to indicate a new tense, as in look-->looked, or a plural form cat-->cats. The affix may attach to a stem (some prefer the term base), i.e., certain combinations of a root and a morpheme which we may consider as a single unit at times, e.g., the adjective careful, which could become an adverb with attachment of the suffix -ly to form carefully, so we consider careful the stem in that case, though careful itself could be further analyzed into

[ ADJ [Ncare] ful ]

where -ful is the suffix that changes the noun care into the adjective careful. In that context, care is the stem, and is also the root, since it cannot be further analyzed. Following 1913 Webster's and An Anglo-Saxon dictionary, by T. Northcote Toller ,1921, we see that the word care derives from the AS (Anglo-Saxon) adjective cear, meaning "sorrowful, anxious, solicitous." AS cear in turn attaches the AS suffix -ful with sense "full" (this suffix terminates many AS adjectives).

It is not impossible, of course, that rival passed from Latin first to French, and from French to English. The 1740 Dictionnaire de l'Académie françoise does have an entry for the noun rival, meaning "competitor in love;" and noun rivalité, "competition between lovers," i.e., our word rivalry, but that does not alter our course here. Also, derivations of new words constructed from rival (e.g., rivalrous) in the French could have come into English, rather than having been constructed within English directly by affix attachment to rival. This is further complicated by the fact that many English affixes were taken from the French and became part of the word construction toolbox of native English. However, we found no evidence of rivalrous in French.

What about the word describing the state of being rivals, e.g., the noun rivalry? Johnson lists rivality and rivalry as synonyms in this sense. He tells us that rivality derives from Latin rivalitas. The Lewis Latin dictionary has an entry for rīvālĭtas, rivalry in love.

Marchand §4.55.1 tells us English -ity forms an abstract substantive (noun) from adjectives, but in §4.55.2 he describes a variant where the English substantive is instead formed directly from a Latin substantive in -itas. We thus have, as Johnson asserts, the English substantive condition of being rivals, i.e., rivality, coming in directly from the Latin rīvālĭtas. We recall from above that the French have the word rivalité, "competition between lovers," i.e., their form for our word rivality (though far more common today in English is rivalry).

Johnson implied that rivalry was derived from rival (which came in from Latin rīvalīs as we discussed above), but he did not state explicitly that the suffix -ry was attached to rival to form rivalry, possibly because he considered it a trivially common derivation. In Dictionary of the English Language he apologizes for some of his perhaps needless concern for primitives, "for who does not see that remoteness comes from remote, lovely from love, concavity from concave and demonstrative from demonstrate?"

rivalry

Marchand comments in §4.1.5 that just as the introduction of foreign words into a language is essentially uncomplicated, so is their combination with native derivative elements. Since there is no structural problem involved in the use of a foreign lexical unit, it can be treated like a native word. Accordingly, native prefixes and suffixes were added to French words almost immediately after their introduction. For example, native suffixes like -ful, -less and -ness were early-on attached to incoming French words like faith to produce faithful, faithless, etc. by 1300.

The case is somewhat different with foreign affixes added to native words, since a structural pattern is involved (e.g., English rival + French -ry). In this case the native speaker must become familiar with the pattern of combination of the foreign affix in the foreign words of its origin in order to learn how to apply the affix to native words. For example, the affix -ous , which Marchand tells us (in §4.70.1 of The Categories and Types of Present-Day English Word-Formation, 1960) came into English through Middle English loans from the French (e.g., advantageous, courageous, etc., which we believe would have been in form avantageux, courageux in the French), was picked up as an English formative, and combined with burden (Anglo-Saxon berðen or burthen), to form the English adjective burdenous (becoming obsolete in favor of burdensome these days apparently).

Marchand, §4.32.1, says the -ery, -ry suffixes originated with French words in -erie forming concrete and abstract substantives from substantives. His examples of direct loans (from French to English) in this context include the word ribaldry. We would add the example of French chevalerie (chivalry) from French chevalier (knight), i.e., English chivalry could have come in more or less directly (with slight form modification of the French noun root cheval meaning "horse," to chival, which apparently did occur in Norman French) from the French chevalerie using the typical affix transformation -erie to -ry. We note that we could find no French rivalerie in current French word lists or in the 1740 Dictionnaire de l'Académie françoise.

Marchand tells us also in §4.32.1 that the word husbandry was first recorded in English in 1290, implying that the originally French suffix -ry had already been established as an English formative by then. This is because husband was a native English word with Old English origin hús-bonda, from page "d0574" of the electronic version of An Anglo-Saxon dictionary, by T. Northcote Toller ,1921. Similarly, in §4.32.2, Marchand tells us that OF (Old French) jeulerie is not quoted before 1434 (i.e., the 15th century), though English jewelry occurs at the beginning of the 14th century (i.e., the 1300's).

