Запета

Overview - logical, not syntactic, punctuation

Serbian punctuation, codified in the Правопис Матице српске, follows слободна (логичка) интерпункција - free, logical punctuation: a comma marks what is inserted, added as an afterthought, or moved out of its expected place, never subordination as such - a subordinate clause isn't automatically comma-marked just for being subordinate. This is a major gap between Serbian and languages whose comma rules are syntactic and near-obligatory, marking every complement and relative clause with a comma regardless of meaning; carrying that reflex into Serbian is one of the most visible errors at this level, precisely because Serbian readers see the comma as meaningful, not decoration.

Because the rule is semantic, not mechanical, this lesson turns on one recurring question: does the comma mark something genuinely inserted, added on, contrasted, or displaced - or is it just a bound clause doing its ordinary job? Every rule below, even the always-comma cases, is an instance of that question, not a separate item to memorize.

Bound clauses take no comma - inversion takes one

A postposed subordinate clause - one that follows the main clause in its expected position - is bound to it and takes no comma: Мислим да је то тачно. (I think that's correct.), Јави се кад стигнеш. (Get in touch when you arrive.), Човек који седи поред прозора је мој брат. (The man sitting by the window is my brother.) In each case the subordinate clause completes the main clause's meaning in its normal slot - nothing inserted or displaced, so nothing marked.

The moment the same clause leaves that postposed slot - typically fronted before the main clause - it counts as displaced, and the logical-punctuation principle marks it with a comma: Кад стигнеш, јави се. (When you arrive, get in touch.), Ако будеш имао времена, дођи. (If you have time, come.) The information is identical to the postposed versions above; only the word order changed, and the comma tracks that change, not the subordination itself.

conditionals-b1's real-condition pattern, ако + present, is exactly this: its example Ако имам времена, доћи ћу на журку. fronts the ако-clause and takes the inversion comma for that reason, not because it's a conditional. Once futur-ii-b1 exists, its subordinate-only tense slots into this same rule unchanged - the comma tracks word order, never which conjunction or tense is involved.

Restrictive vs. non-restrictive који - a meaning-bearing minimal pair

advanced-clause-syntax-b2 already flagged this distinction in passing; here it's the center of the lesson, because the comma is not decoration - it changes what the sentence says. Студенти који су положили славили су. (The students who passed were celebrating.) has no commas: који су положили restricts студенти to a subset - only the ones who passed were celebrating, and the sentence says nothing about the others. Студенти, који су положили, славили су. (The students, who had passed, were celebrating.) sets the same clause off with commas: now non-restrictive, adding incidental information to a group already fully identified - all the students were celebrating, and the writer mentions, as an aside, that they had passed.

Some languages punctuate both sentences identically - a relative clause always takes a comma regardless of restrictiveness - so a reader or writer coming from such a language literally cannot express this distinction in their own punctuation, and has to learn to notice it here as new information, not just a new rule applied to old.

English, by contrast, draws this exact same line - restrictive "who/that" takes no comma, non-restrictive "who, …," does - so English intuition, unusually, helps here rather than betraying it; this is one of the rare places in the course where the English comparison is more reliable than instincts carried over from a syntactic-comma language.

Always-comma cases: vocative, parentheticals, contrast, dislocation

A handful of constructions are always comma-marked, because they always satisfy the underlying principle - something genuinely set apart from the main clause's ordinary flow. The vocative, naming who is addressed, is set off from the rest of the sentence: Марко, дођи! (Marko, come here!), and on both sides mid-sentence: Реци ми, Марко, шта се десило. Parenthetical connectors - међутим (however), дакле (therefore), наравно (of course) - are inserted comments rather than part of the grammatical structure, and are usually set off by commas wherever they land, unless tightly bound in sense and intonation to what they modify: Он је, међутим, одбио понуду.

Contrastive coordinators - а, али, него, већ - always take a comma before them, because a contrast is by definition a break in the sentence's expected direction: Он воли кафу, а она чај. (He likes coffee, and she tea.), Није дошао на време, него је закаснио. (He didn't come on time, but rather was late.) This holds even though plain coordination with и or или takes none (next section) - the contrastive conjunctions mark a turn, not a simple addition.

Dislocated constituents - a phrase pulled out of its ordinary slot, re-placed for emphasis, and resumed by a pronoun - apply the same principle to word order rather than a connector: word-order-information-structure-b2's comma before a fronted, resumed topic is this same rule in action, not a separate one to relearn.

No comma before и/или, never before complement да, and јер as a judgment call

Plain coordination with и (and) or или (or) takes no comma, because nothing is inserted, added on, or contrasted - it's the default way to join two equal things: Купио сам хлеб и млеко.; Он ради у канцеларији и код куће. This is the same principle as the always-comma cases above, just returning the opposite answer: plain addition doesn't satisfy it.

Complement да - introducing what is said, thought, wanted or believed, as in complex-sentences-b1's Желим да дођеш. - never takes a comma before it, in any register: Мислим да је то тачно.; Кажем ти да не бринеш. The complement clause isn't inserted or displaced; it's the verb's own object, filling the grammatical slot the verb requires. This is one of the most common transfer errors at this level for a writer whose native language treats an equivalent complement clause as comma-obligatory for exactly the construction Serbian never marks.

Јер (because) is the one case where the writer genuinely decides, since the same word covers two different relationships. A tightly bound cause, the ordinary reason for the main clause's action, takes no comma: Остао је код куће јер је био болестан. (He stayed home because he was sick.) - the cause is part of the single, unified event. An afterthought explanation, added to justify or comment on a claim already made, takes one: Он ће сигурно успети, јер је увек био вредан. (He'll surely succeed - because he's always been hard-working.) - here јер introduces the writer's own justification, not an event's grammatical cause, and the comma marks it as an added-on comment. The choice is semantic, not lexical: ask what the јер-clause is doing, not which word it is.

One shared convention: the decimal comma

One comma use needs no logical justification at all - it isn't marking anything inserted or displaced, it's a numeral convention: Serbian writes decimals with a comma, 3,14, not a period, like several other European languages. This is the one place in the lesson where that habit transfers directly and safely - worth stating explicitly so it doesn't get swept into the general warning against syntactic-comma reflexes.