A Construction Morphological Investigation of the Polysemy of Persian Similarity Suffixes in the Naab Corpus

Introduction
Polysemy—defined as the coexistence of multiple, related senses that radiate from a shared conceptual core—has been a longstanding topic in Cognitive Semantics. In the context of Persian, where morphological derivation plays a central role in lexical innovation, polysemy becomes particularly intriguing when seen through the micro lens of suffixal constructions. The current study investigates the polysemy of three Persian similarity denoting suffixes—-gūn, -fām, and -vaš—using Construction Morphology as its analytic lens (Booij, 2007, 2010, 2018). While Persian pedagogical grammars often present these suffixes in simplified semantic terms (mainly resemblance in color, form, or texture), this corpus-based study reveals more complex semantic networks, degrees of productivity, and register specific behaviors.
 
 Literature Review
Traditional Persian grammars have tended to view similarity suffixes as fixed meaning bearers with transparent reference scopes (Haghshenas, 2009; Dehkhoda, 1995; Amid, 2010). Yet, the Construction Morphology framework redefines these suffixes as holistic form–meaning pairings—or constructions—that participate in broader morphological networks. Related research in Persian morphology has explored affixal productivity (Kashani, 1992; Kolbasi, 2012) and cognitive semantic expansion (Afrahi & Koushki, 2017; Pourmohammad et al., 2019), but few works have deeply modeled similarity suffixes via radial category structures (Lakoff, 1987; Geeraerts, 2010). For instance, Bamshadi & Ansariyan (2013, 2017) have demonstrated multiple polysemy patterns for other Persian derivational suffixes within Construction Morphology, yet the trio examined here—-gūn, -fām, and -vaš—remains under described in large scale corpus studies.
 
Methodology
Corpus Selection
The study draws on the Naab Corpus, a morphologically annotated Persian mega corpus containing over 15 billion tokens. This extensive coverage is essential for detecting low frequency suffixes, especially -vaš, which verges on obsolescence.
Data Extraction
Corpus searches employed orthography sensitive and morphology aware queries aligned with Naab’s lemmatization standards. These were designed to:

Capture inflected or compound embedded derivatives.
Exclude false positives where the string matched but did not function as a suffix.

For each suffix, exhaustive lists of candidate forms were generated. From these, 1,000 random tokens per suffix were drawn for manual semantic classification.
Semantic Tagging
Using radial category modeling (Brugman & Lakoff, 1988; Geeraerts, 2010), tokens were tagged into literal similarity (physical/perceptual) and extended senses (metaphorical/metonymic). For example:

Literal: golgūn ‘rose colored’
Metaphorical: janggūn ‘war like’ in demeanor

Tagging was validated against lexicographic authorities (Dehkhoda, Amid, Mo’in) and prior Persian morphological studies.
3.4 Productivity Analysis
Following Baayen (1993), both token productivity (frequency) and type productivity (distinct forms) were measured. Register based distribution was also calculated to uncover stylistic tendencies.
 
Results
Suffix -gūn

Productivity: Highest among the three, with over 900 literal similarity tokens.
Semantic Range: Encompasses chromatic resemblance (sanggūn, ‘stone colored’) and conceptual resemblance (doshmangūn, ‘enemy like’).
Register Spread: Found across literary, journalistic, and formal prose.
Radial Network: Multiple semantic branches—color, material, demeanor, degree approximation (“ish” meanings such as talxgūn ‘somewhat bitter’).

Suffix -fām

Productivity: Moderate; ~500 tokens dominated by color similarity (nīlīfām, ‘indigo hued’).
Semantic Core: Specialized in hue based resemblance; rare metaphorical uses (ensānfām, ‘human like’).
Register Spread: Strong presence in technical, artistic, and advertising contexts.

 
Suffix -vaš

Productivity: Marginal; very low token/type counts.
Diachronic Status: Largely fossilized in literary and archaic registers (shīrvaš, ‘lion like’).
Semantic Network: Narrow; retains prototypical resemblance meaning but with no significant new formations.

 
Discussion
The findings confirm a continuum of morphological vitality:

-gūn at the expansionist, high productivity pole;
-fām at a specialized, moderate productivity midpoint;
-vaš at the attritional, near obsolete end.

From a Construction Morphology perspective, these suffixes differ not only in productivity but in schematicity—the degree of abstraction of their morphological schema (Booij, 2018). The -gūn schema connects to multiple higher order constructions, supporting rich polysemic branching. -fām sustains a specialized sub network centered on color, while -vaš embodies a reduced, fossilized network.
Moreover, the corpus based radial analysis reveals how metaphor, metonymy, and degree modification mechanisms vary by suffix. This spectrum reflects broader processes of lexical innovation, semantic bleaching, and diachronic narrowing—phenomena documented across languages but under researched in Persian derivation.
Methodologically, combining large scale corpus frequency profiling with qualitative semantic modeling provides a more robust account than dictionary based or intuition driven studies.
 
References

Booij, G. (2007). “Construction Morphology and the Lexicon.” In F. Montermini, G. Boye & N.R. Hathout (Eds.), Selected Proceedings of the 5th Decembrettes: Morphology in Toulouse(pp. 33‑44). Cascadilla Proceedings Project.
Booij, G. (2018). “Construction Morphology.” In A. Hippisley & G. Stump (Eds.), The Oxford Handbook of Construction Grammar(pp. 255‑274). Oxford University Press.
Baayen, R.H. (1993). “On Frequency, Transparency and Productivity.” In Booij, G. & van Marle, J. (Eds.), Yearbook of Morphology 1992(pp. 181‑208). Springer.
Brugman, C.M., & Lakoff, G. (1988). Cognitive Topology and Lexical Networks. Berkeley Cognitive Science Report 64.
Geeraerts, D. (2010). Theories of Lexical Semantics. Oxford University Press.