This suggests the -ry suffix was already in use by the English to make new derivations. Among those English derivations, Marchand lists rivalry 1598, meaning "competition." We see then that English speakers attached the assimilated French suffix -ry to the word rival (or through French intermediary rivalerie if that existed for a time).

Grammar.morphology

Based on our work above, we can confidently state that rivalr- is not the root in the adjective rivalrous (the root is rival). On the other hand, -rous is not a suffix. Complex lexemes can, over time, become a single unit with specific content, losing their nature as a syntagma or combination of smaller units to a greater or less extent (referring to lexicalization, Claudia Pisoschi in Considerations of Some English Words of Latin Origin citing English Lexicology, Lexical structure, word semantics and word-formation, L. Lipka 2002). However, neither rivalr- nor -rous have developed that character, the root rival still being dominant (prevalent in usage), as is its derived form rivalry.

conglutination, affix reanalysis Martin Haspelmath (The Growth of Affixes in Morphological Reanalysis, in Yearbook of Morphology 1994, 1-29) talked about conglutination in his Section 2.2, the case of affix reanalysis where an inner affix (our -ry) and an outer affix (our -ous) are combined such that the inner affix becomes part of the outer affix, but the meaning of the original outer affix is not changed. -ry + -ous produces -rous, having the same meaning as -ous would normally (as -ous had in infrequent use of the word rivalous). It is possible that English users first became familiar with the forms chivalry and chivalrous (English users may actually have created the form chivalrous, e.g., we finding no occurrence in the 1839 French text of Chanson des Saxons, written c. 1200 by Jean Bodel, of any forms other than cheval, chevalerie, and chevalier) and some feel that rivalry could similarly become rivalrous.

present usage of consonant + suffix -rous form As we mentioned earlier, Marchand (The Categories and Types of Present-Day English Word-Formation ) §4.70.1 tells us that the affix -ous is an adjectival suffix which came into English through Middle English loans from the French (who had adapted Latin -osus and -us word endings), acting as an English formative from the 14th century on. Adjectives constructed with this suffix have the meaning "full of, of the nature, character or appearance of" (Marchand).

Words ending in a consonant + -rous are relatively infrequent and appear to be used primarily in learned words or scientific terms from Latin or Greek. For example, we analyzed (with a computer) 210,687 words (the length of the corpus after removing any proper nouns) in the Words Corpus included in NLTK (Natural Language Toolkit). The Words Corpus is the spell checker word list from the Unix operating system. We first counted all words ending with the suffix -ous, obtaining 6,395 words. Then we extracted all words with a suffix consisting of a consonant followed by -rous (because it would not be unusual to see -ous appended to a word ending in a vowel+r, e.g., -er in murder to murderous, -or in clamor to clamorous). Our list of words ending with consonant followed by -rous totaled 194, i.e., only 3% of the total number of -ous suffix words.

These were almost all scientific, medical, or technical words from Latin and/or Greek. For example, arthrous, which we interpret as meaning "pertaining to the joint," from Greek noun αρθρον or arthron meaning "a joint."

There were also a few appearances in our list of 194 by words like thunder that produce the same adjective in two forms, e.g., thunderous or thundrous, the latter form thereby slipping through our filter which otherwise rejected forms like murderous where the root ends in -er and -ous has been appended. The only words of general usage in this list (words ending in consonant followed by -rous) were rivalrous itself, along with chivalrous, revelrous and ribaldrous.

Perhaps the sound of words like thunderous or thundrous and wonderous or wondrous have conditioned the English ear as it were to preferring that final -rous sound. Though technically possible, rivalous, chivalous, revelous and ribaldous are not often used, e.g., we found zero occurrences in the Brown Corpus and zero in the SUBTLEX-US corpus.

Look at usage rivalrous, chivalrous, revelrous and ribaldrous

If we interrogate Google Ngram 1800-1900 English with

rivalrous/rivalry, chivalrous/chivalry, revelrous/revelry, ribaldrous/ribaldry

We see:

This graph is consistent with the idea that chivalrous satisfies a unique need for this semantic set headed by chivalry, i.e., the construction is not blocked because there is a need for an adjective to describe activity that is honorable and courageous as required under the code of chivalry.

The graph is also consistent with the concept of blocking. Rivalrous is superfluous (one can simply use rival to describe one of the entities involved in a rivalry), revelrous is superfluous (one rarely needs to describe activity at a party as being party-like; the usual declaration is about reveling in some occurrence, or engaging in revelry, rather than describing revelrous behavior at a party; excessive revelry would be more riotous), and ribald behavior is exactly that which ribaldrous would seek to describe, redundantly.

Conclusion

We have shown that the construction of rivalrous is indeed unusual in terms of what is expected in that attachment of the -ous suffix and what is observed in present usage. Our analysis of English corpora demonstrates that, by a large margin, most users of English feel no need for a derivative of rival to characterize entities standing in competition for superiority, the word rival already serving that purpose.